Abstract:Plant and animal phenotypes are quantitative descriptions of their characteristics and traits. Accurate analysis of phenotypic features is an important prerequisite for the development of digital agriculture. The traditional phenotypic analysis task heavily relies on manual identification and measurement by agricultural experts, which is labor-intensive, costly, and sensitive to subjective judgments. Also, the traditional approach can hardly process high-throughput data. Benefited by the rapid development of the deep learning technique, as one of the most representative computer vision models, the YOLO family algorithms have shown excellent performance and great potential in plant and animal phenotypic analysis tasks, including disease diagnosis, behavior quantification, biomass estimation, and so on. In this review, livestock, poultry, crops, fruits, vegetables, and other plants and animals were chosen as the research targets. The research progress of YOLO family algorithm applications was summarized from three aspects, namely, object detection, key point detection, and object segmentation. Along the same lines, some commonly used datasets for plant and animal phenotyping tasks for subsequent researchers were presented. Finally, the potential problems faced by current researching and the future development trend of YOLO family algorithms were highlighted, including lightweight architecture design, accurate detection of small targets, weakly supervised learning, complex scene deployment, and large model for target detection. The research aimed at providing summarization and guidance for plant and animal phenotypic analysis based on YOLO family algorithms and promoting the further development of digital agriculture.

Abstract:Artificial identification of wheat rust is costly and inefficient, and can no longer meet the needs of modern agricultural production. A lightweight dense multi-scale attention network model called Mobile-DMSANet was presented for the automatic identification of rust on wheat leaves (stripe rust and leaf rust) from images of natural scenes taken in the field. In the input layer of the network, a fast subsampling block (FSB) was used to improve the feature expression ability of the network without adding computational cost. In the feature extraction layer, three lightweight blocks called dense multi-scale attention (DMSA) blocks were used to extract the features of rust on wheat leaves. In the DMSA block, a multi-scale three-way convolution (MSTC) layer was designed to get different scales for the receptive fields, in order to improve the expressive ability of the network and its ability to perceive the features of rust disease at different scales. Six MSTC layers were used to achieve feature reuse by dense connections in the DMSA block, an approach that not only greatly reduced the number of parameters of the network but also improved the feature extraction ability for similar diseases. A coordinated attention (CA) block was also introduced to the DMSA block to increase the sensitivity to positional information and suppress background information in the image. The output layer of the network used a Softmax function to classify rust on wheat leaves. The results showed that the recognition accuracy of Mobile-DMSANet model on the test dataset was 96.4%, which was higher than that of other models. Mobile-DMSANet had only 454000 parameters, less than for other lightweight models. The proposed model can be used for the automatic identification of rust on wheat leaves using mobile devices.

Abstract:Rice blast is one of the most serious diseases of rice. It is caused by blast fungus and occurs in different growth stages of rice. The spores of blast can be transmitted through air, which seriously affects food production security. Therefore, the identification of blast spores plays an important role in the early diagnosis and control of rice blast. Based on the YOLO v8 model, an RBS-YOLO method for the detection of rice blast spores was proposed. Firstly, the algorithm introduced the PP-LCNet lightweight network in the backbone network, which used DepthSepConv as the basic block and reduced the computational effort of the model and the size of the model weight file, but hardly increased the inference time. Secondly, the efficient multi-scale attention module was introduced into the neck network, which reshaped some channels into batch dimensions and grouped the channel dimensions into multiple sub-features, so that the spatial semantic features were evenly distributed in each feature group. The information of each channel can be effectively preserved and the computational overhead can be reduced. Finally, the loss function of YOLO v8n was changed to WIOU loss function, which can reduce the impact of low-quality samples on the model during training. WIOU used dynamic non-monootone focusing mechanism to evaluate the quality of the anchor frame, and used gradient gain, which ensured the high-quality effect of the anchor frame and reduced the influence of harmful gradients. The accuracy and mean accuracy of model identification of rice blast spores were improved. The accuracy and average accuracy of the improved RBS-YOLO model were 97.3% and 98.7%, respectively, meeting the demand for the detection of rice blast spores. The weight file size and computation amount were 3.46MB and 5.2×109, respectively, which were 41.8% and 35.8% lower than that of YOLO v8n. In order to verify the detection performance of RBS-YOLO, under the same training environment and parameter configuration, the improved model was compared with the YOLO v5s, YOLO v7 and the original YOLO v8n model, and the computational load was reduced by 67.3%, 95.1% and 35.8%, respectively. Model weight file sizes were reduced by 10.14MB, 67.84MB, and 2.49MB, respectively. The results showed that RBS-YOLO can meet the demand of real-time detection of rice blast spores,which was conducive to deployment to mobile terminals.

Abstract:When identifying rice pests, issues such as targets being obscured, similarity to the background color, and proximity of multiple targets due to the rice field environment can lead to reduced identification accuracy. To address this, a rice pest identification method was proposed based on an improved YOLO v5s. The method enhanced the model’s ability to capture target location information by replacing ordinary convolution in the backbone network with CoordConv. It introduced the CBAM attention mechanism to increase the model’s focus on the target area. The Slim-neck architecture was adopted to enhance feature processing capabilities and reduce computational load. The introduction of the Soft-NMS algorithm optimized the selection of adjacent target bounding boxes, reducing missed detections. Experimental results showed that the improved YOLO v5s model achieved an mAP of 94.3% on the rice pest dataset, which was an increase of 3.8 percentage points over the original model and 1.5, 12.7, 13.6 and 1.9 percentage points higher than that of the other mainstream models such as YOLO v3, YOLO R-CSP, YOLO v7, and YOLO v8s, respectively. Ablation experiments further validated the effectiveness of each component in the improved model. Heat map analysis also demonstrated that the improved model can better learn pest features. In summary, the improved YOLO v5s model proposed achieved significant results in improving the accuracy of rice pest detection, providing a more precise identification method for the prevention and control of rice pests.

Abstract:Maize is one of the most important food crops in China. Leaf diseases of maize can seriously damage its yield, so the correct identification of disease is of great significance. However, the efficiency of traditional manual identification of leaf diseases is low. The resolution of disease images collected from agricultural fixed points or drone monitoring is low, and the key features are not significant, which cannot meet the image resolution requirements of classification and recognition models. The training effect is poor, making it difficult to accurately identify leaf diseases. To this end, a maize disease classification and recognition model based on an improved super-resolution generative adversarial network (SRGR) was designed. The images of maize leaf disease were divided into four types: large spot, rust, gray spot, and healthy leaves. The data set was divided into low resolution (LR) and high resolution (HR) images that corresponded one-to-one. In order to realize the restoration of low-resolution maize spot images to high-resolution images, this model proposed an improved strategy for the enhanced super-resolution generative adversarial networks (ESRGAN) model based on dual attention mechanism. LR images were input into the highfrequency feature reconstruction network, and channel attention(CA) mechanism after each residual dense block (RRDB) was added to extract deep detailed features of the image, making the model highly targeted in reconstructing highfrequency details and reducing the possibility of pseudo texture phenomenon.The generation network was divided into encoding and decoding parts, and the spatial attention mechanism was introduced into U-shaped dense block with skip layers to maximize the retention of maize disease effective features in the middle and low levels of the LR image of maize lesions. The probability value of high-frequency features in the input feature map was calculated to determine the position of reconstructed lesion features in the image.The WGAN-GP loss function was used to train the network to solve the problem of vanishing generator gradients, enhancing the stability of the network. The regenerated lesion images were input into the discriminant network, and images that met HR image standards were input into the ResNet34 classification model to achieve accurate classification and identification of maize leaf lesions, and images that did not meet the standards were returned to the generation network for retraining. The experimental results showed that the addition of the dual attention mechanism and the change of the loss function increased the model’s ability to recover high-frequency features and robustness.Compared with other super-resolution image reconstruction algorithms, the high-resolution reconstructed images generated based on the SRGR model improved peak signal-to-noise ratio(PSNR) and structural similarity index measure(SSIM) values, with an average increase of 2.1dB and 0.049, which was a significant improvement. Four different classification networks were selected for image classification and recognition, and the recognition accuracy of reconstructed images was improved by an average of 28.1 percentage points compared with that of LR images. Among them, the ResNet34 classification model had the highest accuracy compared with AlexNet, VggNet, and GoogleNet models. In the attention module ablation experiment, compared with the other three models, SRGR accuracy in identifying maize lesions exceeded other models by an average of 1.3 percentage points, with an accuracy rate of 97.8%. In the visualization of the recognition results, the heat map of the lesions identified by the SRGR model had the darkest color and the highest recognition degree. In summary, the research result can serve as a reference for accurate identification of low-resolution leaf disease images in crop leaf spot monitoring or drone field monitoring.

Abstract:In order to enhance the effectiveness of identifying pests in fruit trees and promptly implement preventive measures, focusing on six major pests that pose a significant threat to fruit trees, an improved lightweight MobileViT recognition model was proposed for the problems of complex background of fruit tree pest recognition in the natural environment, high difficulty of detecting the small target of the pests, and high feature similarity with the features between different categories. In enhancing the model, the partial convolution (PConv) module was employed to replace certain standard convolution modules in the original MobileViT module. Additionally, modifications were made to the feature fusion strategy within the MobileViT module, involving the concatenation fusion of input features, local expressive features, and global expressive features. The tenth layer MV2 module and the eleventh layer MobileViT module were removed, introducing an improved atrous spatial pyramid pooling (ASPP) module as a replacement, aiming to create multi-scale fusion features. Furthermore, the model adopted the SiLU activation function in lieu of the ReLU6 activation function for computations. Finally, the model underwent transfer learning based on the ImageNet dataset. The experimental results indicated that the recognition accuracy of six categories of fruit tree pest images reached 93.77%, with a parameter count of 8.40×105. In comparison with the previous version, the recognition accuracy was improved by 7.5 percentage points, while the parameter count was decreased by 33.86%. When compared with commonly used pest CNN recognition models, namely AlexNet, ResNet 50, MobileNetV2, and ShuffleNetV2, the proposed model achieved higher recognition accuracy by 8.25, 4.78, 7.27 and 7.41 percentage points, respectively, with parameter counts lowered by 6.03×107, 2.48×107, 2.66×106 and 5.30×105, respectively. Compared with Transformer recognition models such as ViT and Swin Transformer, the accuracy was higher by 19.03 and 9.8 percentage points, respectively, with parameter counts lowered by 8.56×107 and 2.75×107. The research was suitable for deployment in environments with limited resources, such as mobile terminals, which can contribute to the effective identification and detection of small target pests in fruit trees amidst complex backgrounds.

Abstract:Apples have become one of the most popular fruits in the world, and the annual production of apples in China has continued to increase. However, there are certain diseases in the growth process of apple trees, which will affect the quality and yield of apples, resulting in economic losses of fruit farmers. Therefore, in view of the problem that apple leaf diseases have diverse forms and dense distribution, resulting in low detection accuracy, an improved YOLO v7 model was proposed to accurately detect apple leaf diseases. Firstly, bidirectional feature pyramid network (BiFPN) was used to replace the original feature fusion method in YOLO v7 to improve the model’s detection ability of different scale diseases on apple leaves. Secondly, after the ELAN and E-ELAN modules of YOLO v7, an efficient channel attention mechanism (ECA) was added to enhance the ability of the model to extract features of apple leaves disease and improve detection accuracy. Finally, the loss function of YOLO v7 was changed to the SIOU loss function to accelerate the convergence speed of the model. Experimental results showed that the improved YOLO v7 model had a precision of 89.4%, a recall rate of 81.5%, a mean average precision (mAP@0.5) of 90.5%, and a mean average precision (mAP@0.95) of 62.1%. Compared with the original YOLO v7 model, they were increased by 4.9, 5.2, 3.5, and 4.6 percentage points, respectively. Compared with the Faster R-CNN, SSD, YOLO v3, YOLO v5s, and YOLO v7 models, the mAP@0.5 of improved YOLO v7 model was increased by 40.9, 20.3, 4.0, 2.3 and 3.5 percentage points, respectively, and the single image detection speed reached 12ms. The research can provide a feasible technical means for accurately detecting apple leaf diseases.

Abstract:In order to further improve the accuracy and speed of grape disease identification, the YOLO v8 model was improved. Firstly, the GhostNetV2 backbone feature extraction network was introduced to improve the feature extraction ability and recognition performance of the model. Secondly, the SPPFCSPC pyramid pooling was embedded to improve the speed while keeping the receptive field unchanged. Thirdly, the GAM-Attention mechanism was added to reduce the information reduction and amplify the feature information to speed up the recognition. Finally, Focal-EIoU was used as the loss function to improve the bounding box regression performance of the detection model, and finally the grape leaf disease identification model YOLO v8-GSGF was formed. The recognition test verified that the YOLO v8-GSGF model can achieve 97.1% recognition accuracy and 45.3ms inference time, and can achieve high-precision identification of various grape diseases. The results of the ablation test showed that all the improvements had an effect on the recognition performance of the model, and the GhostNetV2 backbone network had the most obvious effect on the model. The YOLO v8-GSGF model can achieve 98.2% recognition accuracy and 43.7ms inference time in the ablation test, which was 8.6 percentage point and 20.4ms higher than that of the original YOLO v8 model. Compared with the current mainstream recognition model, the YOLO v8-GSGF model had better performance, better recognition accuracy and speed, and the curve chart also intuitively showed that the performance of the YOLO v8-GSGF model was superior, and the improvement effect was remarkable, which can meet the needs of grape orchard disease identification and had the potential for practical application.

Abstract:Facial keypoint detection in dairy cows plays a crucial role in the automation of cow farms. It aids in cow face recognition, face alignment, head movement detection, and behavior recognition. In view of the problems of cow face occlusion and weak light in the current dairy farming environment, an improved algorithm of YOLO v7-Pose network model was proposed, which can be used for keypoint detection and head pose recognition of cow face. Firstly, dairy cow facial images were collected from cow farms by using network cameras and a dataset was constructed. Secondly, the SPPFCSPCL structure was integrated into the YOLO v7-Pose network model to enhance its feature extraction capabilities for cow facial keypoints. The WingLoss loss function replaced the OKS loss function for keypoint detection, thereby improving the accuracy of cow facial keypoint detection. Finally, L1 regularization was applied to prune the improved model, reducing the number of network parameters. The experimental results showed that the cow face keypoint detection of improved model YOLO v7-SCLWL-Pose was improved by 5 percentage points and AP0.5 was improved by 2.7 percentage points compared with the original model AP, and the memory occupation of the improved model was only 106.7MB, which was reduced by 33.6%. The keypoint detection was applied to pose recognition, and the experimental results showed that the recognition accuracy of the motions of looking up and looking down reached 95.5% and 86.5%. This research can provide support technology for behavior recognition in dairy cows on farms.

Abstract:With the rapid development of intelligent animal husbandry, bovine face recognition has become the key to intelligent cattle breeding, but the problem of bovine face occlusion in practical application scenarios is more serious, which brings challenges to the performance of the recognition system. To solve this problem, a two-branch network structure based on occlude-assisted bovine face recognition was proposed. Firstly, an improved lightweight U-Net occlusion segmentation model was designed. By adding deep separable convolution and multi-scale mixing pool module, the occlusion segmentation performance of the segmentation network was effectively improved. Secondly, in order to better attenuate the influence of occlusions on bovine face recognition performance, a multilevel mask generation unit was introduced, and masks corresponding to different stages of the recognition network were constructed with different levels of occlusions as input. The damaged feature information caused by occlusions was effectively eliminated in each stage of feature extraction through mask operation. Finally, for the validity and real-time performance of the detection algorithm, the algorithm was verified on the selfmade data set, and compared with a variety of recent typical recognition algorithms. The experimental results showed that the proposed algorithm had an average accuracy of 86.34% on the blocked cow face data set, and the recognition speed was 54 f/s. Compared with the single-scale mask, the average accuracy of multistage mask was improved by 2.02 percentage points, and the recognition effect was better than that of the comparison network under different degrees of occlusion.

Abstract:In large-scale broiler farms, the behavior of broilers is usually observed and analyzed by feeders or professional veterinarians to determine their health status and breeding environment status. However, this method is time-consuming and subjective. In addition, in caged environments, due to the high density of chickens and serious mutual occlusion, the visual features of behavior are not obvious, and traditional detection algorithms cannot accurately identify the behavior characteristics of chickens. Therefore, an improved object detection algorithm for behavior detection of caged white-feather broilers was proposed. The proposed algorithm consisted of two modules: multi-scale detail feature fusion module (MDF) and object relation inference module (ORI). The multi-scale detail feature module fully utilized and extracted the multi-scale detail features contained in the shallow feature maps of the feature extraction network, and integrated them into the corresponding feature maps responsible for detection at the corresponding scale, achieving effective transmission and supplementation of detail features. The relational reasoning module fully utilized the positional relationships between objects for inference and judgment, enabling the model to more fully utilize the potential relationships between objects to assist in detection. To verify the effectiveness of the proposed algorithm, a large number of comparative experiments on both authoritative public datasets in the field of object detection and self-built behavior detection datasets in real large-scale caged white-feather broiler breeding environments was conducted. The experimental results showed that the proposed improved algorithm achieved the best detection accuracy compared with other state-of-the-art models, both in the COCO dataset and the self-built dataset. For the detection of behaviors such as feeding, drinking, moving, and opening the mouth, which were crucial for the health status of broiler chickens, the algorithm achieved accuracy rates of 99.6%, 98.7%, 99.2%, and 98.3% respectively.

Abstract:Poultry behavior is closely related to its physiological state, and behavioral data can be used to assess the health status of poultry. Statistical individual behavioral data is needed for laying hen behavioral identification and individual identification, to address the behavioral identification process, laying hen body size was small, aggregation of shade, breeding environment lighting changes and other factors resulting in the laying hen effective features expression was insufficient, individual behavioral identification effect was not ideal problem, based on the YOLO v8n network to build behavioral identification model, while fusing ODConv, GhostBottleneck, GAM attention and Inner-IoU structure, and the model was improved by reducing image feature loss, amplifying global interaction information, fusing crossstage features, and enhancing the feature extraction and generalization ability, which improved the recognition accuracy of five behaviors of laying hens, namely, feeding, drinking, standing, feather arranging, and stooping to search. Meanwhile, the individual identification network was constructed based on the YOLO v8n model, and the individual identification network model was optimized by introducing the MobileNetV3 module, which improved the statistical efficiency of individual behavioral data. The experimental results showed that the optimized behavior identification model achieved 94.4%, 93%, 90.7%, 91.7%, 86.9% average precision (AP) for the recognition of feeding, drinking, standing, feather arranging, and stooping searching behaviors, respectively, and 91.4% mean average precision (mAP), which was comparable to that of YOLO v5n, YOLO v6n, and YOLO v7-tiny, YOLO v8n, the mean average precision mean (mAP) was increased by 4.8, 4.1, 5.5, and 3.5 percentage points, respectively;the number of parameters and the amount of operations of the individual identification model were reduced by 1.9651×106 and 6.1×109 compared with that of the YOLO v8n model.It was found that by analyzing the behavioral data of the laying hens, the behavioral data were related to the temperature and the individual laying hens themselves, and that when the temperature was decreased, the number of feeding and standing was increased, the number of drinking was decreased, the number of finishing feathers and stooping to search almost did not change, the behavioral data of different individual laying hens varied greatly at the same temperature, and the value of the difference was related to the body size of the laying hens. The results of the experiment laid the foundation for judging the health status of laying hens based on behavioral data, precision breeding on farms and preferential selection of individual laying hens.

Abstract:Using machine vision technology to identify underwater crab targets is one of the effective ways to achieve intelligent crab farming equipment. However, river crab detection methods face challenges in the difficulty of target detection in underwater environments, limited feature information and high complexity of mainstream target detection models. To solve these challenges, a lightweight river crab detection model GC-YOLO v5s (GhostNetV2-CBAM-YOLO v5s) was proposed. These specific enhancements were as follows: an improved image enhancement algorithm was used to preprocess underwater crab images to improve the detection accuracy;in order to reduce model complexity, a G3 module based on GhostNetV2 was proposed to improve the feature extraction network of the model, and Ghost convolution was used to further lightweight the model;the convolution block attention module (CBAM) was introduced to solve the challenge of extracting deep features within underwater environments, which were integrated into the feature extraction network. The experimental results demonstrated the improved model’s mAP50, recall, and precision on the test set, reaching 95.61%, 97.03% and 96.94%, respectively. These metrics displayed enhancements of 2.80 percentage points, 2.25 percentage points and 2.28 percentage points compared with the baseline. Moreover, GC-YOLO v5s parameters, computations, and model size were only 69.1%, 56.3%, and 58.3% of YOLO v5s respectively. Comparative trials against mainstream object detection algorithms showcased the superiority in accuracy and model complexity. While slightly trailing YOLO v5s in detect speed, GC-YOLO achieved 104f/s.

Abstract:In order to achieve rapid and accurate identification of salmon in complex underwater environments, a lightweight salmon detection model, YOLO v7-CSMRep, was proposed based on YOLO v7. Firstly, by adopting the Stem module, the first four convolutional operations in the backbone layer were merged into an efficient convolutional operation, reducing the computational load of the model. Secondly, the ELAN and ELAN-H modules of the YOLO v7 network were replaced with the multi-directional reparameterization (MRep) module, which enhanced the one-way feature extraction capability while greatly reducing parameters and calculations. Finally, at the end of the backbone layer, the convolutional block attention module (CBAM) was integrated to enhance the network’s spatial and channel feature extraction capabilities. The experimental results showed that the improved model’s volume, parameter count, and computational load were reduced by 4.28%, 5.29% and 31.30%, respectively. The F1 score and mAP0.5 were increased by 0.5 and 0.7 percentage points, and reached 93.1% and 97.1%, respectively. Additionally, the frame rate was increased by 15.41%, and reached 140.8f/s. Compared with that of YOLO v5s, YOLO v6s, YOLO v7, YOLO v7-tiny, and YOLO v8s models, the mAP0.5 was improved by 1.0, 2.0, 0.7, 0.8, and 1.2 percentage points, respectively. Therefore, the method proposed can rapidly and accurately identify salmon and provide technical support for biomass monitoring in deep-sea aquaculture.

Abstract:Underwater biological target detection is a crucial technology for achieving automation in underwater robotic fishing. Aiming to address issues such as object overlap, occlusion, and false detections, missed detections caused by small object scales in underwater biological object detection tasks, an underwater biological object detection algorithm, FDC-YOLO v8 was proposed based on an improved YOLO v8n. Firstly, the FDC module was incorporated, which utilized deformable convolution networks in the backbone network to enhance the model’s feature extraction capability and enrich the diversity of extracted features. Secondly, the FrSAConv module, integrating fractional Fourier transform and spatial attention mechanism, was introduced to further separate diverse object features and enhance the model’s perceptual ability towards various features. Finally, the Wise-IoU loss function was introduced as the bounding box loss function to better address issues related to object imbalance and scale differences. The experiments were conducted by using the RUIE dataset, which included four types of underwater organisms: echinus, starfish, holothurian, and scallops. Experimental results demonstrated that the improved FDC-YOLO v8 achieved an mAP of 85.3%, a 2.6 percentage points improvement over the baseline model. The inference speed can reach 769 frames per second, showcasing better performance in underwater object detection of marine organisms with challenged such as object overlap, occlusion, and small-scale objects.

Abstract:Precise feeding technology for fish ingestion is a key technology to achieve intelligent aquaculture. However, most of the precise feeding model is based on indoor aquaculture ponds with clear water quality, which are not suitable for outdoor open farming environments. In view of the actual situation, a set of detailed open pond dataset through water perspective acquisition was constructed, and the dataset was augmented to increase its diversity, and then the BiLSTM bidirectional recurrent neural network was embeded on the basis of the lightweight neural network MobileViT, so as to improve the memory ability of the model for video sequence data in a long period of time, and the CBAM attention module was combined with the MV2 module to design the CBAM-MV2 module, and then the CBAM-MV2 module was added to different layers of the model for experiments to obtain the most reasonable improvement scheme. Finally, an improved MobileViT-CBAM-BiLSTM fish feeding behavior classification model was proposed, which improved the prediction ability, robustness and generalization performance of the model, and realized the three classification of fish feeding behavior. The experimental results showed that the improved MobileViT was significantly better than previous in the collected video frame dataset, with an accuracy of 98.61%, 98.79% for Macro-F1, which was 6.33 percentage points for accuracy, 6.75 percentage points for Macro-F1 compared with the original MobileViT.

Abstract:Phenotypic parameters of plants are quantitatively indicated, describing the morphology, structure, and physiological characteristics of plants, unveiling the growth patterns and relationships with environmental factors. Issues such as significant data errors, plant damage, high costs, and extensive data volume were exhibited by existing manual measurement and laser scanning-based methods for extracting plant phenotypic parameters. Therefore, an automatic extraction method for phenotypic parameters of Anthurium andraeanum Linden plants based on YOLO v8 and CycleGAN was proposed. The method included the follows: YOLO v8 was enhanced with the convolutional block attention module to improve the model’s feature extraction capabilities for detecting and segmenting Anthurium andraeanum Linden leaves;the Grabcut algorithm was utilized to eliminate background features from segmented images, and the VGG model was employed for classification to distinguish intact and missing Anthurium andraeanum Linden leaves;the convolutional block attention module and feature pyramid network were introduced into the CycleGAN generator to enhance multi-scale feature extraction capabilities, incorporating the SmoothL1 loss function to enhance model stability and repair missing Anthurium andraeanum Linden leaves;a phenotypic parameters extraction algorithm (PPEA) was proposed to automatically extract leaf length, leaf width, and leaf area of Anthurium andraeanum Linden plants. The proposed methods were compared and analyzed by using a dataset of 650 self-collected images. Experimental results demonstrated the effectiveness of the proposed approach in automatically extracting phenotypic parameters of Anthurium andraeanum Linden plants.

Abstract:In the face of challenges such as complex terrains, fluctuating lighting, and unstructured environments, modern orchard robots require the efficient processing of a vast array of environmental information. Traditional algorithms that sequentially execute multiple single tasks are limited by computational power which are unable to meet these demands. Aiming to address the requirements for realtime performance and accuracy in multitasking autonomous driving robots within dragon fruit orchard environments. Building upon the YOLOP, focus attention convolution module was introduced, C2F and SPPF modules were employed, and the loss function for segmentation tasks was optimized, culminating in the OrchardYOLOP. Experiments demonstrated that OrchardYOLOP achieved a precision of 84.1% in target detection tasks, an mIoU of 89.7% in drivable area segmentation tasks, and an mIoU increased to 90.8% in fruit tree region segmentation tasks, with an inference speed of 33.33 frames per second and a parameter count of only 9.67×106. Compared with the YOLOP algorithm, not only did it meet the real-time requirements in terms of speed, but also it significantly improved accuracy, addressing key issues in multi-task visual perception in dragon fruit orchards and providing an effective solution for multi-task autonomous driving visual perception in unstructured environments.

Abstract:In order to realize the adaptive grading conveyance and real-time monitoring of harvesting status in the process of Panax notoginseng combined harvesting operation, focusing on the characteristics of Panax notoginseng root-soil complex and the complex field harvesting conditions, a Panax notoginseng object detection method based on YOLO v8s and suitable for deployment on the Jetson Nano was proposed. Based on the accurate recognition of Panax notoginseng by YOLO v8s, the channel pruning algorithm was utilized to formulate a corresponding pruning strategies for its new model structural characteristics, which ensured the accuracy and improved the real-time detection performance at the same time. The improved model was deployed to Jetson Nano by using the TensorRT inference acceleration framework, which realized the flexible deployment of the Panax notoginseng object detection model. The experimental results showed that the mean average precision of the improved PN-YOLO v8s-Pruned model on the host side was 93.71%, although it was decreased by 0.94 percentage points compared with that of the original model, the number of parameters, computational complexity, and model size were 39.75%, 57.69%, and 40.25% of the original model, respectively, and the detection speed was increased by 44.26%. Compared with other models, the improved model demonstrated superior comprehensive detection performance in terms of computational complexity, detection accuracy, and real-time performance. After deployment at the Jetson Nano, the improved model had a detection speed of 18.9 frames per second, which was 2.7 times higher than before acceleration and 5.8 frames per second higher than the original model, and the deployment detection effect was better than the original model. The results of the bench tests showed that the mean average precision of Panax notoginseng detection was more than 87% under four conveyor separation harvesting conditions. The average accuracy of the Panax notoginseng counting under different conveyor separation harvesting conditions and different flow levels reached 92.61% and 91.76%, respectively. The field test results showed that the mean average precision of Panax notoginseng detection was more than 84%, and the average accuracy of the Panax notoginseng counting reached 88.11%, which could meet the detection requirements of Panax notoginseng under complex field harvesting conditions, and could provide technical support for the monitoring system of harvesting quality and the adaptive grading transportation system of combined harvesting operation based on edge computing equipments.

Abstract:Detecting missed tassels is crucial for assessing the quality of aritificial emasculation in maize seed production fields. Aiming at the problems of large parameter quantity, low detection efficiency and poor accuracy of the existing maize tassel detection models, a lightweight tassel detection model based on RTMDet-tiny, named MLCE-RTMDet, was proposed. The model used the lightweight MobileNetv3 as the feature extraction network to effectively reduce the model parameters. The CBAM attention module in the neck network was integrated to enhance multi-scale feature extraction capability for tassel objects, overcoming potential performance losses caused by the lightweight networks. Simultaneously, the EIOU Loss was adopted, replacing the GIOU Loss, which further improved the accuracy of tassel detection. Experiments on the self-built dataset showed that the improved MLCE-RTMDet model reduced model parameters to 3.9×106, while the number of floating point operations was lowered to 5.3×109, resulting in a 20.4% reduction in parameters and a 34.6% decrease in computational complexity compared with that of the original model. When evaluated on the test set, the model’s mean average precision (mAP) reached 92.2%, reflecting a 1.2 percentage points improvement over the original model. The inference speed was increased to 41.9 frames per second (FPS), representing a 12.6% enhancement. Compared with current mainstream detection models such as YOLO v6, YOLO v8, and YOLO X, MLCE-RTMDet demonstrated superior overall detection performance. The improved high-accuracy lightweight model offered technical support for tassel re-inspection and emasculation quality assessment in maize seed production fields following artificial emasculation.

Abstract:As an important feature of plants, leaf veins contain physiological and genetic information. Aiming at the problems of blurred edge segmentation and low segmentation accuracy of small veins in complex leaf texture state, a GAN-SA-UNet vein segmentation algorithm was proposed with tobacco leaves as the research object. The spectral information of veins and leaves was obtained by hyperspectral imaging technology, and the principal component analysis ( PCA ) was used to reduce the dimension and obtain the composite map. On this basis, the spatial attention mechanismwas introduced to capture the key spatial features and improve the segmentation accuracy. At the same time, the adversarial network was introduced to optimize the generated results and improve the robustness of vein segmentation. The results showed that the interpretation rate of the first three principal components of PCA of leaf vein and leaf surface spectrum was 95.71%, and the spectral characteristics of the two after dimension reduction showed obvious separability. The first three principal components composite map could highlight the difference between leaf surface and leaf vein, and highlight the characteristics of leaf vein. The GAN-SA-UNet segmentation algorithm can capture the vein features in complex leaf texture images. The segmentation accuracy and intersection over union were 98.93% and 66.23%, respectively. Compared with the original model, they were increased by 0.18 percentage points and 4.21 percentage points, respectively. The inference time of single image was 4ms. It showed strong generalization ability and efficient and accurate recognition ability in the verification test of different producing areas, parts, grades and types of tobacco leaves.

Abstract:In view of its advantages such as small ground pressure, good climbing performance and flexible turning, agricultural track chassis is currently a mobile agricultural power machinery widely recognized by farmers, which has been widely used in various aspects of agricultural production such as farming, planting, field management, harvesting, transportation, and gradually developing in the direction of automation and intelligence. The development status at home and abroad was mainly expounded from the application of agricultural track chassis, stability theory and control technology, drive system and steering technology, autonomous navigation and intelligent control technology, and agricultural track chassis-soil interaction theory. The application progress of stable leveling, efficient transmission, smooth steering and autonomous driving in agricultural track chassis was summarized. Combined with different agricultural operations, the application characteristics of agricultural track chassis in related fields were illustrated. At last, according to the current and future needs of agricultural track chassis in China, the future development direction of agricultural track chassis was prospected from the aspects of strengthening the optimization of high-stability walking system, creating efficient transmission and flexible steering system, overcoming the technology of autonomous driving and intelligent control, and researching the basic theory of track-soil system. The research result can provide a good reference for the future technical research of agricultural track chassis.

Abstract:Aiming at the issues such as prolonged path planning time, low efficiency and poor success rate of the picking manipulator in the apple picking task as a consequence of the complex natural picking environment, an improved fusion and switching path dynamic planning algorithm was proposed. The algorithm introduced a dynamic threshold goal bias sampling strategy and artificial potential field to alter the generation position of new nodes, increasing the purposiveness of sampling and improving convergence speed. A relative distance was incorporated into the repulsive potential field coefficient to overcome the problem of unreachable targets by considering the distance to the goal. To enhance the algorithm’s robustness, a threshold was set to partition the spatial region, dynamically switching to the failure-guided adaptive sampling region RRT algorithm (FGA-RRT) based on the current node expansion state to address narrow passage issues and increase planning success rates. The greedy algorithm was utilized to optimize the resulting path tree, removing redundant nodes, further shortening the path length, and optimizing path smoothness to ensure the stable movement of the picking robot arm. Simulation experiments were conducted for the RRT algorithm, RRT* algorithm, GB-RRT algorithm, common fusion algorithm and the improved fusion and switching algorithm respectively in simple obstacles, narrow channels, complex obstacles and simple three-dimensional spaces. The results showed that the improved fusion and switching algorithm had good adaptability in different environments, with high planning efficiency, few iterations and high path quality. Based on the established 6-DOF robot arm motion planning simulation environment and laboratory environment, obstacle avoidance picking tests were conducted. The improved hybrid switching algorithm’s picking efficiency was increased by 74.74%, path length was decreased by 32.03%, and picking success rate was improved by 8 percentage points compared with that of the RRT algorithm. The experimental results demonstrated that the proposed algorithm had stronger search capabilities in apple-picking scenarios, providing a reference for improving the operational efficiency of picking robot arms.

Abstract:In order to solve the problem that the magnitude and frequency of excitation force output by existing exciter cannot be adjusted independently, the amplitude and frequency of excitation force cannot meet the demand of fruit vibration harvesting at the same time, TRIZ theory was used to construct the model of exciter and fruit tree matter-field. According to the matter-field model and the physical conflict solving method, the eccentric block layout of the excitation force amplitude and frequency of the eccentric block exciter was obtained, and the schemes of the four eccentric block exciter and the radius adjustable exciter were proposed. By comparing the dynamic performance of the two designs, it was found that the moment of inertia of the radius adjustable exciter was lower than that of the four eccentric blocks. The smaller the moment of inertia of the shaker was, the smaller the torque and power required during the rotation of the shaker was. When the four eccentric blocks were rotating, the phase angle size of each eccentric block was changed with time, so it was very difficult to adjust the size of the phase angle φ. In the radius adjustable exciter, the eccentric block and the rotating axis were relatively fixed in the radial direction, and the eccentric block radius was relatively static in the axial direction, and it was relatively easy to adjust the eccentric block radius in the axial direction. Therefore, the radius adjustable exciter was more suitable for the amplitude-frequency regulation mechanism of exciting force in the process of fruit vibration harvesting. Three different frequencies and amplitudes were applied to citrus trees. Citrus trees were used as test subjects to apply three different frequencies and amplitudes of excitation forces to different fruit trees with adjustable radius exciter. The results showed that the detachment rate of fruit was increased with the increase of the frequency of excitation force. At the same time, when the exciting frequency was constant, the amplitude of the exciting force was increased, and the detachment rate of the fruit also showed an increasing trend. When the exciting frequency was 19Hz and the stroke of the screw nut was 70mm, the average detachment rate of the fruit was 86.3%. The results showed that the effect of excitation frequency on the detachment rate was greater than that of excitation force amplitude. The radius adjustable exciter can change the amplitude of the exciting force and improve the fruit harvest rate. At the same time, the amplitude-frequency independent regulation characteristic of the radius adjustable exciter can reduce the amplitude to prevent the damage of fruit trees when the excitation force frequency was large.

Abstract:In order to obtain the discrete element simulation parameters of coastal saline soils, the soil discrete element parameters were calibrated by combining the experiment and discrete element simulation using the soil of the Yellow River Deltd Agricultural Highland Area as an example, and Hertz-Mindlin with JKR in EDEM was selected as the simulation contact model. The significance analysis was carried out through the Plackett-Burman test to explore the factors that had a significant effect on the soil stacking angle, the steepest climb method was used to further determine the range of factor values, the Box-Behnken test was applied to establish a quadratic polynomial regression model of the three significance factors and the simulated stacking angle of the soil, and the regression model was performed with the measured soil stacking angle of 33.6° as the target. The optimal combination of soil-soil static friction factor of 0.546, soil-soil recovery coefficient of 0.358, and soil surface energy of 3.207J/m2 in the JKR model was obtained. Under the conditions of the optimal parameter combination, the simulation results of the outer diameter of the top of the hole and the longitudinal depth of the hole had an error of 4.04% and 3.47% from the test, respectively, which verified the accuracy of the soil parameter calibration. The discrete element parameter calibration method and parameter values proposed for saline alkali soil can provide theoretical basis and support for the discrete element simulation of the interaction between soil contact components and soil under saline alkali working conditions, as well as the research and design of specialized agricultural machinery for saline alkali soil.

Abstract:In view of the lack of cotton field fertilization combined soil preparation machinery and the low control precision of fertilizer application in Xinjiang, a variable fertilization combined soil preparation machine was designed, which could complete the operations of harrowing, variable fertilization, soil leveling, soil crushing and suppression at one time. The key components such as fertilizer discharge device, double disc trenching mechanism, notch rake group, soil crushing and pressing components were designed. The key factors affecting its working performance were obtained through static analysis, and the structural parameters of key components were determined. Based on the Beidou automatic navigation driving system, the driving speed was obtained, and the variable fertilization control was realized with STM32F405 as the core processor. The field experiment was carried out with the consistency of fertilizer amount in each row, the control accuracy of fertilizer amount at different driving speeds and the qualified rate of fertilization depth as the evaluation indexes of fertilization performance, and the stability of rake depth, standard deviation of surface flatness and rate of broken soil as the evaluation indexes of soil preparation performance. The test results showed that when the rotation speed of the fertilizer shaft was 10~60r/min, the maximum coefficient of variation of the consistency of each row of fertilizer was 4.27%. When the driving speed was 4km/h, 7km/h and 10km/h, the minimum control precision of fertilizer discharge was 96.70%, 95.35% and 94.14%, respectively. The qualified rate of fertilization depth was 87.04%. The maximum coefficient of variation of rake depth stability was 5.34%. The maximum standard deviation of surface roughness was 4.69mm. The broken soil rate was 96.16%. All indexes met the requirements of fertilization machinery and soil preparation machinery standards.

Abstract:During the operation of the seeder, the discharge device will experience non-stationary random vibration, which significantly affects seed discharge performance and holds great importance for acquiring and analyzing vibration signals. A denoising method was proposed that combined dragonfly algorithm (DA), variational mode decomposition (VMD), and wavelet threshold to continuously update the location and speed of dragonfly individuals through iterative processes. The optimal parameter combination for VMD decomposition effect was determined. A simulated random road signal in the time domain served as the initial signal and underwent denoising by using DA-VMD combined wavelet threshold, wavelet threshold denoising, empirical mode decomposition (EMD), VMD, and wavelet combined EMD methods respectively. The results demonstrated that the proposed method achieved superior denoising effects on non-stationary random vibration signals with post-denoising signal-to-noise ratio, root-mean-square value, and correlation number measuring 21.570, 0.094, and 0.833, respectively. Furthermore, vibration signals from seeders under different surface conditions and operating speeds during field seeding were collected and subjected to denoising by using the DA-VMD combined wavelet threshold denoising method. The effectiveness of denoising was evaluated based on smoothness index, signal energy ratio, and noise mode indices. The results indicated smoother signals with higher signal energy ratios after denoising across various working conditions.

Abstract:It is difficult to guarantee the stability of soil compaction system under the condition of high speed no-tillage seeding. Therefore, a technical scheme of automatic soil compaction control was proposed, and the electro-hydraulic control system for soil compaction of maize planter was designed. The overall structure of the system was proposed, and the pressure control process was determined through the mechanical analysis of the movement of the seeding monomer and the interaction between suppression wheel and soil. The electro-hydraulic control system design and hardware selection were carried out. AMEsim simulation analysis and step response test were used respectively to design the hydraulic actuator and electronic control system. The results showed that compared with the traditional mechanical regulation, the adjustment accuracy and stability of the hydraulic actuator using PID control were improved. The mean adjustment time of the control system for soil compaction was 1.9s, the mean steady-state error was 1.9N, and the mean overshoot was 2.0%. It was obviously better than mechanical control. Sinusoidal surface soil trough verification test was adopted, and comparative experimental study was carried out on the influence of different adjustment methods and surface conditions on the stability of soil compaction. The results showed that when the target pressure was 300N, the RMSE of the pressure using electro-hydraulic control was 30.1% lower on average and the variation coefficient of soil compaction was 24.46 percentage points lower than that by using traditional mechanical control. When the maximum surface displacement in the vertical direction of sine curve was 0mm, 20mm and 40mm, respectively, the variation of RMSE was not significant with the increase of the target force, and the maximum difference was 39.2N. The automatic soil compaction control system of maize planter based on electro-hydraulic control can guarantee the compaction operation quality under different work conditions. It provided technical and equipment support for the construction of seed bed and maize sowing under the condition of high speed and no tillage in arid region.

Abstract:In order to further improve the accuracy and stability of the longitudinal feeding mechanism of vegetable pot seedling automatic transplanter,a four-link longitudinal feeding mechanism driven by CAM link and sector gear was proposed.First of all,by analyzing the process of longitudinal disk feeding,the design requirements and overall scheme of the mechanism were determined,the ideal longitudinal disk feeding trajectory was planned,the comprehensive model of longitudinal disk feeding mechanism trajectory was established,and the preliminary solution was carried out by genetic algorithm.Then,based on Matlab GUI,the visualization computer aided optimization analysis software of the vertical disk feeding mechanism was developed,and the design parameters of the vertical disk feeding mechanism satisfying the design requirements were obtained.Finally,the structure design,simulation analysis and test verification of the longitudinal disk feeding mechanism were carried out.The results showed that the actual trajectory of the test was basically consistent with the theory and simulation trajectory.When the transplanting efficiency was 80 plants/(min·row) and 90 plants/(min·row),the single feeding error of the mechanism was within±1.5mm,and there was no cumulative error after multiple feeding.At the same time,it was observed that the longitudinal feeding mechanism and the seedling taking mechanism were well coordinated through the seedling taking test,which verified the rationality of the mechanism design.

Abstract:Aiming at the problem that the reverse design of the non-circular gear train can not achieve the single-row two-stage gear ratio distribution, an optimal subgear ratio solution method was proposed, and a small plant spacing light simple non-circular gear train planting mechanism was designed. The poses of the planting mechanism at four key moments, namely, embedding, planting, exhuming and grafting, were selected on the ideal seedling trajectory of vegetable small plant distance transplanting. A finite separable four-position kinematic solution model for a planar 2R open-chain linkage assembly was established, and the initial design parameters of the mechanism were obtained. Based on the kinematic model of the seedling planting mechanism with non-circular gear planetary gear train, the optimization objective function was determined, and the improved whale optimization algorithm was used to solve the objective function. A visualization assistant software for the non-circular gear planetary gear train planting mechanism was built, and finally, a seedling planting mechanism with an optimized trajectory was designed. The non-circular gear planetary gear train seedling planting mechanism was three-dimensionally modeled and topology optimized, and kinematic simulations were conducted. The simulated trajectory was compared with the theoretical trajectory to verify the correctness of the three-dimensional design. Finally, field transplanting experiments were conducted. Under the conditions of 30r/min for the mechanism speed and 120mm/s for the machine forward speed, the measured planting spacing was 120mm, which was consistent with the theoretical design spacing. The planting efficiency was 60 seedlings/min, and the average seedling planting success rate over five trials was 95.4%, indicating that the seedling planting mechanism had high operational performance and practicality.

Abstract:In response to the shortage of self-propelled Panax notoginseng combine harvesters, poor overall passability, and weak driving stability, a tracked self-propelled Panax notoginseng combine harvester was designed and research on overall driving passability and stability under hilly and sticky soil conditions was conducted. Firstly, the walking system of the self-propelled Panax notoginseng combine harvester was designed. Secondly, by using the centroid intersection method, the centroid range of the entire machine for stable driving under various working conditions was obtained, and the overall layout plan was determined. Finally, simulation and field experiments were conducted on the self-propelled Panax notoginseng combine harvester under four operating conditions: straight, turning, climbing, and obstacle crossing. The simulation results showed that the combined harvester had good straight and unilateral braking steering performance, and met the design requirements for stable passage through longitudinal slopes of 20° and below, transverse slopes of 15° and below, 350mm field ridges, and 900mm trenches. The field test results showed that the straight deviation rate of the combine harvester was 3.79%, the minimum turning radius was 1480mm, and it can stably pass through longitudinal slopes of 20° and below, transverse slopes of 15° and below, 350mm field ridges, and 900mm trenches. The variation laws of its motion posture, pitch angle, and rolling angle curves were consistent with the simulation results, verifying the accuracy of the simulation results, which demonstrated that the combine harvester had good passability and can meet design requirements. The designed tracked self-propelled Panax notoginseng combine harvester had strong passability and can achieve stable driving in sticky soil and gentle slope sections, providing a theoretical basis and reference for the design and manufacturing of root and stem combine harvesters in hilly and mountainous areas.

Abstract:To address the issues of high labor intensity, low efficiency, and lack of mechanical equipment in the harvesting of sandy willow growing in complex sandy terrains and dispersed distributions, the theoretical analysis, orthogonal experiments, response surface analysis, and field test methods were used, a sandy willow cutting harvester with seven degrees of freedom and a counter-rotating dual saw blade cutting device were designed. Subsquently, the prototype was fabricated and subjected to field performance tests. The harvester was driven by a hydraulic system, and it mainly comprised a frame, a walking mechanism, a cutting device, a telescopic device, and a rotating device, etc. The orthogonal experiments results and response surface analysis indicated that the saw blade rotate speed, feed rate, and number of saw blade teeth were key factors affecting the cutting force and power. The number of saw blade teeth and saw blade rotate speed significantly influenced the cutting force and power, as well as the notable interactions between the saw blade teeth and rotate speed, saw blade teeth and feed rate. The field test results showed that, when the cutting device operating parameters were saw blade tooth number of 120, feeding speed of 10mm/s, and sawing speed of 1400r/min the average stubble breakage rate, missed cut rate, re-cut rate, and stubble height qualification rate of the sandy willow harvester were 4.02%, 4.19%, 0, and 94.33%, respectively, meeting the technical requirements for mechanized sandy willow cutting operations.

Abstract:Aiming at the problem of mixture separation with significant differences in various shapes after the Potosia brevitarsis larva biotransformation the residual film mixture to meet the premise of biological activity, a segmented drum type pneumatically assisted Potosia brevitarsis larva biotransformation mixture separation device was designed. Through physical experiments and theoretical calculations, the resistance coefficient partitions and theoretical suspension speeds of Potosia brevitarsis larva, frass, and residual films were determined. Its accuracy was verified by using a suspension speed bench test, which can provide data basis for subsequent fluid-structure interaction simulations. Through theoretical analysis and EDEM-Fluent coupling method, the screening process of residual film, larvae and frass was simulated, and the main structural parameters and working parameters of the device were determined. The rotation speed of the drum screen, the inclination angle of the drum screen and the fan wind speed were selected as test factors. The residual film impurity content rate and the frass film content rate were tested as test indicators. A single factor test was conducted to determine the reasonable range of the levels of each factor. Based on the results of the single factor test, a three-factor and three-level quadratic regression response surface test was designed and a regression model was established. The test results showed that the order of factors affecting the impurity content of the residual film was fan speed, drum screen inclination angle, and drum screen speed. The order of factors affecting the film content of frass was drum screen rotation speed, drum screen inclination angle, and fan wind speed. After optimization, the optimal working parameter combination was drum screen rotation speed of 21.79r/min, drum screen inclination angle of 3.58°, and fan wind speed of 5.52m/s. The material screening test was carried out with this parameter combination, and the average impurity content rate of the residual film and the film content rate of the frass were 8.96% and 1.52%, respectively. The relative errors with the theoretical optimization values were less than 5%. The research can provide a reference for the design of separation devices for mixtures containing biological activity and significant differences in shape.

Abstract:Aiming at the problems of unreasonable distribution of wind field and large loss of liquid, an air delivery system of tower sprayer was designed and its structural parameters were optimized. ANSYS software was used to simulate the air delivery system of sprayer, and the optimal parameters of the rectifier plate structure, deflector taper and deflector on both sides were determined by comparing the wind speed difference of the two sides outlet and the wind speed of the upper outlet under different structural parameters. The experimental results showed that the influence of the structure parameters of the rectifier on the symmetry of wind field on both sides of the sprayer from large to small was as follows: guide vane angle, guide vane number and guide vane length. The diversion table greatly improved the symmetry of wind field. The length and angle of the baffle had great influence on the wind speed of the upper outlet. When the number of guide blades of the rectifier plate was 12, the length was 200mm, the angle was 0°, and the taper of the guide table was 60°, when the contraction angle of air duct 1 was 2°, the symmetry of wind field and vertical distribution uniformity on both sides of the sprayer were optimized. According to the optimal parameters, the sprayer air delivery system was set up and the wind speed test was carried out. The results showed that the relative error between the simulated wind speed and the test value was less than 10%, and the simulation results were reliable. Field experiments showed that the difference of droplet deposition density on both sides of the sprayer was less than 9%, that was, the droplet deposition on both sides was uniform. The average density of droplets deposited in the canopy of fruit trees was 75.69 particles/cm2, which indicated that the optimized tower sprayer had good droplet penetration. The vertical distribution of fog droplets in the lower layer was the largest, followed by the middle layer and the smallest in the upper layer, which was consistent with the distribution of high spindle tree canopy. The symmetry of the wind field and droplet deposition characteristics on both sides of the optimized tower sprayer was high, and the research results can provide reference for the optimization of the air delivery system of tower sprayer.

Abstract:As an important energy storage technology, pumped storage power plants have both pump and turbine operating modes. In order to meet the requirements of energy collection and grid scheduling, the pump turbine operating conditions change frequently, the number of start-ups and shutdowns increases, and the instability of the unit is prominent. Due to the lack of detailed knowledge of the operating parameters such as pressure in the mechanical action of the unit, the boundary conditions are set differently from the actual operating conditions, resulting in the discrepancy between the field test and the numerical simulation and model test, especially the increase in the complexity of the unit’s characteristic parameters during the start-stop transient process. A prototype energy storage unit was tested in the field, and pressure and acceleration sensors were installed on each component of the unit to obtain pressure signals and vibration acceleration signals, and based on the experimental data collected in the field, the operating characteristics of the unit were analyzed in the start-up and shutdown processes of the turbine operating conditions. The results showed that during the hydraulic turbine start-up process, the instantaneous stability of the start-up was good, after which the pressure pulsation of the unit during the runner acceleration stage was manifested as a broadband noise, and the pressure pulsation during the runner deceleration stage was manifested as the impeller rotational frequency and its octave frequency. Under no load with fixed guide vane opening, the periodic pressure oscillations at each measurement point of the unit disappeared which accompanied by changes in the amplitude of the mixed frequency. Under the governor dynamic mode, the frequency component of the unit was dominated by the natural frequency. During the turbine shutdown process, the pressure pulsation was insensitive to the guide vane action, and the intensity of the pulsation was gradually decreased with the speed reduction. The lower frame and top cover vibrated more strongly by the mechanical action, after which they were decreased with the decrease of rotational speed. The top cover modal frequency line, which was independent of the speed change, still existed after the butterfly valve was closed.

Abstract:The research on the trade-off synergy relationship of multifunctional cultivated land is of great significance for protecting cultivated land resources, ensuring food security, maintaining ecological security, and promoting rural revitalization. Taking 105 counties in Henan Province as the research object, and methods such as Spearman rank correlation analysis and K-means clustering analysis were used to study the multifunctional level of cultivated land in Henan Province from 2000 to 2020 and its trade-off and synergy relationship. The results showed that from 2000 to 2020, the multifunctional index of cultivated land in Henan Province showed an overall upward trend, showing a spatial pattern of “high in the east and low in the west”, and different spatiotemporal differentiation characteristics were observed among various functions. The grain production function in the eastern region was strong, while the ecological service function in the western region was excellent. The overall economic contribution function was declining, and although the social security function was weak, it was gradually improving. The spatiotemporal differences in the trade-off synergy relationship of multifunctional cultivated land were significant, and the interactions between multiple functions showed a characteristic of weakened synergy and intensified trade-offs, especially the significant trade-off between food production and economic contribution functions, as well as ecological service functions. Based on the dominant functional types of cultivated land and the characteristics of multifunctional coupling coordination, Henan Province was divided into four types of functional zones: modern agricultural demonstration zone, urban agricultural leisure zone, ecological agricultural construction zone, and modern agricultural construction zone, and differentiated control measures were proposed. The research result had certain theoretical and practical significance for guiding the rational allocation of cultivated land in major grain producing areas and improving the efficiency of diversified utilization of cultivated land.

Abstract:Studying the spatial-temporal divergence characteristics of cultivated land use functions and spatial forms in different functional areas can provide scientific basis for sustainable management and utilization of cultivated land, as well as regional economic and social sustainable development. Taking all districts (counties) in Hubei Province as the research object, cultivated land use functions and spatial forms of different main functional areas were evaluated from 1995 to 2019 by using data from social economy, natural geography, agricultural production etc., and the spatiot-emporal evolution characteristics were analyzed. The results indicated that from 1995 to 2019, the production, social, ecological and overall functional forms of cultivated land showed characteristics of inverted “N” shaped, “W” shaped, “V” shaped, and “V” shaped changes over time respectively. In terms of spatial distribution characteristics, the production, ecological and overall functions were stronger in western Hubei than that in eastern Hubei, while the spatial differentiation of social functions was relatively small. On the main functional area, the production function was not significantly different between the key development areas and the main agricultural production areas, but it was significantly stronger than the ecological function areas;the social function was the strongest in ecological function areas and the weakest in the key development areas, while the ecological function and overall function were the strongest in the main agricultural production areas and the weakest in the ecological function areas. From 1995 to 2019, the number of cultivated land, landscape pattern, and spatial form exhibited characteristics of inverted “N” shaped, “V” shaped, and “N” shaped changes over time respectively. In terms of spatial distribution characteristics, the quantity form showed that the central Hubei region were stronger than the western and eastern Hubei regions, while there was a little difference between the western and eastern Hubei regions. The landscape pattern and spatial form showed that the central Hubei region was the highest and the western Hubei region was the lowest. The quantity form values were relatively high in the key development areas and the main agricultural production areas, and the landscape pattern and spatial form were manifested as the largest agricultural product production area, followed by key development areas, and the smallest ecological function area, and the phenomenon was very obvious. The overall form of cultivated land use showed a “V” shaped trend over time;in terms of spatial distribution characteristics, under the combined influence of cultivated land use function and spatial form, the overall form of cultivated land use showed a differentiation pattern with the highest in central Hubei and relatively lower in eastern and western Hubei. The main function were manifested as the largest agricultural product production area, followed by key development areas, and the smallest ecological function area. It was recommended that different functional zones had significant heterogeneity in social and economic conditions, functional positioning and cultivated land resource endowment, so differentiated regulatory strategies should be adopted to optimize the functional and spatial forms of cultivated land use, serving the sustainable use of cultivated land and the sustainable development of Hubei Province’s economy and society.

Abstract:Aim at the frequent soil drought disasters and the limited monitoring area of ground soil moisture monitoring stations in Henan Province, the meteorological drought index and remote sensing monitoring model were combined to predict soil drought. It was based on the standardized precipitation evapotranspiration index calculated by meteorological data from 2012 to 2021, and the drought monitoring effects of four indices of commonly used remote sensing models, namely crop water scarcity index, vegetation supply water index, temperature vegetation drought index and vegetation temperature condition index were evaluated. Taking 2019 as a typical drought year, the differences among the four indices were compared, and the spatial distribution and change trend of TVDI in Henan Province from 2012 to 2021 were analyzed. Finally, ARIMA model was used to predict soil drought in 2022. The results showed that the research results of CWSI, VSWI and VTCI were different from the actual results. Only the TVDI value was consistent with the change trend of soil moisture recorded in the field, and showed an increasing trend with time in the northwest, central and northern parts of Henan Province. The spatial evolution results showed that the coverage pixels of arid areas in 2019 accounted for 76%, which accounted for the largest proportion in this decade, and the soil moisture predicted by the ARIMA model in 2022 was consistent with the reality. On the basis of soil drought prediction, it can provide reference for the precise management of agricultural production in Henan Province.

Abstract:To determine the operation technology of root cutting and nurturing in jujube orchards in southern Xinjiang, physical parameters such as soil moisture content, solidity, porosity, bulk density, and particle size distribution were measured. The root aggregation depth and soil root content of the jujube orchard’s root cutting area were also calculated. The agricultural technical requirements were determined that the root cutting area should be located 750mm away from the tree trunk and the depth of the root cutting operation should be adjustable within the range of 0~200mm.The research results indicated that the root soil composite had anisotropy. When the sampling angles θ were 0°, 45° and 90°, the shear strength T relationship of the root soil composite showed trend of T0°

Abstract:In order to rationally develop and utilize brackish water resources, the structure and operating parameters of a continuous first-but-mix stirring reactor at the front end of an agricultural irrigation system were optimized. The effects of mixing speed and paddle width on the mixing performance were investigated by using a double-layer paddle mixer with a diameter of 290mm. Realizable k-ε model and Eulerian multiphase flow model were used for numerical simulations and validated by experimental data. The analysis of turbulent kinetic energy, velocity flow field, brackish water volume fraction distribution, mixing time and power consumption of the agitator showed that the increase in mixing speed and paddle enhanced the turbulence and flow velocity for both the threebladed open type and the combined anchor type. It was found that the rotational speed had a greater effect on the studied parameters than the paddle width. Overall, the effect of rotational speed on the studied parameter was significantly more than paddle width, in addition to the significant difference in the effect of these two factors on the two types of agitators. Compared with the three-bladed open agitator, the combined anchor agitator reduced the agitation time by 6.8%, but increased the agitation power consumption by about 10%. Taking into account of crop irrigation requirements, mixing efficiency and equipment energy consumption, it was recommended to use a three-blade open agitator with a lower speed (90r/min) and a wide paddle (50mm). This configuration can effectively balance the mixing efficiency and energy consumption of a continuous salt and freshwater mixing and stirring system. The results can provide a reference for the design optimization of continuous brackish freshwater mixing and stirring reactors.

Abstract:It is of great significance to explore the influences of climate change on crop phenology and yield of the rotation system of winter wheat and summer maize in the Huang-Huai-Hai Plain, a major grain production base in China, for guaranteeing the food security of China. The APSIM-Wheat and APSIM-Maize models (V 7.6) were calibrated and verified based on experimental data of multiple years and multiple sites, which were obtained based on literature review. Then, future meteorological data predicted by ten different global climate models (GCMs) in the CMIP6 dataset were used to drive the verified APSIM models to simulate the changes of phenology and yield of winter wheat and summer maize in the time periods of 2021—2060 (2040s) and 2061—2100 (2080s) under two greenhouse gas emission scenarios of SSP2-4.5 and SSP5-8.5. Based on the analyses with multiple linear regression and random forest model, the positive and negative effects of climatic factors and change of crop reproductive stage on crop yield were analyzed and their importance was clarified. The result showed that compared with the baseline period (1981—2020), the vegetative stage of winter wheat was shortened, the reproductive stage was prolonged, and wheat yield was increased. These changes were more obvious under the SSP5-8.5 than that under the SSP1-2.6 scenario. The vegetative and reproductive stages of summer maize were both shortened, and maize yield was increased. However, compared with SSP2-4.5, maize yield would be reduced under the SSP5-8.5 scenario. Compared with SSP2-4.5, the total growth period of winter wheat-summer maize rotation system was shortened, the annual yield was increased, and the proportion of winter wheat yield was increased under SSP5-8.5 scenario. In the future, winter wheat yield was mainly positively correlated with solar radiation, daily mean temperature, and cumulative precipitation during the whole growing season. However, the increase of daily mean temperature and cumulative precipitation was unfavorable to yield increase in 2080s under the SSP5-8.5 scenario. Summer maize experienced the similar changes as winter wheat under future climate change, but daily mean temperature had a negative effect on maize yield. Based on the random forest model, the length of winter wheat reproductive stage and accumulated precipitation in the whole growing season had the greatest impacts on winter wheat yield. At the same time, CO2 concentration, daily average temperature, and accumulated precipitation in the whole growing season had the greatest impacts on summer maize yield. Future climate change would prolong winter wheat reproductive stage and shorten summer maize reproductive stage, but increase winter wheat and summer maize yields in the Huang-Huai-Hai Plain. However, the positive effects of temperature and precipitation on crop yield would become negative over time, resulting in a reduction of summer maize yield in 2080s under the SSP5-8.5 scenario. In general, crop yield mainly would depend on the synergistic effect of climate change and the change of crop growing stage. The results would provide a scientific base and theoretical guidance for the management and the adaption to future climate change of the rotation system of winter wheat and summer maize in the Huang-Huai-Hai Plain of China.

Abstract:The acidified liquid of food waste had a high concentration of organic acids, providing an additional carbon source for the denitrification process in wastewater treatment, potentially alleviating the common problem of carbon source scarcity in urban sewage treatment plants in China. Based on fully mixed anaerobic fermentation reactors, the effects of different pH values, organic loading rates (OLR), hydraulic retention times (HRT) on the anaerobic acidification characteristics of food waste under continuous operation were investigated. By regulating the types of anaerobic acid production, different types of acidification liquids were obtained to evaluate their performance as additional carbon sources for wastewater denitrification. The results indicated that controlling the pH value at 6.0 and the OLR at 15.0g/(L·d) yielded the maximum production of TVFAs at 50.05g/L, with butyric acid fermentation being predominant. When the pH value was controlled at 5.0, acetate-type fermentation occurred, exhibiting the optimal denitrification rate of 7.62mg/(g·h), which was between that of glucose (5.39mg/(g·h)) and sodium acetate (9.47mg/(g·h)). At equivalent COD levels, the denitrification capability of the acidified liquid of food waste (0.21g/g) was 84% of that of sodium acetate. Therefore, food waste acidification liquid represented a sustainable, low-energy and cost-effective renewable carbon source with significant market and application potential.

Abstract:With the aim to clarify the effect of plasma treatment on the storage quality of Tibetan sheep meat, the hind legs meat of Tibetan sheep was selected as the material. After plasma treatment for different times (0min,2min,3min,4min), the meat samples were stored in a refrigerator at 4℃ for 0d,1d,3d,5d,7d. The total number of colonies, pH value, color, texture, cooking loss, TBARS value and carbonyl content were measured at different storage time points.The results showed that on the 7d of storage, the total number of colonies in the Tibetan sheep group with plasma treatment time of 2min,3min and 4min was 18.56%, 23.08% and 27.09% lower than that of the control group.The pH values were 1.53%, 2.21% and 1.02% lower than those of the control group.The a* values were 4.44%,11.71% and 21.62% lower than those of the control group. The hardness values were 5.79%,26.18% and 26.43% lower than those of the control group.The cooking loss was 1.66 percentage points,5.26 percentage points and 2.71 percentage points lower than that of the control group, respectively (P<0.05). In addition, the TBARS content of Tibetan sheep group with plasma treatment time of 2min and 3min was 1.89% and 13.21% lower than that of the control group, respectively, and the carbonyl content was 11.33% and 13.33% lower than that of the control group, respectively. The TBARS content of Tibetan sheep group with plasma treatment time of 4min was 13.21% higher than that of the control group, and the carbonyl content was 13.33% higher than that of the control group (P<0.05). It can be seen that plasma treatment can effectively reduce the total number of colonies in Tibetan sheep and improve tenderness and water holding capacity. However, long-term treatment had a negative impact on meat color.On the whole, when the plasma treatment time was 3min, the total number of colonies of Tibetan sheep was decreased significantly, the tenderness and water holding capacity were increased, and the oxidation of lipid and protein was delayed. The research result can provide a theoretical basis for the improvement of the quality of mutton during storage by atmospheric plasma technology.

Abstract:In response to the challenges posed by the sinking of agricultural walking wheels on wet and soft terrain, as well as the absence of a comprehensive interaction mechanics theory, the typical foot-terrain interaction models for pressure and shear mechanics were modified. Numerous foot-terrain interaction tests were conducted by using a universal testing machine to study the pressure and shear resistance-displacement of various foot designs on different types of wet and soft ground, including sand and soil with varying humidity levels. At the same time, the particle velocity field and motion trend of different types of foots on different wet and soft ground were investigated by means of EDEM discrete element simulation, which was used to observe the fine-scale behavior of wet and soft terrain. In the pressure test, the cylindrical foot was easier to sink compared with the rectangular foot. Combined with the simulation, the cylindrical foot was more likely to damage the soil structure under the foot, and the anti-subsidence performance of the rectangular foot was better than that of the cylindrical foot. The difference in intrusion resistance between rigid and rigid-flexible foots was not significant. The bearing capacity of sand was gradually increased with the increase of humidity, while the bearing capacity of soil was gradually decreased. In the shear test, the foot shear resistance of rigid cylindrical and rectangular foots under sand and soil with different humidity was related to the normal load, and both of them were increased with the increase of normal load. The foot shear resistance had little relationship with the medium humidity. Combined with the simulation, there was more soil accumulation in front of the foot in the process of foot shear, and the foot continuously pushed the soil above the slant, which needed the influence brought by the pushing effect to be further considered. Based on the pressure test data, the typical pressure-bearing model was modified, the subsidence-depth relationship of sand and soil with different humidity was supplemented. Based on the shear test data, the typical shear model was modified, and the shear resistance-displacement relationship was obtained by considering the influence of pushing soil. It can provide a design reference and theoretical basis for the development of a walking wheel foot on wet and soft terrain.

Abstract:Aiming at the problems of poor optimization ability, easy deadlock, and low search efficiency of the traditional ant colony optimization (ACO) when applied to mobile robot path planning, a multilevel field of view adaptive ant colony optimization (MLFVAACO) algorithm was proposed. Firstly, on the basis of ACO, the two levels field of view was expanded sequentially to make the planned path smooth. Secondly, an adaptive global initial pheromone update strategy was designed, which not only avoided the blind search phenomenon of ants in the early stage of the algorithm but also strengthened the guiding role of ants in selecting optional areas. Then the deadlock ants in the algorithm iteration process were optimized to improve the utilization of the ant colony and increase the diversity of search solutions. Finally, the state transition rule of ants was improved to prevent ants from falling into the local optimal solution. The optimal parameters of the MLFVAACO algorithm were selected through simulation analysis, and the feasibility and effectiveness of the MLFVAACO algorithm were verified by comparing it with the traditional ACO algorithm, the improved ACO algorithms, and the graph search algorithms, respectively, in two kinds of grid maps with different levels of complexity. The simulation results showed that in simple and complex environments, compared with the traditional ACO algorithm, the optimal path of the MLFVAACO algorithm was shortened by 12.74% and 4.38%, respectively, the turning points of the path were reduced by 50% and 63.16%, respectively, the ant utilization rate was increased by 99.99% and 99.95%, respectively, and the search efficiency was increased by 60.14% and 62.17%, respectively. Compared with the improved ACO algorithms and the graph search algorithms, MLFVAACO algorithm can plan the shortest path with better path smoothness, while the quality of the search solutions was also better. This fully validated the excellent performance of MLFVAACO algorithm when applied to mobile robot path planning.

Abstract:Humanoid robots, with their human-like form, can more easily integrate into human daily life and adapt to existing infrastructure environments. The study of their kinematics and dynamics theories, along with methods for disturbance rejection control, has been a focal point of research among numerous scientists and engineers around the world for nearly half a century. Due to disturbances from external uncertainties, the motion state of humanoid robots may undergo significant changes in a short period, often leading to difficulties in maintaining continuous walking and resulting in falls. Firstly, optimization adjustments were made to the relationship equation between the walking parameters and the target foot placement in classical gait planning methods based on the linear inverted pendulum, aiming to achieve a more coordinated walking gait. Secondly, a gait planning method based on optimizing the deviation of foot placement within one and two steps was proposed by generating walking patterns in two-step cycles and anticipating the subsequent two target foot placements. Substantial acceleration/deceleration walking simulations and experiments were conducted on a small humanoid robot. The experimental results showed that the improved gait planning method can significantly reduce the maximum deviation of the landing points, reducing the deviation of two consecutive steps during motion state transitions from 1.1cm and 0.8cm to 0.6cm and 0.7cm, respectively, compared with the classical method. Moreover, the improved gait planning method also mitigated the impact of inertial forces on trunk stability, decreasing the maximum change in trunk pitch angle caused by the classical method from 7.8 to 6.0°.

Abstract:A configuration of 3-(PRRPR)RC parallel platform was proposed to meet the dual requirements of orientation adjustment and vibration isolation. The mechanism consisted of three limbs, in which the large-stroke actuator and the passive vibration isolation unit existed in the form of closed-loop sub-chain. In the case of passive vibration isolation, it was assumed that the actuator was instantaneously locked, and the equivalent motion screw was used to describe the closed-loop sub-chain and the orientation adjustment and vibration isolation platform was instantaneously equivalent to a 3-R(RC parallel mechanism. Based on the screw algebra and QR decomposition, the first and second order influence coefficients of the mobile platform and the limb links in the passive vibration isolation mode of the orientation adjustment and vibration isolation platform were derived. Based on the dynamic model, the vibration differential equation of the platform was derived when the foundation was excited by simple harmonic motion. The vibration response was solved by the superposition method of vibration mode, and the corresponding numerical examples were given. The prototype of the orientation adjustment and vibration isolation platform was designed, the experimental system was built, and the passive vibration isolation experiments in x, y and z directions were carried out. Numerical examples and experimental results showed that the vibration transmissibility of the prototype in three directions was less than 55%, that was, more than 45% of the vibration was isolated, which verified the effectiveness of the passive vibration isolation of the designed parallel platform.

Abstract:To meet the reliability requirements of thermal actuators, a method for reliability-based topology optimization of thermally actuated compliant mechanisms based on interval non-probabilistic model was proposed. The interval non-probabilistic model was adopted to describe the uncertainties of thermal load. The functional function was constructed by the output displacement of thermally driven compliant mechanisms. The objective function was used to minimize the volume of the thermally actuated compliant mechanisms, and the reliability index was used as the constraint. The model for reliability-based topology optimization of thermally actuated compliant mechanisms based on interval non-probabilistic model was established. The method of moving asymptotes was applied to update the design variables. Compared with the results of deterministic topology optimization, the volume of thermally actuated compliant mechanisms obtained by reliability-based topology optimization was increased, and the reliability index constraints can be effectively met. The theoretical results of thermally actuated compliant mechanisms obtained by reliability design had an error of less than 5% relative to the finite element analysis results. The effectiveness of the proposed design method for thermally actuated compliant mechanisms was demonstrated. The influence of different output stiffness and thermal load intervals on the results of reliability-based topology optimization of thermally actuated compliant mechanisms was analyzed.

Abstract:The bionic undulating propulsion method has the advantages of good manoeuvrability, strong adaptability, and environmental compatibility, which shows broad prospects in the application of underwater robots. To enhance the propulsion speed, reduce the weight and volume, and improve the reliability of the bionic undulating fin, a modular bionic undulating propeller based on cam mechanism was proposed. Active disturbance rejection control (ADRC) method was used to achieve the continuous accurate control of wave frequency. The structure of the undulating fin was designed. The mathematical model of the sinusoidal wave generating mechanism was established. The instantaneous efficiency formula of the cam mechanism was derived, resulting in an average efficiency of 83.6%. The motion simulation was carried out and the result proved the feasibility and the correctness of the cam parameters calculation. Taking the output frequency as the control target, the control model of the undulating fin was established. The theoretical derivation and computational fluid dynamics simulation analysis showed that both the internal characteristics and the external loads of the undulating fin suffered from nonlinear time-varying effects at constant frequency. To overcome the negative effect of internal and external disturbances, the linear active disturbance rejection controller was designed for frequency control. Based on STM32 single chip microcomputer, the experimental measurement and control system of the undulating fin was realized. The experimental results showed that the undulating fin could accurately track the expected frequency at both low and high frequencies. The response curve was smooth and continuous without overshoot, and the steady-state fluctuation error was less than 2.3%. When the frequency was 3Hz, the output fluctuation errors of the active disturbance rejection controller and the proportion integration differentiation controller were 6.3% and 2.1%, respectively, and the control accuracy was improved by 66.7%. The research results showed that the modular biomimetic undulating fin propeller had satisfactory control precision and good integration. As it can be configured to biomimetic underwater vehicles in any numbers, the modular undulating fin had good application value.