Joint intent detection and slot filling plays an important part in natural language understanding for knowledge answering, but it is still in its infancy in the field of agricultural diseases and pests. In addition to lack of corpus, it also faces several challenges such as task independence, mutual correlations ignoring, and intent embedded information neglection. To address the above questions, a novel joint intent detection and slot filling model based on intent embedding and slot-gated mechanism, named AgIG-IDSF was proposed. Firstly, the attention mechanism, which could extract the context-aware features, was introduced into shared encoder to further enrich the contextual semantic features. Secondly, an intent-slot interaction method with intent embedding and the slot-gated mechanism was designed to enhance the ability of intent detection to guide the slot filling task. Finally, the comparative experiments from various aspects were conducted on a self-constructed corpus named AGIS, which mainly contained 22 intent categories, 10 slot categories, and 11.976 annotated samples. The experimental results showed that AgIG-IDSF achieved the intent detection accuracy of 94.41%, slot filling F1-score of 94.01%, and overall semantic accuracy of 88.07% on the self-constructed corpus, which were significantly better than a variety of benchmark models, including bidirectional mutual models. It demonstrated the effectiveness of AgIG-IDSF in jointly identifying the intent and slots in the field of agricultural diseases and pests. In addition, the experimental results on public datasets, i.e., ATIS and SNIPS also showed that the model had a certain generalization ability.