Traditional monitoring systems realize water quality monitoring with a large number of monitoring nodes placed in the aquaculture area and communicating with wireless sensor network. The monitoring nodes combine multiple sensors to collect water quality data. The monitoring data changes on each node in the integrated water area reflects water quality status. But the quantity of nodes, monitoring scale and water coverage are limited. While insufficient number of monitoring nodes could not represent the entire water area, but increasing the density of monitoring nodes would increase the complexity of the system and cost. Therefore, the wider range of water quality data was collected, the more intuitive water quality status distribution in the water area reflected. Expanding the monitoring scope could avoid the abnormality or missed inspection of water quality caused by inadequate coverage. A dynamic monitoring system for aquaculture water quality monitoring was designed based on unmanned surface vehicle. The system expanded the monitoring scale, increased the monitoring range and collected more extensive water quality information by dynamic monitoring. It also expanded the current water quality monitoring and anomaly detection programs. The dynamic monitoring system consisted of unmanned surface vehicle, shore-based console, manual remote controller and cloud monitoring server. The unmanned surface vehicle integrated Raspberry Pi and multiple sensors to collect water temperature, pH value and water turbidity, and the data was sent back to the shore-based console and uploaded to the cloud server. The dynamic monitoring system designed a data acquisition scheme based on hover sampling by unmanned surface vehicle, the returned data composed longitude, latitude, roll angle, pitch angle, yaw angle, ultrasonic distance, water temperature, turbidity value and pH value in sequence. The data returned the shore-based console and performed effective filtering on all received packets. The vehicle ran a path planning strategy based on map analysis. It calculated the position and heading angle in real time to assist in automatic navigation to improve monitoring efficiency, and designed an obstacle avoidance system to detect obstacles in front of the hull. After testing and verifying the feasibility of the system and optimizing the monitoring efficiency of the system, the deployment experiment was carried out in Zhoushan aquafarm. It was verified by experiments that the absolute value of the relative error of water temperature was not more than 0.5% compared with the traditional method, the absolute value of the relative error of the pH value was not more than 1.43%, and the absolute value of the relative error of turbidity was not more than 4.9%, all the data was within the accuracy range of the sensor and met the monitoring needs. The system was deployed in aquaculture waters, collecting 731 sets of valid data within 9800m2, covering approximately 68% of the water surface range, which reflected the overall water quality information of the water area and provided abnormal conditions in the water surface area. The dynamic monitoring system improved the shortcomings of current aquaculture water quality monitoring methods and expanded application of the Internet of Things technology in the field of agricultural engineering. Compared with current monitoring program, the scope of water quality monitoring was expanded, the monitoring efficiency was improved, and the monitoring cost was greatly reduced. It could be regarded as monitoring strategy and technical means for aquaculture water quality monitoring which had better application and promotion value and still had somespace to be improved.