In order to improve the efficiency and reliability of edge-tracking and feeding operation of shrimp ponds, an unmanned differential-speed feeding ship was taken as the research object, and a control algorithm based on variable integral PID attitude control and margin interval search control dynamic switching was proposed. Based on the establishment of the motion model and attitude feedback model of the unmanned differential feed-dropping ship, the algorithm tracks the distance information and angle information of the unmanned ship's side patrol process to maintain the correct attitude of the feed-dropping ship. At the same time, considering the nonlinear characteristics of unmanned ship control, a margin interval search control algorithm is designed to make the hull quickly converge within the PID adjustment range and reduce the complexity of the algorithm. Through Matlab simulation and analysis of the effectiveness of the algorithm under different parameters and boundaries, it is verified that the maximum edge-tracking error can be controlled within 0.3m and the maximum angle tracking error can be controlled at 9° under the edge-tracking speed of 0-3m/s. The research shows that the proposed edge-tracking algorithm meets the patrol index requirements of shrimp pond feeding equipment and has good practical value.