Research on Path Tracking Method of Unmanned Surface Vehicle Based on Particle Swarm Optimization PID Control and Surrogate Model Accelerated Optimization

With the development of intelligent ships and autonomous navigation technology, real-time performance and computational efficiency of path tracking control have become key challenges in engineering applications. Traditional PID parameter optimization relies on high-fidelity simulation models, whose computational cost increases linearly with the number of iterations, making it difficult to meet real-time requirements. This paper proposes an acceleration optimization method based on Particle Swarm Optimization (PSO) and Deep Neural Network (DNN) agent models: by replacing high-fidelity simulations with DNN, rapid tuning of PID parameters is achieved. Experiments show that the agent model reduces the time for a single parameter evaluation from 0.41 seconds to 0.21 seconds, reducing total optimization time by 47%, with control accuracy loss ranging from 0.3% to 1.8%. In disturbance tests, the maximum heading error is 0.34 rad, with a recovery time of 8 seconds, demonstrating the robustness of the method. This research provides an efficient solution for real-time optimization of ship path tracking and explores its potential for expansion under complex sea conditions.