A control strategy is proposed for the position control loop of permanent magnet synchronous motors (PMSM) servo systems to address issues related to deteriorating dynamic characteristics and poor disturbance rejection when system inputs vary. This strategy combines particle swarm optimization (PSO) with a Fuzzy PID controller and utilizes deep Q-Network (DQN) compensation techniques. Initially

the PSO algorithm is used to fine-tune critical parameters of the Fuzzy PID controller. This process involves searching for optimal control parameters within the error space to enhance the system's adaptability under complex operating conditions. Subsequently

the DQN deep reinforcement learning algorithm is applied. An exponential function is introduced to design the system's reward function

increasing the differentiation of reward signals. Extensive training on replayed empirical data approximates the target value function

allowing fine-tuning and compensation of the Fuzzy PID controller's output. Finally

simulation and experiments validate the effectiveness of the proposed control strategy. The results demonstrate significant improvements in the system's adaptability

rapid response

and steady-state accuracy

along with enhanced robustness when dealing with load disturbances.