Optimization of Fuzzy PID Temperature Control Algorithm Based on Particle Swarm Optimization (PSO)
Abstract
Fuzzy PID controllers are widely adopted in the field of industrial manufacturing. However, when applied to temperature control systems characterized by large inertia, significant time delay, and potentially time-varying delay, conventional fuzzy PID controllers still face considerable challenges in balancing dynamic response and steady-state accuracy. Nevertheless, they suffer from several drawbacks: the design of membership functions and scaling factors heavily relies on expert experience, parameter tuning is time-consuming, and it is difficult to acquire the global optimal solution. To tackle the above limitations, this paper proposes an offline optimization scheme for fuzzy PID temperature control based on Particle Swarm Optimization (PSO). This method adopts a product-type fuzzy PID structure, and performs joint optimization on 62-dimensional parameters including the central values, half-widths of input and output membership functions as well as scaling factors. To enhance the overall control performance of the system, the integral of time multiplied by absolute error (ITAE) and overshoot penalty term are combined to construct the fitness function. Meanwhile, automatic interactive simulation with Simulink is integrated to search for the optimal control parameters and the shapes of membership functions. The results demonstrate that the optimized fuzzy PID controller significantly improves dynamic performance: the ITAE index decreases by approximately 8.4% (from 49350.6 to 45194.5), the settling time is shortened from 27.97 s to 24.34 s (a reduction of about 13.0%), and the overshoot drops markedly from 0.46 °C to 0.18 °C (a reduction of about $60.9 \%$). Although the steady-state error increases slightly from 0.0011 °C to 0.0158 °C, it remains within the acceptable tolerance for temperature control and still meets the practical requirements. Accordingly, the proposed optimization method can improve the dynamic response performance while guaranteeing system stability. This approach enables automatic tuning of fuzzy PID parameters, reduces reliance on empirical parameter adjustment, and boosts the efficiency of parameter optimization.