Intelligent Optimization of PID Controller Parameters Using Enhanced Parallel Genetic Algorithm

Abstract

Traditional Proportional-Integral-Derivative (PID) parameter tuning often employs the Ziegler-Nichols method, but it suffers from limitations such as reliance on experience and trial-and-error. With the development of intelligent algorithms, although the Simple Genetic Algorithm (SGA) can achieve automatic parameter adjustment, it still faces issues like insufficient global search capability and premature convergence. This paper proposes an Enhanced Parallel Genetic Algorithm (EPGA), which constructs a parallel evolutionary architecture and a multi-dimensional performance index system. By integrating dynamic tournament selection, adaptive mutation, and periodic elite migration mechanisms, and designing a diversity reward function, EPGA effectively balances exploration and convergence capabilities. Simulation experiments show that for a typical first-order inertial system with time delay, the PID controller optimized by EPGA can achieve no overshoot dynamic response, fast stability, and precise steady-state control performance. Compared with traditional genetic algorithms and empirical tuning methods, this optimization strategy significantly improves control quality indicators. The study indicates that EPGA suppresses premature convergence by constructing a parallel population topology network and accelerates the propagation efficiency of excellent parameters by combining dynamic gene selection mechanisms, demonstrating excellent global search capability and convergence characteristics in control system parameter optimization. This method provides a new solution for high-precision control requirements in complex industrial scenarios, balancing dynamic response quality and system robustness.