Dynamic Particle Swarm Optimization for PID Controller Design of AVR Systems

Abstract

Particle swarm optimization (PSO) is a robust technique in optimization for challenging tasks. Several state-of-the-art PSO variants have emerged in the last two decades. Despite their potential performance, the premature convergence issue is still alive due to insufficient population diversity in the later phase. Consequently, performance is degraded and misses the opportunity of a promising solution. Therefore, to rectify premature convergence issues and build overall optimization performance, Dynamic PSO (DPSO) is introduced in this work for optimal tuning of proportional-integral-derivative (PID) controller parameters in an automatic voltage regulator (AVR) system. Introduced DPSO is based on dynamic parameter adjustment in the velocity update equation, as well as the trajectory modulation factor (TMF) in the position update equation. DPSO consist of (i) constant inertia weight, to reduce redundant momentum, (ii) acceleration coefficients (time-varying), to endorse better transition from exploration to exploitation, and (iii) TMF, to dynamically shrink the particle trajectory and attain proper diversity. The optimal values of the PID controller parameters are evaluated using DPSO according to the objective function ITAE. Compared results illustrate that DPSO delivered better control outcomes for the PID controller parameter than competing approaches in measures of settling time, overshoot, and statistical performance.