Metaheuristic-based PID Control: Evaluating the Effect of Parameter Settings

Abstract

Proportional-Integral-Derivative (PID) control method has been utilized in many industrial control applications, which has its limitations for the optimized control. Manually calculated controller tunings by using fixed formulas are only adequately to give basic control but not well-competent to non-linear process with various operating levels. Optimization analysis via metaheuristic approach has substantially obtained the best solutions and has proven its credibility to provide the better controller values. Nevertheless, the optimization analysis might provides bad result if the determinant parameter settings are not properly adjusted during the optimization analysis. This paper objectively presents a direct way to determine the best determinant parameter settings for the optimization analysis, whereby the Genetic Algorithm (GA) and Particle Swarm Optimization (PSO) are chosen for the research analysis. In detail, parameter settings that include upper and lower bounds (UB, LB), maximum iteration (MaxIt), population size (nPop), mutation rate (mu) and damping ratio (wdamp) are analyzed. Ultimately, analysis results are compared with manually calculated controller tunings by using Level Process Control Training Module, SE-207. Both performance indexes and response curves showed that PSO and GA were better performed than the conventional tunings. While comparing optimization analyses, PSO reacted more aggressively than GA corresponding to slight overshoots but both optimizations produced very close error values. In conclusion, metaheuristic approach with proper parameter settings produces better control tunings as for the controlled loop demands aggressive control action would prefer PSO, and in contrast, a stabilized control prefers to apply GA for the nonlinear process.