An Artificial Immune System Algorithm with Social Learning and Its Application in Industrial PID Controller Design

Abstract

A novel artificial immune system algorithm with social learning mechanisms (AIS-SL) is proposed in this paper. In AIS-SL, candidate antibodies are marked with an elitist swarm (ES) or a common swarm (CS). Correspondingly, these antibodies are named ES antibodies or CS antibodies. In the mutation operator, ES antibodies experience self-learning, while CS antibodies execute two different social learning mechanisms, that is, stochastic social learning (SSL) and heuristic social learning (HSL), to accelerate the convergence process. Moreover, a dynamic searching radius update strategy is designed to improve the solution accuracy. In the numerical simulations, five benchmark functions and a practical industrial application of proportional-integral-differential (PID) controller tuning is selected to evaluate the performance of the proposed AIS-SL. The simulation results indicate that AIS-SL has better solution accuracy and convergence speed than the canonical opt-aiNet, IA-AIS, and AAIS-2S.