Impact of Performance Evaluation Criteria on Intelligent Tuning

Abstract

This paper addresses the problem of automatically tuning in an intelligent manner so that a balance between efficiency and computational speed is reached. In this paper a new proposed technique which uses Particle Swarm Optimization (PSO) to compute the best optimal value for the PID parameters are presented. In this study two different performance criteria are used simultaneously for the optimization problem, namely Integral of Time-weighted Absolute Error (ITAE) and an output response based performance criteria (Fitness Function). The integration between the two performance criteria produces two distinct tuning techniques called Error-Fitness PSO (EFPSO) and Fitness-Error PSO (FEPSO). This paper also proposes new modified Time Varying Acceleration Coefficients (TVAC) that is used in the PSO algorithm. Finally, simulation experiment on a single degree of freedom robotic arm shows that the proposed techniques can produce optimal PID gains with good computational efficiency and improved step response characteristics. The proposed integration techniques can highly improve the PID tuning optimization in comparison with the one that use only one technique.