A Comparison of Intelligent PID Position Controllers With Autotuners for a Pneumatic System

Abstract

This paper reports on a study whose objective is to explore the potential of intelligent algorithms such as fuzzy rules and neural networks (NN) as applied to the control of pneumatic servosystems. Current application is position control of a pneumatic gantry robot. Preliminary experiments with adaptive fuzzy and NN controllers showed improvement in tracking performance by upwards of 70%. This level of improvement was expected given the adaptive nature of both controllers. However, both methods also required significant effort to setup. An on-line autotuner was developed to improve the ease of implementation. Comparative results are given for five controllers: 1) manually tuned PID, 2) autotuned PID, 3) adaptive NN PID, 4) fuzzy adaptive PID and 5) autotuned fuzzy adaptive PID. Experiments were conducted at 2 different supply pressures and 3 different tracking frequencies. Once again the fuzzy adaptive PID controller improved performance over fixed gain PID, this time by upwards of 80%, even when both were given the benefit of autotuning.