An Intelligent PID Control using Neural Networks for Pneumatic Servo Systems

In this paper a novel control scheme in which the variable gain type PID controller is constructed combined with a Neural Network (NN) optimized by a Particle Swarm Optimization (PSO) algorithm. To improve the control performance, overcoming the disadvantages caused by low rigidity, response delay, fluctuations in pneumatic servo systems (PSSs), the proposed method is designed to minimize error by the intelligent scheme based on PID control using the optimized gains tuned by the PSO type NN. Different from the conventional fixed gain type PID control, which requires gain tuning according to some rules, or trials and errros, or experience, the proposed is with the advantages from PSO type NN to search proper gain values automatically. According to the experimental results in this paper, the optimal or “near optimal” gains were confirmed. The effectiveness of the proposed scheme is confirmed by experiments using the existent pneumatic servo system.