Dead-zone in the valve degraded the performances of the Electro-Pneumatic Actuator (EPA) system. It makes the system difficult to control, become unstable and leads to chattering effect nearest desired position. In order to cater this issue, the EPA system transfer function and the dead-zone model is identified by MATLAB SI toolbox and the Particle Swarm Optimization (PSO) algorithm respectively. Then a parametric control is designed based on pole-placement approach and combine with feed-forward inverse dead-zone compensation. To reduce chattering effect, a smooth parameter is added to the controller output. The advantages of using these techniques are the chattering effect and the dead-zone of the EPA system is reduced. Moreover, the feed-forward system improves the transient performance. The results are compared with the pole-placement control (1) without compensator and (2) with conventional dead-zone compensator. Based on the experimental results, the proposed controller reduced the chattering effect due to the controller output of conventional dead-zone compensation, 90% of the pole-placement controller steady-state error and 30% and 40% of the pole-placement controller with conventional dead-zone compensation settling time and rise time.
Direct Adaptive Pole-Placement Controller using Deep Reinforcement Learning: Application to AUV Control
