Design of a performance-driven CMAC-PID controller

To operate process plants under restricted condition such as high-quality and reduced cost, Control Performance Assessment (CPA) that assesses performance of the control systems from the operation data has been actively studied in recent years. However, owing to the non-linearity included in the general system plants, it is difficult to tune the control parameters regularly. The scheme to tune PID parameters using Cerebellar Model Articulation Controller (CMAC) that is a kind of artificial neural networks has been proposed to improve transient-state response for the non-linear plant. In this research, a design of PID control system using the CMAC based on CPA is proposed. Consequently, the desirable control performances in both transient-state and steady-state are obtained by using the proposed controller. Through a numerical simulation, usefulness of the proposed scheme is verified under non-linear system.