Rule-based control of Wiener-type nonlinear processes through self-organizing

Abstract

This paper considers the application of rule-based fuzzy control techniques to the problem of controlling Wiener type nonlinear systems in general and pH neutralization processes in particular. Instead of relying on human experts or a process model to design the controller, we explore the possibility of constructing a controller automatically by a paradigm referred to as active learning from the environment. To be more specific, a rule-based feedback controller is built with the required control rules being self-organized by means of a self-growing adaptive quantization algorithm with the aid of a learning mechanism which provides necessary control information. To overcome the learning difficulty arising from the presence of strong nonlinearity, we propose two schemes to improve the learning process. The simulation results on the control ofa pH process are given.