A 2 DOF predictive control based on evolving fuzzy model

In this paper we present the design of a two degree-of-freedom (2 DOF) control algorithm, developed for a class of single-input multiple-output non-linear systems. The feedforward and feedback control loops are developed based on the known Takagi-Sugeno fuzzy model of the system. The fuzzy model is part of the control law and it is obtained by evolving fuzzy modelling. Although only a single system output is controlled, the clustering in evolving fuzzy model is made on multiple measurable system outputs. The control law is designed in a way that enables clustering in the evolving fuzzy model on multiple outputs. However, every input-output connection is described by a non-linear autoregressive system with exogenous inputs. The purpose of the feed-forward loop is to bring the controlled output close to the desired reference signal, and the feedback loop is used to eliminate the reference tracking error that may occur due to imprecise system modelling, noise or any other disturbances. Linearisation around the reference signal enables the design of an error-model predictive control in the feedback loop. The presented control approach was experimentally validated in simulation environment on a system of three water tanks. The experimental results confirm the applicability of the approach.