Membership function-based fuzzy model and its applications to multivariable nonlinear model-predictive control

Abstract

The nonlinear processes in direct digital control systems can be modeled by the membership function-based fuzzy models proposed in this paper. The two-dimensional membership functions used by this paper are identified by using limited process response data. Instead of using membership functions to represent the belonging to a set this paper uses the membership functions to represent the gradual deviation from the known states. The membership function-based fuzzy models are effective nonlinear models which can be used for multivariable nonlinear predictive control in which the process interaction is used to enhance the control action rather than being decoupled like in linear control.<>