One-step ahead predictive fuzzy controller

A new type of predictive fuzzy relational controller is proposed. The system dynamics is described by a fuzzy relational model, developed on the basis of the input-output description of the process behavior in an open-loop identification experiment. The desired process behavior is defined by a set of fuzzy criteria induced either by a reference model or by general technical requirements. The control algorithm operates by trying, at each sample, a set of predefined control actions on the model. Possible outcomes are determined and the possibility of obtaining the desired process behavior is calculated The proposed algorithm then combines the obtained possibilities and maps them to the control signal, with the aim of reinforcing the possibility of the desired process behavior. As an example, the controller is designed for a specific desired process behavior. Practical aspects of the controller design are discussed along with some digital simulation results. (C) 2001 Elsevier Science B.V. All rights reserved.