One-step ahead predictive fuzzy controller (Q5947546)

A predictive fuzzy relational controller for nonminimum phase systems is proposed. The applied probabilistic model of the system is based on the fuzzy discretization of the input-output domains. The conditional possibility measure of the output with respect to the referential fuzzy sets is estimated in an open-loop experiment. The proposed controller tries at each sample instant a set of predefined control actions on the relational array model. The set of outcomes is obtained, and the possibility measure that the control signal will lead to the desired process behavior is evaluated. These possibilities are then combined and mapped into the controller output. The desired process behavior is defined by a given set of fuzzy criteria. Digital simulation results illustrate the advantages and drawbacks of the proposed new type of predictive fuzzy controllers.