A Belief Rule Inference Approach for Controlling the System with Uncertain Parameters

This paper presents a belief rule inference method for controlling the system with uncertain parameters. It uses the belief rule base (BRB) to model nonlinear relationship between system deviation and its integral, differential (inputs) and the controlled variable (output), in which the antecedent and consequent attributes of rule are the referential values of the inputs and output respectively. For the rules activated by the inputs, the evidential reasoning (ER) algorithm is used to combine the belief structures of these rules so as to reason out the value of the controlled variable. Moreover, a nonlinear optimization strategy is given to enhance the accuracy of initial parameters of BRB. A representative experiment on the separately excited DC motor with uncertain parameters illustrates that the proposed method can give more accurate control actions than the classical fuzzy-PID control method.