Weighted CRI Method for Fuzzy Controller

A traditional fuzzy control system the Max-Min CRI (Compositional Rule of Inference) method, which equivalently combines rules for an inference, for fuzzy inference suffers from a critical problem. It produces a big error for a premise which is very similar or identical to one of premises in rules. It is because the importance of rules is not taken into the consideration for inference. In this paper, we propose a novel fuzzy inference method to reduce the error by providing weights to composition of rules according to the similarity of premises. In the experimental section, we show that the proposed similarity measure can reduce a lot of error regions compared with the conventional Max-Min CRI method using the example of D.C. series motor controller.