Generating fuzzy control rules by a clustering algorithm based on a grey relational measure

A method is proposed for generating appropriate fuzzy control rules directly from some gathered input-output data of the system considered. First, a clustering algorithm using a grey relational measure as the similarity measure is proposed. The grey relational measure is adopted such that the scale problem can be effectively processed when the considered data have different features and ranges. The proposed clustering algorithm can effectively cluster and correctly classify the data with different sizes and shapes of clusters. Then, a method based on the clustering algorithm is proposed for generating appropriate fuzzy control rules from the gathered input-output data. Finally, an inverted pendulum control problem is used to illustrate the efficiency of the proposed method.