A new approach for online fuzzy identification by potential clustering including rule reduction

This paper uses potential clustering approach to perform online fuzzy clustering. This method is an improvement of the subtractive clustering which is a noniterative clustering algorithm and so is suitable for online applications. In Spite of all capabilities of the potential clustering, this method suffers from a major disadvantage. The number of clusters grows fast when the sensitivity of the algorithm is increased. In this article an innovative technique has been proposed to reduce the number of clusters. The proposed method is applied to the Macky-Glass benchmark. It is shown although the number of clusters is reduced; the resulting performance will not be affected.