Fuzzy Modeling Based on Noise Cluster and Possibilistic Clustering
Isei Ohyama, Yukinori Suzuki, Sato Saga, Junji Maeda · 2006
We propose new fuzzy modeling methods using noise cluster and possibilistic clustering. These modeling methods are based on a switching regression model and a T-S fuzzy model. Since one of the major problems in using a fuzzy clustering algorithm is noise in given data, we employed the noise cluster proposed by Dave to construct a fuzzy model to identify processes of nonlinear plants. Another problem is derived by probabilistic constraint of the FCM algorithm. To solve these problems, we propose a fuzzy model using possibilistic clustering. Fuzzy models using these clustering methods arc proposed in the present paper. Furthermore, computational experiments were carried out to show the effectiveness of the proposed models