An Enhanced Objective Cluster Analysis-Based Fuzzy Iterative Learning Approach for T-S Fuzzy Modeling

This work presents an Enhanced Objective Cluster Analysis-based fuzzy iterative learning approach for T-S fuzzy modeling. In this method, the Enhanced Objective Cluster Analysis including the Dipole Partition, the Relative Dissimilarity Measure and the Enhanced Consistency Criterion are incorporated with the Fuzzy c - Means algorithm for the robust and compact fuzzy partition in the input space. For improving accuracy of the model, iterative learning strategy with Covering Measure is adopted to repartition the dissatisfying fuzzy subspaces according to the user´s requirement. By the Stable Kalman Filter algorithm, the consequent parameters are efficiently estimated. The Box-Jenkins example demonstrates the power of our method.