An easily implemented approach to fuzzy system identification

An easily implemented approach to fuzzy system identification is proposed. The proposed approach consists of two phases. The first phase involves a baseline design to effectively identify a prototype fuzzy system for a target system from a collection of input-output data pairs. This is implemented by incorporating the subtractive clustering method to determine the number of clusters and the fuzzy c-means (FCM) clustering algorithm to build the actual clusters. If the resulting prototype fuzzy system cannot describe the target system appropriately with respect to the input-output data pairs, the second-phase fine-tuning process to adjust the parameters identified in the baseline design is necessary. This process can be realized by using the steepest descent and recursive least-squares estimation methods. The proposed approach is validated by applying it to a function approximation type of problem: the truck backer-upper problem with an obstacle in the center of the field. For further confirmation, the learning behaviour of the proposed approach for this problem is studied.