The influence of sampling points on the descriptive performance of T-S fuzzy modeling
Yanliu Zhang · Caai Transactions on Intelligent Systems · 2008
The number of sampling points in fuzzy modeling has a substantial influence on the accuracy of models.If the sampled data is limited and its distribution not properly controlled,choice of the optimal number of sampling points creates significant problems in fuzzy identification.The author proposed a fuzzy identification algorithm with varied sampling points to investigate the influence of the number of sampling points on descriptive performance.Based on the T-S fuzzy model,we extracted the fuzzy rules by using the symmetrical triangular fuzzy division and the net-diagonal method.By modeling the DISO system and the Mackey-Glass chaotic time-series,we concluded that training and testing performance indexes in fuzzy models will increase with increased numbers of sampling points.