On-Line Identification of Takagi-Sugeno Model Based on Improved Density-Based Clustering Algorithm

摘要

For the complexity and time-consuming of the density-based clustering algorithm (DBSCAN), an improved DBSCAN algorithm is proposed to simplify the clustering, and on this basis present a new algorithm for on-line identification of Takagi-Sugeno (TS) fuzzy model. The new on-line clustering based on improved DBSCAN algorithm is adopted in structure identification, rules can be added, modified and deleted dynamically to reflect the influence of the new arriving data timely, and then update consequent parameters according to the number of rules, finally the recursive least square method is applied to identify the consequent parameters. The approach has been tested on data from Box-Jenkins gas furnace and a second-order nonlinear uncertain system, and the results show the viability and effectiveness of the approach.