New Concepts for the Estimation of Takagi-Sugeno Model Based on Extended Kalman Filter

摘要: This chapter describes new approaches to improve the local and global approximation (matching) modeling capability of Takagi-Sugeno (TS) fuzzy model. The main aim is obtaining high function accuracy fast convergence. problem encountered that TS identification method cannot be applied when membership functions are overlapped by pairs. restricts application because this type has been widely used during last two decades in stability, controller design systems popular industrial control applications. approach developed here can considered as a generalized version with optimized performance approximating nonlinear functions. We propose noniterative through weighting parameters an iterative algorithm applying extended Kalman filter, based on same idea parameters’ weighting. show filter effective tool An illustrative example inverted pendulum chosen evaluate robustness remarkable proposed locally globally comparison original Simulation results indicate potential, simplicity, generality algorithm. In we prove these algorithms converge very fast, thereby making them practical use.