Acceleration of Levenberg-Marquardt Method Training of NSV Attitude System Fuzzy Modeling

Takagi-Sugeno fuzzy modeling is well known as a practical approach for nonlinear system modeling. In this approach, a selection of membership functions makes much effect upon the model performance. It is usually determined by the experts' knowledge for the objective systems. However, it is often difficult to give appropriate membership functions for complex nonlinear system without any prior information. In this paper, a modification on the Levenberg-Marquardt (L-M) algorithm for Takagi-Sugeno fuzzy training is proposed. It has global and locally quadratic convergence and raises the convergence speed greatly. This method can adjust the parameters of each linear polynomials and fuzzy membership functions on line without relying on experts' knowledge. Then, an application to identification of NSV (near space vehicle) attitude system is considered and the availability of the proposed method is illustrated by the simulation results.