Fuzzy prediction of chaotic time series based on SVD matrix decomposition
Hongwei Wang, Hong Gu, Zhe-Long Wang · 2005
A learning algorithm of fuzzy modeling based on fuzzy competitive learning and singular value decomposition (SVD) is proposed in this paper. First, fuzzy competitive learning is used to confirm the fuzzy space of input variables. In addition, the recursive least square based SVD method is used to confirm the consequent parameters of fuzzy model for the sake of accumulating and transferring of the errors of recursive least square. The structure and parameters of fuzzy model are confirmed by means of the proposed algorithm. To illustrate the performance of the proposed method, simulations on the chaotic Mackey-Glass time series prediction are performed. Combining either off-line or on-line learning with the proposed method, the simulating result shows that the chaotic Mackey-Glass time series are accurately predicted, and demonstrate the effectiveness.