The two-stage Recursive Identification Algorithm for theModel of Nonlinear Output Error

The output error model of nonlinear system is a kind of common models in actual industrial production. A class of two-stage recursive extended least squares algorithm based on auxiliary model is proposed for the nonlinear output error model with colored noise in this paper. According to the auxiliary model ideas and decomposition techniques, the identification of nonlinear complex system is decomposed into sub-system models and noise models. Then the sub-models parameters are separately identified through the ideas of the least squares, where the unpredictable noise term exists in the noise information vector are replaced by its estimated values. Finally, the algorithm is compared with the recursive augmented least squares algorithm in respects of the parameter estimation accuracy and convergence rate. The simulation results show that the algorithm has high precision, fast convergence rate and a small amount of computation.