Hits:

Indexed by:会议论文

Date of Publication:2021-04-26

Page Number:250-257

Key Words:nonlinear system; fuzzy neural network; hybrid signal; non-uniformly sampled data

Abstract:Aiming at the identification issues of non-linear systems with non-uniformly sampled data in the actual industrial processes, a separated identification method based on hybrid signal source is proposed. First of all, a hybrid signal source is used for separating the static nonlinear part from the dynamic linear part in Hammerstein system. Second, the recursive least square algorithm based on the auxiliary model is used for identifying the parameters of the dynamic linear model. Third, the fuzzy neural network model is used for approaching the static non-linear model, and the consequent parameters of the fuzzy neural network are confirmed by the recursive algorithm. Finally, the effectiveness of the proposed method is verified by a non-uniformly sampled nonlinear system.