Hierarchical gradient based parameter identification for non-uniformly sampling wiener systems

摘要

In view of the stochastic gradient identification method and the hierarchical identification principle, a hierarchical multi-innovation stochastic gradient (HMISG) identification algorithm for non-uniformly sampling Wiener nonlinear (NUSW) systems is presented in this paper. Furthermore, a piecewise forgetting factor is introduced to improve the convergent rate and disturbance rejection. The proposed algorithm can estimate system parameters directly by decomposing the system into two subsystems, and has less computation. The simulation results shows that the NUSW systems can be identified effectively using developed algorithm.