Identification of nonlinear stochastic systems described by a reduced complexity Volterra model using an ARGLS algorithm

This paper proposes a stochastic identification algorithm of a model describing non linear stochastic system. The identified model known as SVD-PARAFAC-Volterra model [1] results from tensor decomposition of kernels of classical Volterra model. The proposed algorithm uses the Recursive Generalized Least Square (RGLS) method in alternative way to estimate the parameters of the model. The algorithm validation is ensured by simulation results.