Latest estimation based recursive stochastic gradient identification algorithms for ARX models

A modified recursive stochastic gradient identification algorithm is presented for ARX models. In the presented algorithm, the hierarchical identification principle is first used, and then the unknown true parameters are replaced by their latest estimation. The convergence analysis of the proposed algorithm is given. In addition, a simulation example is employed to show the advantage of the proposed identification algorithms in convergence rates and estimation accuracy compared with some existing algorithms.