System Identification via Simultaneous Perturbation Stochastic Approximation

抄録

The simultaneous perturbation stochastic approximation(SPSA) is an extension of Kiefer-Wolfowitz stochastic approximation (KWSA) algorithm. In SPSA, since all parameters is perturbed simultaneously, it is possible to modify parameter with only two measurements of the evaluation function regardless of the dimension of parameter. We propose a parameter identification algorithm using SPSA. Simulation result shows a feasibility of the identification proposed here.