Simultaneous perturbation extremum seeking method for dynamic optimization problems

A method of dynamic optimization using simultaneous perturbation principle is proposed and illustrated with numerical examples. The method is developed from a simultaneous-perturbation stochastic approximation (SPSA) recursive algorithm and is intended for problems with continuous measurements. The method is suitable for large scale dynamic optimization problems. Problems requiring continuous measurements appear, for example, in the area of neurocontrol, where, for some algorithms, the updates of the network gains depend on the integral of the square tracking error with respect to time.