On search techniques in switching environments

A study is made of the applicability of two direct search techniques to the problem of optimizing the performance of a system operating in a switching environment. The techniques are modified to extend their applicability in such environments. The methods considered are (1) the stochastic automaton method and (2) the adaptive random optimization method. Pattern recognition techniques are used to identify the state of the environment, and a separate search is performed for each state of the environment. Experimental results are presented to compare the convergence properties of the methods with pattern recognition with that of the stochastic automaton method without pattern recognition for several different switching rates.