Particle swarm optimization for worst case tolerance design

Abstract

Worst case tolerance analysis is a major subtask in modern industrial electronics. Recently, the demands on industrial products like production costs or probability of failure have become more and more important in order to be competitive in business. The main key to improve the quality of electronic products is the challenge to reduce the effects of parameter variations, which can be done by robust parameter design. This paper addresses the applicability of particle swarm optimization combined with pattern search for worst case circuit design. The main advantages of this approach are the efficiency and robustness of the particle swarm optimization strategy. The method is also well suited for higher order problems, i.e. for problems with a high number of design parameters, because of the linear complexity of the pattern search algorithm.