Adaptive System Design: A Genetic Approach

Abstract

An unconventional approach to adaptive system design is presented which has considerable promise for complex automation problems. The heart of this approach is a sophisticated search technique called a reproductive plan which is based on information processing models from population genetics. Both simulation and analysis results are presented which suggest that reproductive plans out-perform existing techniques on complex process response surfaces. An example is presented to provide an intuitive idea of how reproductive plans operate, and some observations are made concering the applicability of this approach to various problems in adaptive system design.