Incorporating Problem Specific Information in Genetic Algorithms

This paper describes an approach to incorporating domain knowledge into genetic algorithm search. We use genetic algorithms to attack system connguration design problems; speciically, the structural design and optimization of trusses. Since there exists a large amount of domain knowledge on this problem, we describe the incorporation of this knowledge for guiding genetic search. We outline the problem, the heuris-tics used, and the encoding of the problem in light of available domain knowledge. Preliminary results point toward the eeectiveness of combining genetic algorithms with knowledge-based systems.