Genetic Algorithms in Problem Space for Sequencing Problems

In this paper, genetic algorithms will be developed for sequencing type problems of importance in manufacturing systems. The proposed algorithms are based on an auxiliary problem domain called “problem space” [15, 17]. Problem space provides a framework in which problem-specific information can be incorporated explicitly into local search heuristics. The proposed space has been found to be well suited for search by genetic algorithms perhaps because standard crossover can be used. In this paper, properties of problem space will be discussed, then three test cases will be presented which illustrate the usefulness of the method. The test problems we present are 1) the number partitioning problem, 2) the classic job shop scheduling problem (JSP) with minimum makespan as the objective, and 3) the “standard cell placement problem” which arises in the design of VLSI circuits.