GENETIC ALGORITHM-BASED APPROACH TO BI-LEVEL LINEAR-PROGRAMMING

This paper reports on the use of a genetic algorithm based technique, GABBA, to solve bi-level linear programming (BLLP) problems. GABBA is used to generate the leader's decision vector, and the follower's reac tion is obtained from the solution of a linear program. GABBA is diffe rent from the usual genetic algorithms because we only use mutations, alleles of base-10 numbers, and a survival strategy that is suited to BLLP. Results show that, while it takes more cpu time, GABBA gets clos er to the global optimum than Bard's [1983] grid search technique for problems of most sizes.