Local Search Methods to Solve Multiple Objective Function

Abstract

           In this paper we considered the problem of scheduling n jobs on a single machine. Our aim in this study is to find the near optimal solution to minimize the cost of total flow time and maximum earliness with unequal ready times.

            Different local search methods: (Descent Method, Adjacent Pairwise Interchange Method, Simulated Annealing, Genetic Algorithm) are developed, compared, and tested for the problem. We investigate the influence of the parameters variance for these local search methods, and empirically analyze their starting solutions. Computational experience found that these local search algorithms can solve the problem up to (23000) jobs with reasonable time. Also we found that: the Genetic algorithm is the best local search heuristic algorithm for our problem when the size is less than or equal to (1500) jobs, and for problems of large size the Simulated Annealing was recommended.