Hierarchy Genetic Algorithm to Solve Multi-Objective Scheduling Problems Involving Various Types of Assignments for Parallel Processing System

Abstract

This paper examines multi-objective scheduling problems involving various types of job assignments. A hierarchy type of genetic algorithm is proposed to search for Pareto solutions for multi-objective functions in an effective manner. The algorithm consists of job assignment control and control of condition for job assignments, and these controls are constructed hierarchically. The structure of controls in the proposed algorithm is adaptable to parallel processing systems to reduce computational time. In addition, various types of procedures are adoptable in order to control conditions for job assignments. In this study, a hybrid type of local search method is introduced as an effective procedure to search for solutions. The local search method is constructed of two local search procedures to optimize different objective functions, and these procedures are alternately executed to obtain optimal Pareto solutions. In this paper, the characteristics of the proposed hierarchy type of genetic algorithm are described. A parallel processing system for the algorithm is then developed and examined on a scheduling problem involving worker assignments and job assignments to evaluate its performance.