Scheduling of Manufacturing Systems Based on Extreme Value Theory and Genetic Algorithms

Two scheduling methods based on Extreme Value Theory (SEVAT) and Genetic Algorithms (GA) are developed. The SEVAT approach is a schedule building approach that creates a statistical profile of schedules through random sampling and predicts the potential' of a schedule alternative. The GA approach, on the other hand, is a schedule permutation approach, in which a population of schedules are initially generated, and through some operations, good traits of these schedules are combined to produce better schedules. These two scheduling methods were applied to two static benchmark job shop problems (the Muth and Thompson 6x6 and lOx 10 problems). The results compare favorably with the optimal solutions, and the solutions obtained using some common dispatch rules and a scheduling approach based on Fuzzy Logic, which is representative of current scheduling research. A dynamic scheduling problem was designed to reflect a real job shop scheduling environment closely. Two performance measures, viz, Mean Job Tardiness and Mean Job Cost, were used to demonstrate multiple criteria scheduling. Three factors were identified, and varied between two levels each, thereby spanning a varied job shop environment. A factorial design of experiments comprising of 8 experiments were then designed. The SEVAT and GA approaches were applied to these 8 experiments and the results compared with several common dispatching rules and the Fuzzy Logic Approach. The results of this extensive simulation study, overwhelmingly indicate that the SEVAT and GA scheduling approaches produce better scheduling performance than the other methods. Thesis Supervisor: Professor George Chryssolouris Title: Associate Professor of Mechanical Engineering