A new memetic global and local search algorithm for solving hybrid flow shop with multiprocessor task scheduling problem

Abstract

Hybrid flow shop (HFS) scheduling problem is combining of the flow shop and parallel machine scheduling problem. Hybrid flow shop with multiprocessor task (HFSMT) scheduling problem is a special structure of the hybrid flow shop scheduling problem. The HFSMT scheduling is a well-known NP-hard problem. In this study, a new memetic algorithm which combined the global and local search methods is proposed to solve the HFSMT scheduling problems. The developed new memetic global and local search (MGLS) algorithm consists of four operators. These are natural selection, crossover, mutation and local search methods. A preliminary test is performed to set the best values of these developed new MGLS algorithm operators for solving HFSMT scheduling problem. The best values of the MGLS algorithm operators are determined by a full factorial experimental design. The proposed new MGLS algorithm is applied the 240 HFSMT scheduling instances from the literature. The performance of the generated new MGLS algorithm is compared with the genetic algorithm (GA), parallel greedy algorithm (PGA) and efficient genetic algorithm (EGA) which are used in the previous studies to solve the same set of HFSMT scheduling benchmark instances from the literature. The results show that the proposed new MGLS algorithm provides better makespan than the other algorithms for HFSMT scheduling instances. The developed new MGLS algorithm could be applicable to practical production environment.