Dynamic flexible job shop scheduling method based on improved gene expression programming

Chunjiang Zhang,Kunlei Lian,Kunkun Peng,Yin Zhou,Xinyu Li,Suyan Zhang

doi:10.1177/0020294020946352

Abstract

Dynamic scheduling is one of the most important key technologies in production and flexible job shop is widespread. Therefore, this paper considers a dynamic flexible job shop scheduling problem considering setup time and random job arrival. To solve this problem, a dynamic scheduling framework based on the improved gene expression programming algorithm is proposed to construct scheduling rules. In this framework, the variable neighborhood search using four efficient neighborhood structures is combined with gene expression programming algorithm. And, an adaptive method adjusting recombination rate and transposition rate in the evolutionary progress is proposed. The test results on 24 groups of instances with different scales show that the improved gene expression programming performs better than the standard gene expression programming, genetic programming, and scheduling rules.

Highlights

Production manufacturing process is vital to a manufacturing company’s survival and growth, where scheduling plays a key role in determining the overall efficiency and productivity
An improved gene expression programming (GEP)-based dynamic scheduling method was proposed for solving the dynamic flexible job shop scheduling problem (DFJSP) considering setup time and random job arrival
The variable neighborhood search is embedded into the GEP, and multiple neighborhood structures are designed for improving the local search ability

Summary

Introduction

Production manufacturing process is vital to a manufacturing company’s survival and growth, where scheduling plays a key role in determining the overall efficiency and productivity. To the best of the authors’ knowledge, Nie et al.[21] is the only attempt in the literature to use GEP for flexible job shop dynamic scheduling problems. The improved GEP for flexible job shop scheduling problem (FJSP) is presented in detail, followed by the description of the variable neighborhood search algorithm that is used to improve the local search capability of GEP, and the transposition operators used for promoting population diversity. Before each operation of a job is processed, some preparation work will be done on the corresponding machine, such as adjustment and clamping This period is usually called the setup time.[28] In the traditional scheduling problem research, the setup time is usually ignored or used as a part of the processing time. These operators are introduced briefly as follows: 1. Selection: the roulette wheel selection with elitism is adopted in the original GEP

Recombination

Mutation

Result analysis

Conclusion and future work