A Memetic Algorithm Approach for the Job-Shop Scheduling Problem with Variable Machine Efficiency and Maintenance Activities

Abstract

Variable machine efficiency (VME) and maintenance activities (MA) are critical factors often unexplored in job scheduling problems. This paper introduces a new problem termed the job-shop scheduling problem with variable machine efficiency and maintenance activities (JSSP-VME-MT), wherein, unlike the traditional JSSP, machine efficiency and maintenance activities are explicitly incorporated into the scheduling process. The study proposes a novel memetic algorithm (MA) underpinned by a variable neighborhood descent (VND) local search strategy to address this complex problem. This methodology demonstrates significant improvements, achieving mean makespan reductions ranging from 2.22% to 5.77% across diverse problem instances with varying numbers of machines and jobs. Key contributions include the development of an encoding scheme to model maintenance activities and machine-specific constraints, along with the design of a hybrid metaheuristic framework combining global exploration and local refinement. This work provides a foundation for future comparative studies, algorithm enhancements, and practical industrial applications. The approach offers a scalable and flexible solution to job-shop scheduling challenges involving dynamic efficiency and planned maintenance activities.