Data-Mining-Based Real-Time Optimization of the Job Shop Scheduling Problem

Abstract

In the job-shop scheduling field, timely and proper updating of the original scheduling strategy is an effective way to avoid the negative impact of disturbances on manufacturing. In this paper, a pure reactive scheduling method for updating the scheduling strategy is proposed to deal with the disturbance of the uncertainty of the arrival of new jobs in the job shop. The implementation process is as follows: combine data mining, discrete event simulation, and dispatching rules (DRs), take makespan and machine utilization as scheduling criteria, divide the manufacturing system production period into multiple scheduling subperiods, and build a dynamic scheduling model that assigns DRs to subscheduling periods in real-time; the scheduling strategies are generated at the beginning of each scheduling subperiod. The experiments showed that the method proposed enables a reduction in the makespan of 2–17% and an improvement in the machine utilization of 2–21%. The constructed scheduling model can assign the optimal DR to each scheduling subperiod in real-time, which realizes the purpose of locally updating the scheduling strategy and enhancing the overall scheduling effect of the manufacturing system.