A Genetic Algorithm Based Artificial Neural Network for Production Rescheduling Problem

Production rescheduling plays an essential role in endorsing the effectiveness of a dynamic manufacturing environment. When the significant disruptive changes invalidate the original schedules, the rescheduling system should be adopted by responding quickly to lessen the effects on the performance of the production. Among the fourth industrial revolution, digital technologies (e.g., Internet of Things or IoT) and machine learning are creating new opportunities to execute production rescheduling. This paper presents a rescheduling approach based on a genetic algorithm (GA) and artificial neural network (ANN) to address the problem of flow shop scheduling with machine disruption. The objective is to find a new sequence or schedule of jobs that minimize makespan in satisfactorily computational time. This study first generates simulated scenarios of the interruptions. Then, we propose GA for solving each scenario. Secondly, we apply ANN to store the knowledge from simulated scenarios that can provide initial solutions for novel GA. It is found that the GA-based knowledge from ANN renders the new schedule 35.8% faster than the standard GA. Through observing the results, the proposed rescheduling methodology for flexible manufacturing not only has a productive performance in handling machine disruption in a scheduling problem but also contributes a faster new schedule to fill the gaps in state-of-the-art heuristic approaches whose computational time is inapplicable in implementation.