A new customer-oriented multi-task scheduling model for cloud manufacturing considering available periods of services using an improved hyper-heuristic algorithm

Mengjiao Chen, Jiyuan Xu, Wenyu Zhang, Zhenghui Li

Abstract: Efficiently assigning services to multiple tasks decomposed from manufacturing orders is a critical challenge in cloud manufacturing. While multi-task scheduling has gained attention, existing approaches often overlook key factors such as different expectations of customers for various non-functional attributes of services and limitations on available periods of services, which hinder effective and efficient service scheduling. This challenge is particularly pronounced in industries such as automotive and healthcare manufacturing, where timely service delivery and individualization are crucial. To address this, a new customer-oriented multi-task scheduling model is proposed, which incorporates service availability and uses soft constraints for customer expectations that are balanced by a penalty function to improve customer satisfaction. An improved hyper-heuristic algorithm with a new encoding and decoding scheme is developed, featuring three new low-level heuristics and a reinforcement learning-based high-level strategy to enhance scheduling performance. Comparison experiments using the Gurobi solver validate the model’s solution accuracy on small-scale instances. Moreover, experiments against five baselines for moderate-scale and large-scale instances demonstrate that the proposed model improves solution quality by an average of 31.60% and reduces computational time by an average of 46.35%, highlighting its potential to optimize multi-task scheduling on cloud manufacturing platforms while meeting diverse customer requirements.