A Survey of AI-enabled Dynamic Manufacturing Scheduling: From Directed Heuristics to Autonomous Learning

As one of the most complex parts in manufacturing systems, scheduling plays an important role in the efficient allocation of resources to meet individual customization requirements. However, due to the uncertain disruptions (e.g., task arrival time, service breakdown duration) of manufacturing processes, how to respond to various dynamics in manufacturing to keep the scheduling process moving forward smoothly and efficiently is becoming a major challenge in dynamic manufacturing scheduling. To solve such a problem, a wide spectrum of artificial intelligence techniques have been developed to (1) accurately construct dynamic scheduling models that can represent both personalized customer needs and uncertain provider capabilities and (2) efficiently obtain a qualified schedule within a limited time. From these two perspectives, this article systemically makes a state-of-the-art literature survey on the application of these artificial intelligence techniques in dynamic manufacturing modeling and scheduling. It first introduces two types of dynamic scheduling problems that consider service- and task-related disruptions in the manufacturing process, respectively, followed by a bibliometric analysis of artificial intelligence techniques for dynamic manufacturing scheduling. Next, various kinds of artificial-intelligence-enabled schedulers for solving dynamic scheduling problems including both directed heuristics and autonomous learning methods are reviewed, which strive not only to quickly obtain optimized solutions but also to effectively achieve the adaption to dynamics. Finally, this article further elaborates on the future opportunities and challenges of using artificial-intelligence-enabled schedulers to solve complex dynamic scheduling problems. In summary, this survey aims to present a thorough and organized overview of artificial-intelligence-enabled dynamic manufacturing scheduling and shed light on some related research directions that are worth studying in the future.