Abstract

The development of Industry 5.0 focuses on customization, personalization in production, and the innovative thinking of employees, elevating the value of human contribution. Design, being an innovation-driven domain, demands greater flexibility in resource allocation. Consequently, rapidly and effectively allocating cloud service resources for personalized design tasks becomes crucial. With the emergence of the Industrial Metaverse, which blurs the boundaries between real and virtual design and manufacturing, it is gaining increasing attention. To embrace the advent of Industry 5.0 and the Industrial Metaverse, swift collaborative cloud services for design and manufacturing resources are essential. In this context, this article introduces a novel approach combining complex networks with the Non-dominated Sorting Genetic Algorithm III (NSGA-III), aimed at rapidly optimizing the dynamic allocation of distributed design resources (DRs). Initially, a multipartite graph is created from raw data and mapped to multiple bipartite graphs to identify key nodes in the network through intersection. Subsequently, these key nodes are used as reference points in the NSGA-III algorithm to achieve high-quality cloud service combinations, meeting the needs of design tasks with multiple subtasks, and related multi-objective optimization, including time, cost, reliability, maintainability, and reputation associated with the design. Finally, the Pareto service combinations obtained are used to construct a new complex network and employ the Girvan-Newman algorithm based on edge betweenness to identify community structures. In case of anomalies in the best service combination, alternative options can be swiftly searched from the identified communities, thereby enhancing the resilience of the cloud service process. Experimental results demonstrate the method's advantages in recovery and robustness, contributing significantly to the optimization of rapid cloud service allocation for DRs in the context of Industry 5.0.