A Bi-Level Optimization Model for Service Composition in Cloud Manufacturing Based on Global Benefit and Correlation Effect

In cloud manufacturing, manufacturing service composition (MSC) involves multiple associated services and multiple stakeholders such as service demanders, platform operators, and resource providers. The benefits and goals of all stakeholders as well as the correlations between services are critical factors influencing the efficiency and feasibility of MSC allocation. However, existing research focuses on allocating the MSC against demanders’ satisfaction with the highest quality of service (QoS, e.g., time, cost, quality) by assuming that each service is an independent entity. They have seldom considered the global interests of stakeholders and service correlations simultaneously. Addressing this gap, this study aims to (1) propose a comprehensive evaluation system based on the perspectives of demanders, operators, and providers, simultaneously. (2) define four types of correlations to assess the correlation effect between services in CMfg in terms of alliance, mutex, reciprocity, and competition. (3) establish a bi-level optimization model for MSC to maximize the benefits of all stakeholders and characterize the influence of service correlations on the combination feasibility and QoS value. (4) design a fast non-dominated sorting genetic algorithm with preconcentration and advancement (PANSGA-II) coupled technique for order preference by similarity to the ideal solution (TOPSIS) to solve the model. The preconcentration and advancement mechanisms facilitate the quality of the initial population and improve the connection between lower and upper-level models, respectively. Finally, numerical experiments and a case application demonstrated the effectiveness and practicability of the presented model and algorithm.