Multi-objective lot-streaming scheduling in cloud manufacturing system with variable sublots and intermingling setting considering service time availability

Recently, cloud manufacturing (CMfg) has been attracting increasingly attentions for its effectively integrating available services of various decentralized manufacturing resources on-demand. Considering that CMfg service providers (CMSPs) tend to place higher priorities on their local orders, they process the assignments from CMfg platform only with their remaining capacity (a series of available manufacturing time windows). Due to the diversity of customers’ demands, it is usually difficult to fit the CMfg assignments well into CMSP's fragmented available time slots. Especially the short windows are often ignored when they are not adequate for a task lot, inevitably leading to the waste of available manufacturing resources. To address this issue, we formulate the problem using a lot-streaming scheduling model for CMfg systems with manufacturing time availability, where lots are splitted into different sizes to fit the length of the available manufacturing time. Additionally, variable sublots, intermingling setting, and sequence/service-dependent setup times are considered. A multi-objective mixed-integer programming model is developed to optimize the makespan and total cost. Owed to the NP-hard nature of the problem, we design a multi-objective discrete differential evolution algorithm with variable-length individuals, in which a problem-specific encoding/decoding scheme, novel difference-based learning procedure, and critical path-based local search strategies are elaborated. Extensive experiments on the well-known Brandimarte benchmark confirm the proposed method's effectiveness and efficiency, yielding improvements of 3.07 %, 7.23 %, and 13.33 % in HV, GD, and IGD respectively compared to competing approaches. Its efficacy is further validated by a real-world case, reducing makespan by 36.21 % and total cost by 12.26 %.