A robust soft decision-making method for uncertain production scheduling in semiconductor manufacturing

In this study, a robust soft decision-making method (RSDM) is proposed to solve the uncertain work area scheduling problem with random processing times in semiconductor manufacturing. To address the limitations of the conventional robust proactive approach, an improved two-layer soft scheduling framework is introduced into RSDM, which combines the concept of robust scheduling with a soft scheduling mechanism. In the offline layer of RSDM, a new form of soft schedule called the operation-priority based soft schedule is designed to preassign a global priority to each operation, and a two-stage genetic algorithm (2S-GA) is developed to obtain an initial soft schedule. Then in the online layer, a corresponding online dispatching rule is designed to make the remaining decisions dynamically on the premise that the initial soft schedule must be observed. Furthermore, a risk measure, called robust conditional value-at-risk (RCVaR), is introduced into the soft scheduling mechanism to evaluate the robustness of the soft schedule so that RSDM can optimise scheduling performance while also considering robustness. The comparative experimental results demonstrate the superiority of RSDM in addressing uncertain scheduling problems, especially in terms of robustness.