基于多树遗传编程的存储仓与组合设备集成系统优化

打开文本图片集

中图分类号：TP309 文献标志码：A 文章编号： 1000-5013（2026）01-0020-08

Abstract：To addressthe complex scheduling problem stocker-integrated cluster tools under frequent recipe switching in advanced semiconductor manufacturing，a multi-tree genetic programming method is proposed with theoptimization objective minimizing the makespan. Considering multiple practical constraints，including recipe switching，chamber cleaning，stocker capacity，and robot-handling operations，several representative operating scenarios are constructed to verify the effectiveness the proposed method. The experimental framework adopts a dual-tree structure：one syntax tree determines the wafer lot entry sequence，while the other specifies the assignment strategy processing chambers. The two syntax trees are jointly optimized through a cooperative co-evolution mechanism，achieving a favorable balance between search eficiency and solution quality. The proposed method is evaluated on three representative case studies and compared with manually designed scheduling rules. The experimental results show that the proposed method demonstrates superior scheduling performance and robustness across different operating conditions，reducing the average makespan byapproximately 25%-45% ：

Keywords： cluster tool； multi-tree genetic programming；scheduling； semiconductor manufacturing

随着半导体制造工艺复杂性和设备集成度的不断提高，组合设备因其并行处理能力和灵活配置方式被广泛应用于刻蚀、沉积、扩散等关键工艺环节[1]。（剩余11304字）