Scholarlink

Job shop scheduling in Computer Numerical Control (CNC) manufacturing enterprises is a highly complex problem, particularly under stochastic disruptions such as machine breakdowns, fluctuating processing times, and unpredictable job arrivals. To tackle these issues, this paper presents the Generative AI Scheduler, a scheduling system built on an augmented Conditional Generative Adversarial Network (CGAN) reinforced with Recurrent Neural Networks (RNN). The proposed Generative AI Scheduler effectively captures the temporal dependencies in job sequences, which also dynamically adjusts to production environment uncertainties. According to the simulation results, the proposed Generative AI Scheduler outperforms conventional heuristic techniques like Genetic Algorithms (GA) and Shortest Processing Time (SPT) in terms of robustness and efficiency while handling stochastic disruptions. The proposed framework consistently outperforms conventional scheduling adaptability and production stability methods across a range of CNC job shop scenarios. These findings highlight the potential of RNN-enhanced CGANs to transform job shop scheduling by enabling a data-driven, intelligent solution for autonomous production planning in mechanical manufacturing enterprises.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Conditional Generative Adversarial Networks for Robust Job Shop Scheduling in CNC-Based Manufacturing Systems

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Ngoc Cuong Truong,
  • Xuan Van Tran,
  • Quoc-Chi Nguyen

摘要

Job shop scheduling in Computer Numerical Control (CNC) manufacturing enterprises is a highly complex problem, particularly under stochastic disruptions such as machine breakdowns, fluctuating processing times, and unpredictable job arrivals. To tackle these issues, this paper presents the Generative AI Scheduler, a scheduling system built on an augmented Conditional Generative Adversarial Network (CGAN) reinforced with Recurrent Neural Networks (RNN). The proposed Generative AI Scheduler effectively captures the temporal dependencies in job sequences, which also dynamically adjusts to production environment uncertainties. According to the simulation results, the proposed Generative AI Scheduler outperforms conventional heuristic techniques like Genetic Algorithms (GA) and Shortest Processing Time (SPT) in terms of robustness and efficiency while handling stochastic disruptions. The proposed framework consistently outperforms conventional scheduling adaptability and production stability methods across a range of CNC job shop scenarios. These findings highlight the potential of RNN-enhanced CGANs to transform job shop scheduling by enabling a data-driven, intelligent solution for autonomous production planning in mechanical manufacturing enterprises.