以生产趋势预测为基础的主动式调度方法
Initiative Scheduling Method Triggered by Production Trend Prediction

为应对自适应制造需求由被动式调度向主动式调度的转变，提出一种以生产趋势预测为基础的主动式调度。该方法首先构建了实时状态模型，从制造资源、制造资源组合以及生产任务3个层面对生产过程的历史信息和实时状态进行建模，形成了具有时间序列的生产过程状态信息；针对生产过程的不确定性特征，采用贝叶斯网络推理方法，以生产过程状态信息作为输入，获取生产趋势的预测结果；将获取的异常趋势作为主动式调度的触发条件，通过扩展蒙特卡洛树搜索算法，利用其序贯决策能力生成以生产趋势预测为基础的主动式调度方案，从而实现生产过程的自适应制造。将提出的方法应用在某航天机加车间的实时生产过程调度中，验证了方法的有效性。
To satisfy the requirements for transition from passive scheduling to active scheduling, an active scheduling method based on production trend prediction was proposed. Its real-time model was established first. For manufacturing resources, their combination and production task, for historic information and real time state, for time series process information on production; for the uncertainty feature of production processes, the Bayesian network inference method was adopted, using the state information on production processes as input to infer the production trend and obtain the abnormal production trend as trigger conditions of active scheduling. By extending the Monte-Carlo tree search algorithm and using its sequential decision-making capacity, the initiative scheduling scheme is generated based on production trend prediction, and adaptive manufacturing is realized. Finally, the active scheduling method was applied to the real time production process analysis of aerospace machining job shop scheduling, thus verifying its effectiveness