基于大数据的运载火箭总装系统智能优化调度理论与方法

项目名称： 基于大数据的运载火箭总装系统智能优化调度理论与方法

项目编号： No.U1537110

项目类型： 联合基金项目

立项/批准年度： 2016

项目学科： 机械、仪表工业

项目作者： 赵新明

作者单位： 上海交通大学

项目金额： 52万元

中文摘要： 为了满足航天产品高密度发射和快速响应需求，项目以运载火箭总装系统为研究对象，针对火箭总装系统具有的混流生产、多重资源约束、交货期固定和随机动态性等特点，突破传统调度理论和方法的瓶颈，建立数据驱动的“预测调度+实时预警+逆调度”机制。通过建立总装过程多维大数据模型，为调度提供数据基础；研究基于时间序列分析的预测优化调度方法，在零部件到达时间预测的基础上，实现总装物料消耗平准化和工位负载均衡化；研究基于关联关系分析的总装系统异常预警方法，挖掘总装完工时间关键影响因素，构建总装系统异常评估模型；研究基于无监督数据挖掘的智能逆调度方法，通过数据聚类和无模型自适应控制实现总装系统的逆调度。通过以上研究形成基于大数据的运载火箭总装系统智能优化调度理论与方法体系，为缩短运载火箭制造周期、实现生产持续优化奠定了坚实的理论和方法基础，对提升我国航天产品制造的智能化水平具有重要意义。

中文关键词： 火箭总装系统；预测调度；逆调度；大数据

英文摘要： In order to satisfy the demand of high density launch and quick response, this project investigates the scheduling problem of the rocket final assembly system. Against the characteristics of mixed flow, multiple resource constraints, fixed delivery time, and stochastic dynamics, the project breaks through the bottleneck of traditional scheduling theory and methods, and proposes a data driven mechanism of “predictive scheduling + real time monitor + inverse scheduling”. A multi-dimension big data model is developed to support the decision for the scheduling of rocket final assembly. The time series analysis based predictive scheduling approach is proposed to smooth the material consumption and to balance the workload. The correlation analysis based abnormal detecting method is proposed to real time monitor the final assembly system. The unsupervised data mining approach is established to realize the inverse scheduling. On basis of the above research, this project constructs the big data based intelligent scheduling theory and methodology for the rocket final assembly which lays a solid theoretical and technical foundation for the improvement of rocket production efficiency and constant optimization, and plays a significant role to realize the intelligent manufacturing of our country’s aerospace industry.

英文关键词： Final assembly system of rocket;Predictive scheduling;Inverse scheduling;Big data