大数据环境下车间生产异常的智能发现与响应方法研究

发布时间：2018-05-07 21:58

本文选题：云制造 + 作业计划与调度　；参考：《哈尔滨理工大学》2017年硕士论文

【摘要】：云制造及工业大数据等新兴技术为制造企业带来机遇的同时,也为制造执行系统(Manufacturing Execution System,MES)的更新与升级带来了新的挑战。作业计划与调度作为贯穿制造执行系统的内部过程,是将MES与外部系统相连的主线。作业计划与调度优化是体现MES先进性的技术核心。首先,为了解决MES中作业计划与调度优化问题,采用基于软件体系结构优化的设计思路,对体系结构中的关键点进行深入分析。并设计了从作业计划静态化制定,到作业执行实时监控与主动感知,再进行生产异常智能响应,最后到作业调度动态调节的完整体系结构,并给出了调度优化问题的全局性解析,保证了理论研究与实践应用之间的高契合度。其次,对车间生产异常发现、异常智能化处理两大关键问题进行大数据环境下的拓展研究,利用数据挖掘的预测分析能力,打造具备生产异常发现与处理和自我分析能力的制造系统。针对生产异常发现问题,综合时间序列和因果关系两个维度,建立了基于时序序列上多决策树的车间异常事件预警模型,保证了预测异常结果的准确性和可靠性。针对生产异常智能化处理问题,借鉴预测式制造的先进想法,参照制造状况评估、早期异常诊断、未来失效时间推断以及主动维护的过程提出解决方案。并对切削关键设备的刀具组件进行衰退状态监测与量化,提出了基于时序的多神经网络决策模型,实现了多层次的刀具组件寿命预测目标,为故障诊测与健康管理在MES中应用提供了思路。最后,应用某电机厂为应用案例,利用IEC/ISO 62264标准、数据挖掘以及由虚拟化、服务化和面向服务的体系结构(Service-Oriented Architecture,SOA)等组成的云计算技术实现了MES作业计划与调度综合优化系统,验证了上述理论与方法的有效性。
[Abstract]:New technologies such as cloud manufacturing and industrial big data have brought opportunities to manufacturing enterprises, and at the same time, have brought new challenges to the updating and upgrading of manufacturing Execution systems. As an internal process running through manufacturing execution system, job planning and scheduling is the main line that connects MES with external system. Job scheduling and scheduling optimization is the core of MES technology. Firstly, in order to solve the problem of job scheduling and scheduling optimization in MES, the key points in the architecture are deeply analyzed by using the design idea based on software architecture optimization. And designed from the job plan static formulation, to the job execution real-time monitoring and the active perception, then carries on the production unusual intelligence response, finally to the job scheduling dynamic adjustment complete system structure. The global analysis of the scheduling optimization problem is given to ensure the high consistency between theoretical research and practical application. Secondly, to the workshop production abnormal discovery, the abnormal intelligence processing two big key question carries on the big data environment expansion research, uses the data mining forecast analysis ability, Build a manufacturing system with the ability to detect and process abnormal production and self-analysis. Aiming at the problem of abnormal discovery of production, combining two dimensions of time series and causality, a workshop anomaly event early warning model based on multiple decision trees on time series is established, which ensures the accuracy and reliability of predicting abnormal results. In view of the problem of intelligent processing of abnormal production, the advanced ideas of predictive manufacturing, the evaluation of manufacturing condition, the early abnormal diagnosis, the estimation of future failure time and the process of active maintenance are proposed. The tool component of cutting key equipment is monitored and quantified, and a multi-neural network decision model based on time series is put forward, which realizes the multi-level tool component life prediction goal. It provides some ideas for the application of fault diagnosis and health management in MES. Finally, using IEC/ISO 62264 standard, data mining and cloud computing technology composed of virtualization, service-oriented architecture and Service-Oriented Architecture (SOA), an integrated optimization system of MES job planning and scheduling is implemented. The validity of the above theory and method is verified.
【学位授予单位】：哈尔滨理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TB497

【相似文献】