基于大数据的复杂生产线系统健康度实时评估方法  

Real-time health level assessment for complex production line system based on big data

作　　者：刘博元[1] 王焕钢[1] 范文慧[1] 肖田元[1] 

机构地区：[1]清华大学自动化系,北京100084

出　　处：《清华大学学报（自然科学版）》2014年第10期1377-1383,共7页Journal of Tsinghua University(Science and Technology)

摘　　要：复杂生产线系统由于设备规模大、影响因素多,导致运行过程存在着较强的不确定性。系统部署的传感器网络产生的实时大数据(流数据)可以作为系统状态监测的来源,然而传统数据处理方法并不能很好地对系统的健康度作出实时评价。该文以复杂生产线系统运行中的实时大数据(流数据)为基础,基于信息熵原理通过大数据分析方法量化分析系统内部属性间的行为模式和相关性关系,提出了一种基于大数据的复杂生产线系统健康度实时评估方法。磨矿生产线案例表明,该方法可以对复杂生产线的系统健康度作出实时评估。Complex production lines are usually of large scale and influenced by many factors,which leads to a strong uncertainty in system operation.The real-time big data（data stream）provided by the sensor network can serve as a source for status monitoring,but traditional data analytic method cannot well evaluate the system health.In this paper,based on the big data（streaming data）in the production line,the interior behavior patter and correlativity of a system was quantified by information entropy,with a real-time health level assessment method developed based on big data. Grinding production line cases show that the real-time health of a system can be evaluated using this method with credible results.

关 键 词：大数据 信息熵 相关性分析 实时评估 

分 类 号：TN914.42[电子电信—通信与信息系统]