面向多源传感事件序列的时序依赖关联挖掘方法

Temporal Dependency Mining from Multi-sensor Event Sequences for Predictive Maintenance

随着物联网的发展,工业设备上安装了大量的传感器,用来监控工业设备的状态.为了提高设备的可靠性,需要通过对多源传感数据的分析,来预测设备的故障,以便安排预测性维护.工业设备之间的相互作用往往具有传递性,存在复杂的时序依赖关系.因此本文旨在挖掘传感器异常事件之间复杂的时序依赖关系,找到设备故障的根本原因,从而避免故障再次发生.本文通过识别多源传感器事件的频繁共现模式,提出一种新的时序依赖关联挖掘算法“GFE-mining”,挖掘事件之间复杂的时序依赖关系,形成异常预测模型.并通过实验来证明该方法的有效性.

With the development of the IoT,a large number of sensors are installed on industrial equipment to monitor the status of industrial equipment.In order to improve the reliability of the equipment,it is necessary to predict the failure of the equipment through the analysis of multi-source sensor data in order to arrange predictive maintenance.Industrial equipment often has a transitive interaction,and there is a complex time-dependent relationship.Therefore,the purpose of this paper is to mine the complex time dependence between sensor abnormal events and find the root cause of equipment failure,so as to avoid the reoccurrence of failure.In this paper,we propose a new time-dependent association algorithm“GFE-mining”by identifying frequent co-occurrence patterns of multi-source sensor events,mining the complex time-dependent relationship between events,and forming an anomaly prediction model.And through experiments to prove the effectiveness of the method.