多维时间序列关联分析方法在电力设备故障预测中的应用
习伟;李鹏;郭晓斌;许爱东;蒋愈勇;张利强;吴玉生
【期刊名称】《电网与清洁能源》 【年(卷),期】2014(000)012
【摘要】提出基于多维时间序列关联分析的电力设备故障预测方法：将电力设备拓扑网络设备节点的历史时间序列数据进行规范化，运用时间序列分解算法将时间序列进行分解；用一种时间序列模式表示方法，提取关键设备发生故障之前网络拓扑中设备节点的特征事件；并采用关联分析的方法挖掘设备指标运行趋势与设备工况之间存在的隐含关系，达到对故障或冲击进行有效预测的目的。实验证明，该方法能充分利用时间序列数据，发挥数据挖掘对于不确定性关系的分析和表达的优势，能够准确、有效地进行复杂电力设备故障预测。%This paper puts forward a prediction method for the equipment failure based on multidimensional time series which can normalize the historical time series data of power equipment topology nodes. The characteristic events of equip-ment topology nodes can be extracted before the key equipment failure. For the effective prediction of failures and shocks,this method is used to find out the relation between the operation trend of equipment indexes and the working conditions of the equipment. The experiment has proved that this method can make full use of time series data with advantages of analysis and expression of uncertainty relations in data mining. The method helps to make an accurate and effective prediction of the com-plex equipment failure.
多维时间序列关联分析方法在电力设备故障预测中的应用
