基于聚类分析的电厂设备监测参数异常搜索方法研究

Study on the anomaly search method of power plant equipment monitoring parameters based on cluster analysis

针对电厂设备异常参数识别精度不高、搜索定位不准确、响应速度慢等问题,对电厂设备异常参数搜索方法展开研究。基于聚类分析算法提出了一种电厂设备监测参数异常搜索方法,对其相关性进行了分析,确定全局指标排除异常相关值,并在某电厂中压网络电厂水电站电池储能系统和20 kV线路进行现场试验与调试。试验结果表明:聚类分析算法可准确提取异常参数信息,对异常值进行快速搜索,可有效提高电力系统数据传输可靠性。当系数为0.756时,数据集中度较差,异常数据分布距离标准线的偏差较远,短期闪烁指标越小,数据越集中,验证了设计的合理性。

In view of the problems of low accuracy in detecting abnormal parameters,inaccurate search and positioning and slow response in power plant equipment monitoring,based on cluster analysis algorithm,a abnormality search method for monitoring parameters of power plant equipment is proposed.The correlation was analyzed,determine the global index exclusion of abnormal correlation values.Results of the field test and commissioning test on the battery energy storage system and a 20 kV line in a medium voltage network power plant show that the cluster analysis algorithm can accurately extract the abnormal parameter information.When the coefficient is 0.756,the data shows low concentration and the abnormal data distribution is far from the standard line.And the smaller the short-term flicker indicator is,the more concentrated the data is.It can be concluded that the anomaly search method for power plant equipment monitoring parameters,based on cluster analysis greatly improves the reliability of data transmission in power systems,realizes the reliable data interaction,and verifies the rationality and technical advantages of the algorithm design.