基于CFSFDP算法的边缘电力数据异常检测

随着智能电网的不断发展,电力设备产生的数据量逐渐增多,如何利用电力数据成为电网发展的关键。为了保障电力数据的准确性,在边缘端快速检测并处理异常数据,提出了一种基于CFSFDP算法的电力数据异常检测的方法。该方法基于CFSFDP的假设,将局部密度较低且距高密度点较远的样本点定义为异常值,并创新使用了一种根据前后k值自动选择异常值的策略,解决了人工选择时存在主观因素影响的问题。通过与DBSCAN和LOF的比较表明,该方法能够快速、高效地找出电力数据中的异常值,适用于边缘电力数据异常检测。

Detection of Edge Power Data Anomaly Based on CFSFDP Algorithm

With the continuous development of smart grid, the amount of data generated by power equipment is gradually increasing. How to use power data becomes the key to the development of power grid. In order to ensure the accuracy of power data and detect and process abnormal data quickly at the edge, a detection method for power data anomaly based on CFSFDP algorithm is proposed. Based on the hypothesis of CFSFDP, the sample points with low local density and far away from high density points are defined as outliers, and a new strategy of automatically selecting outliers based on the k values before and after is used to solve the problem of subjective factors in manual selection. The comparison with DBSCAN and LOF shows that the proposed method can quickly and efficiently find the outliers in power data, and is suitable for outlier detection of edge power data.