基于熵值法和DEA的电力监测异常数据自动识别算法研究

Research on Automatic Identification Algorithm of Abnormal Data in Power Monitoring Based on Entropy Method and DEA

为了有效提高电力监测异常数据识别准确率,确保异常数据识别效果,提出了基于熵值法和DEA的电力监测异常数据自动识别算法。应用数据脱敏、数据填补、颠簸去除等方式处理电力监测异常数据,还原原始电力监测数据,提升脱敏后数据完整性,降低颠簸数据对电力监测异常数据精度的影响。以经过数据处理的电力监测异常数据为基础,通过熵值法确定目标的属性和权重作为度量标准,采用基于熵值法改进的最近邻聚类算法,实现数据聚类,以数据聚类结果为输入,构建DEA模型,实现电力监测异常数据自动识别。实验结果表明,该算法数据聚类后的平均检测率约为92%,异常数据识别正判率约为95%,误判率约为3.5%,可有效识别出异常月负荷曲线的异常数据点和不同负荷量的异常数据,且识别结果与实际负荷曲线的趋势一致,具有较好的识别效果。

In order to effectively improve the recognition accuracy of abnormal data in power monitoring and ensure the recognition effect of abnormal data,an automatic recognition algorithm of abnormal data in power monitoring based on entropy method and DEA is proposed.Through data desensitization,data filling and bump removal,the abnormal data of power monitoring are processed,the original power monitoring data are restored,the integrity of desensitized data is improved,and the impact of bump data on the accuracy of abnormal data of power monitoring is reduced.Based on the abnormal data of power monitoring after data processing,the attribute and weight of the target are determined by entropy method as the measurement standard,and the improved nearest neighbor clustering algorithm based on entropy method is used to realize data clustering.Taking the data clustering results as the input,DEA model is constructed to realize the automatic identification of abnormal data of power monitoring.The experimental results show that the average detection rate of the algorithm after data clustering is about 92%,the positive judgment rate of abnormal data recognition is about 95%,and the false judgment rate is about 3.5%.It can effectively identify the abnormal data points of abnormal monthly load curve and abnormal data of different load quantities,and the recognition result is consistent with the trend of actual load curve,which has a good recognition effect.