终端用电信息智能检测与精准计量数据异常识别

Intelligent detection of terminal electricity consumption information and accurate identification of anomalies in metering data

为保证终端用电计量数据采集的精准性,对终端用电信息采集过程中异常计量数据精准识别方法进行研究。分析计量装置采集过程中的异常情况,结合前馈神经网络和局部离群因子算法,构建异常计量数据精准识别模型;通过双层前馈神经网络模型,识别计量装置采集过程中的电能质量扰动类型结果后,将识别结果输入到局部离群因子算法中,通过计算局部离群因子,精准确定异常计量数据。测试结果显示,该方法能够获取各个扰动状态的数据点数量;有效计算所有数据的局部离群因子,确定数据中的异常计量数据,并呈现异常计量数据的精准识别结果。

In order to ensure the accuracy of terminal electricity metering data collection,the accurate identifica⁃tion method of abnormal metering data in the process of terminal electricity consumption information collection was studied.The anomalies in the collection process of the metering device were analyzed,and the accurate identifica⁃tion model of abnormal metering data was constructed by combining the feedforward neural network and the local outlier factor algorithm.Through the double-layer feedforward neural network model,the results of power quality disturbance types in the acquisition process of metering devices were identified,and the identification results were input into the local outlier factor algorithm,and the abnormal metering data was accurately determined by calculat⁃ing the local outlier factors.The test results showed that this method could obtain the number of data points for each disturbance state.Effectively calculated the local outlier factors of all data,determined the outlier metric data in the data,and presented accurate identification results of outlier metric data.