Research On Power-stealing Behavior Mining Strategy Based On Power Consumption Characteristics

Since the State Grid Company issued the China Smart Grid Plan,with the promotion of the State Grid Company,smart meters have gradually entered millions of households.After the popularity of smart meters,it is increasingly difficult to implement illegal acts such as electricity theft in technology.In order to avoid the investigation of users’ electricity information collection system and other business applications,electricity theft often depends on high-tech means,so it gradually becomes concealed,which causes great difficulties to the operation of power supply companies.In view of the high rate of power theft by specialized users and the difficulty of detecting the mode of power theft in abnormal electricity usage detection,and the insufficient training set in the process of finding power theft users by using the power theft detection model,this paper puts forward an unsupervised method of power theft mining strategy based on power usage feature analysis,quantifies the abnormal degree of different power usage behaviors in massive data,and extracts it as power usage Charge sequence,and according to the different measurement methods and power characteristics of specialized users,a set of power characteristics is constructed,including the power unbalance feature sequence calculated by outlier algorithm,the voltage anomaly feature sequence considering the duration of voltage loss,the current anomaly feature sequence based on the clustered results of daily current curve,and the evaluation index of abnormal power consumption is calculated.An abnormal electricity behavior detection model is designed.Through an example,it can be verified that this detection model can accurately find power users and time of power theft,and provide strong support for enterprises to detect power theft.In addition,according to the requirements of the current national power enterprise to improve the information management level,this paper designs a set of abnormal electricity query system,which can visualize the algorithms and detection models involved in the previous section.First of all,this paper analyses the functions that the abnormal electricity query system needs to achieve,and the abnormal electricity query system passes through Spring Boot,an open source framework on Java platform,completes the platform building.With the help of the image display function in Echart open source visualization library,the information required by power supply enterprises is displayed.My SQL database is introduced to facilitate the storage and interaction of power information,and the development of Web page client is realized.The development of Web client enables staff to realize data mining and visualization without technical operation,which facilitates the mining of abnormal electricity usage behavior.