数据驱动下窃电智能分析系统的研究

Research on the Electricity Stealing Intelligent Analysis System Driven by the Data

随着智能电表的推广及用电信息采集系统建设的逐步完善,电网公司积累了海量用户用电数据,为大数据技术在电力领域的应用提供了基础。基于数据挖掘思想针对行业分类及用户特征,对海量的多元计量数据进行窃电用户分析,从中挖掘出窃电用户的关联性多维度特征参量,建立神经网络构建数据驱动下的窃电预测模型。系统测试验证了预测窃电值与实际窃电值有很高的吻合度,在一定程度上反映了模型预测的准确性。

With the widely use of the smart meters and the gradual improvement of the instruction of power user electric energy data acquire system,a mass of data is accumulated by the state grid which is the basis of the application of the big data technologies on the electricity field.In this paper,the mass multivariate measurement data is analysed for the industry classification and custom characteristics based on the data mining ideas.Then the related multi-dimensions characteristics of the electricity stealing customs are mined and the electricity stealing prediction model is built based on the neural network driven by the data.The system test shows that the predicted value of electricity stealing is in good agreement with the actual value,which reflects the accuracy of the model prediction to a certain extent.