Anti-Stealing Method Based On Electricity Information Collection System And Application

The anti-stealing intelligence has gradually become the inevitable way for the grid power supply company to ensure the steady development of the power economy in the Taiwan area under its jurisdiction.In this paper,the anti-stealing research based on K-means and LSTM is adopted.According to the status quo and existing problems of power grid anti-stealing decision-making,the anti-stealing method and algorithm of anti-theft are researched and summarized.And the research status of various preventive measures,K-means clustering algorithm and long-short memory network algorithm.Firstly,the related information of the electricity information collection system,the calculation principle of the high-voltage metering device,the common methods of stealing electricity and the characteristics of the electricity theft are studied in detail.The tampering and stealing characteristics of common tampering methods were studied.Secondly,a data-driven anti-stealing decision model is established.Through K-means clustering algorithm,the pre-processing of clustering of electricity information big data is used to reduce the dimension,and the anti-stealing decision model of long and short memory neural network is established to construct a deep learning for each group of historical data.The model is trained.Finally,the validity of the tampering of the sample data based on the anti-stealing decision model of K-means and LSTM is verified by case analysis.The results show that the anti-stealing decision model can diagnose the true rate of thief abnormal data up to 95.3%,the true negative rate is as high as 99.55%,while the negative positive rate is only 0.39%,and the accuracy of the tampering method is as high as 95.77%,which is in line with the actual anti-theft production requirements.At the same time,the guiding role of K-means and LSTM anti-stealing decision model system is verified by three actual field cases in the application report,and the effectiveness of the anti-stealing decision model is further verified.