Pattern recognition of potential stealing behavior based on outlier data mining algorithm

The power theft not only seriously damages the economic interests of the power enterprises, but also poses a potential threat to the safe and stable operation of the power grid. In order to effectively cope with this challenge, this paper puts forward the pattern recognition of potential power stealing behavior based on outlier power consumption data mining algorithm. Sensor nodes are arranged in the monitoring area to collect user electricity consumption information, and according to the collection results, the characteristic data of potential electricity stealing behavior are preprocessed; The outlier data mining algorithm is applied to construct the error objective function, and the calculation process is controlled and standardized. According to the activation function, the pattern recognition results of potential electricity stealing behavior are obtained. The experimental results show that this method can operate normally, and the user's stealing behavior can be quickly identified by using the significant change of voltage.