Multi-feature Fusion based Anomaly Electro-Data Detection in Smart Grid

In recent years, great losses have been caused to power supply company by abnormal electricity consumption. Especially, electricity stealing behavior not only damaged the power facilities, but also easily triggered the fire, and threatened the safe and stable operation of the power grid. Thus there arises the need to develop a scheme that can detect these thefts precisely in the complex power grid. Hence, this paper proposes a novel method which is base on multi-feature fusion to detect anomaly electro-data. This method adopted an unsupervised learning, which can be better applicable to the situation of few samples of the anomaly electro-data. Further, through the analysis of the related electrical parameters, such as voltage, current and power factors, this method can detect the most anomaly electro-data. The results of the case analysis show that our method can detect more exact and stable than the traditional methods.