Non-Destructive Detection of Dismantled and Returned Electricity Meters Based on Power Images and Deep Learning

Information mining and utilization of the large and rapidly growing number of dismantled energy meters will provide valuable auxiliary decision-making information. However, currently there is a lack of intelligent detection methods and the detection efficiency is low. To this end, this article proposes a non-destructive detection method based on electrical power imaging and deep learning. This method obtains the transient power response of the electric energy meter under different input signals, thereby generating an electric power portrait representing the internal structural characteristics of the electric energy meter, which is represented in the form of a grayscale image. Finally, the fault is detected through Convolutional Neural Networks (CNN). The electric energy meter performs rapid automatic identification. This method can achieve non-destructive testing without opening the cover, reduce the risk of damaging the original structure of the electric energy meter, and save testing costs and time. Without opening the cover, the electric power portrait grayscale image is a better choice to fully express the internal structural characteristics of the electric energy meter. Combined with the fault identification method based on deep learning, it has high identification accuracy. Experimental results on real data sets verify the effectiveness and stability of this method.