Abstract:The identification of the abnormal state of power is the key and difficult point for electricity link. In this paper, based on Metrological automation systemcan only collect the electricity consumption big data collected by the terminal smart energy meter, and study the method of identifying the abnormal status of electricity consumption. Firstly, based on the theory of large-scale data and high-dimensional random matrices, the characteristic spectrum distribution of the covariance matrix of large-dimensional random matrices is studied and analyzed; then, a power abnormal state identification method based on large data matrices is proposed based on the statistical properties of the matrix. Finally, a simulation study was conducted using the actual electricity data in Guizhou as an example. Simulation results show that the proposed method not only satisfies the urgent requirements of the grid for visibility, timeliness, reliability, and security, but also provides a new idea of intelligent and visual monitoring of data-driven power usage.