Aiming at the problems such as the existence of monitoring blank spots in the operation state of the electric energy meter, the difficulty in obtaining relevant power data due to the professional barriers in business management, and the single existing monitoring method, a data-driven online monitoring method of running electric energy meter state is proposed. The method covers the low-voltage side, the medium-voltage side and the gateway side.On the low-voltage side, the misaligned electric energy meter is located by the method of sub-component comparison method, which provides a new calculation method for the online monitoring of the running state of the electric energy meter on the low-voltage side. On the medium-voltage side, considering the external characteristics of the transformer, the three-phase line voltage on the high-voltage side of the transformer is obtained by recursion. The optimization goal is to minimize the sum of the squares of the difference between the recursive line voltage and the line voltage at the reference point. Constraints includes angular offset and other electrical node constraints. On the high-voltage side, It proposed three-level power comparison models which includes main transformer model, bus level and line level electricity comparison model. Then, cross-location method is used to calculate the operating error of the power meter at the gateway side. Finally, the validity of the proposed method is verified by practical examples.