Online error correction method of PMU data based on LSTM model and Kalman filter

The popularization of synchronous phase measurement unit (PMU) in distribution network improves the observability of power system effectively. In this paper, an online PMU data error correction algorithm based on LSTM model and Kalman filter is proposed. Firstly, the LSTM prediction model is trained by the historical amplitude data of PMU under the normal operation of power grid, and the PMU amplitude in the next period is estimated. Then the fusion algorithm proposed in this paper is used to fuse the estimated value of PMU amplitude and the actual measured value. Finally, the fused PMU amplitude data is compared with the measured value. When the error between the two is within a reasonable range, it is considered that there is no error in the measured value and no need to change. However, when the error exceeds the set threshold range, the measurement value acquisition error at this time is considered, and the fusion value is used to replace the measurement data. Through this method, real-time online error correction for PMU data is realized.