A Novel State Estimation Method for Modern Power Systems Based on Multi-Source Data Cleaning

This paper proposes a state estimation method based on multi-source data cleaning and fusion to address issues of poor data quality, low estimation accuracy and low efficiency in the measurement of renewable energy distribution grids. First, a method is proposed for identifying and correcting poor data using a Temporal Convolutional Network (TCN) and a Bidirectional Long Short-Term Memory Network (BILSTM), to clean real-time, multi-source measurement data. Secondly, a hybrid linear state estimation method considering renewable energy grid connection is employed to reflect the real-time state of the distribution grid. Simulation results demonstrate that the proposed data cleansing method exhibits high identification rates and correction accuracy, while the proposed state estimation method has high accuracy and real-time performance. This provides a solid foundation for the management and operation of distribution grids and smart power systems.