State Estimation for Low-Voltage Distribution System with High Proportion Distributed Energy Resource Based on Invariant Risk Minimization

Accurate perception of low-voltage distribution system state is critical for the control and operation of power grids. Recently, learning-based state estimation methods have been seriously challenged by the data shift induced by a high proportion of distributed energy resources. To address this issue, this paper proposes an improved learning-based low-voltage distribution system state estimation method. Firstly, a state estimation model for the low-voltage distribution system based on an adaptive neural network is established by utilizing the historical data of smart meters. Then, to eliminate the effects of data shift, a state estimation accuracy improvement method based on invariant risk minimization is proposed. Finally, the effectiveness of the proposed method is verified in the actual distribution network.