Multi-Area Distribution System State Estimation Using Decentralized Physics-Aware Neural Networks

The development of active distribution grids requires more accurate and lower computational cost state estimation. In this paper, the authors investigate a decentralized learning-based system estimation (DSSE) approach for large grids. proposed decomposes feeder-level DSSE into subarea-level problems that can be solved independently. method is pruned physics-aware neural network (D-P2N2). physical grid topology used to parsimoniously design connections between different hidden layers D-P2N2. Monte Carlo simulations based on one-year load consumption data collected from smart meters three-phase power flow are developed generate measurement voltage data. IEEE 123-node selected as test benchmark algorithm against classic weighted least squares state-of-the-art approaches. Numerical results show D-P2N2 outperforms methods in terms accuracy efficiency.