Static Grid Equivalent Models Based on Artificial Neural Networks

Power systems are rapidly and significantly changing due to the increasing penetration of distributed energy resources (DERs) rapid growth widespread grid interconnections. An number operators is thus interested in reduced equivalent representation a large, interconnected power system reduce amount required computational data exchange, e.g., between operators. However, state-of-the-art equivalents become more inapplicable since they analytically calculated for one specific state. They cannot properly be adapted state changes behavior increasingly used controllers, such as reactive controllers DERs. Therefore, we propose an innovative based on artificial neural networks (ANN) which overcomes drawbacks follows: 1) Using supervised ANNs with feedforward recurrent architectures, can equivalently represented adaptively accurately. 2) A feature selection method identifies elements high sensitivity boundary enabling reduction ANN-based equivalent. 3) To guarantee confidentiality cybersecurity, additional unsupervised ANN, Autoencoder, obfuscation proposed exchanged among operators, while relevant information original preserved, maintaining estimation accuracy. Our approach analyzed evaluated two German benchmark grids representative scenarios. The simulation results demonstrate that outperforms radial independent method.