The increasing number of distributed generators connected to distribution grids requires a reliable monitoring of distribution grids. Economic considerations prevent a full observation of distribution grids with direct measurements. First approaches using a limited number of measurements to monitor distribution grids exist, some of which use artificial neural networks (ANN). The current ANN-based approaches, however, are limited to static topologies, only estimate voltage magnitudes, do not work properly when confronted with a high amount of distributed generation and often yield inaccurate results. These strong limitations have prevented a true applicability of ANN for distribution system monitoring. The objective of this paper is to overcome the limitations of existing approaches. We do that by presenting an ANN-based scheme, which advances the state-of-the-art in several ways: Our scheme can cope with a very low number of measurements, far less than is traditionally required by the state-of-the-art weighted least squares state estimation (WLS SE). It can estimate both voltage magnitudes and line loadings with high precision and includes different switching states as inputs. Our contribution consists of a method to generate useful training data by using a scenario generator and a number of hyperparameters that define the ANN architecture. Both can be used for different power grids even with a high amount of distributed generation. Simulations are performed with an elaborate evaluation approach on a real distribution grid and a CIGRE benchmark grid both with a high amount of distributed generation from photovoltaics and wind energy converters. They demonstrate that the proposed ANN scheme clearly outperforms state-of-the-art ANN schemes and WLS SE under normal operating conditions and different situations such as gross measurement errors when comparing voltage magnitude and line magnitude estimation errors.

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

The increasing penetration from intermittent renewable distributed energy resources in distribution grid brings along challenges in grid operation and planning. To evaluate the impact on the grid voltage profile, grid losses, and discrete actions from assets (e.g. transformer tap changes), quasi-static simulation is an appropriate method. Quasi-static time series and Monte-Carlo simulation requires a tremendous number of power flow calculations (PFCs), which can be significantly accelerated with a parallel High-Performance Computing (HPC)-PFC solver. In this paper, we propose a HPC-PFC-solver-based grid simulation (parallel simulation) approach for a multi-core CPU platform as well as a greedy method, which can prevent the errors caused by simultaneous parallel simulation. The performance of the proposed approach and the comparison is demonstrated with two use cases.