A Comprehensive Approach to Implement Monitoring and State Estimation in Distribution Grids with a Low Number of Measurements

This work addresses the monitoring and state estimation of electrical grids, especially at the distribution grid level. This topic becomes relevant because today’s distribution grids are not always designed for the current expansion of decentralized feed-in, which can lead to congestion within the grid. If congestion is detected by suitable monitoring, it can be resolved by measures, e.g., adjustment of generator power. For economic and technical reasons, however, grid monitoring cannot be implemented with a similarly high measurement density as in transmission grids. Two new monitoring methods, which are designed for low measurement density, are therefore presented for use in real-time grid operation. First, a heuristic monitoring method (HMM) is presented, which does not require pseudo-measurements and estimates voltage magnitudes and line loadings. Second, a monitoring method based on artificial neural networks (ANN) is presented. With appropriate training, the method can estimate grid variables, e.g., voltage magnitudes or line loadings, with high accuracy. Various extensions of the method allow the estimation of the complete grid state. The methods are tested on thousands of test scenarios using a comprehensive evaluation methodology. A reference state estimation algorithm, the HMM, and three variants of the ANN-based monitoring are compared for different measurement locations with typical measurement errors, extreme measurement errors, and topology errors. Each method is classified according to the most appropriate application. For measurement infrastructure planning, a concept is presented to determine suitable measurement locations for the use of one of the monitoring methods. A user defines the required accuracy of the monitoring. Using the current grid model and information about potential measurements, heuristic optimizers or a rule-based strategy can identify additional measurement points that fulfill multiple requirements: the relevant estimation errors should be below a certain threshold. At the same time, capital expenditures for the installation and operation of the measuring equipment should be minimal. After completion of the optimization, several possible measurement configurations are presented with their average and maximum errors and the projected capital expenditures. To this end, a web application is developed.