Neural Network-Driven State Estimator for European-Type Unbalanced LV Grids

The imperative to swiftly estimate the state of unbalanced low voltage distribution grids is paramount for effective flexibility management, where machine learning presents an efficient possibility suitable for real-time applications. This research focuses on the design and evaluation of a neural network-based state estimator tailored for unbalanced LV distribution grids. In addition, an approach to generate training datasets utilizing real-world grid profiles is proposed. The estimator’s performance undergoes thorough evaluation across various simulated scenarios conducted on a real grid located in northern Spain. These assessments showcase its resilience and precision under diverse operational conditions.