Artificial Intelligence-Based State Estimation in Power System and Phasor Measurement Units

Abstract

In the dynamically varying modern power system, it is challenging to monitor and estimate state variables of the network. In this work, state variables such as voltage magnitude and phasor angle are estimated using artificial intelligence (AI) techniques such as machine learning (ML) and deep learning (DL) methods. Among various ML models available, the few models considered in the proposed work for the state estimation (SE) in power system are decision tree, support vector machine, Ensemble boost, Ensemble bag (Eboost/bag), and artificial neural network (ANN). Similarly, DL models such as gate recurrent unit (GRU), long short-term memory (LSTM), and bidirectional LSTM are proposed for the SE. Among all these AI techniques, ANN and GRU gives better results than other models. Among them, the performance of GRU, a DL tool is found as best when compared to ANN, the ML. The accuracy of AI techniques are measured using evaluation metrics. The obtained results from AI compared to conventional techniques, weighted least square, and regularized least square method. Phasor measurement units is contributing more to minimize the SE errors based on its maximum observability and proved by using IEEE 14 and 30 bus systems.