Robust and Fast Data-Driven Power System State Estimator Using Graph Neural Networks

Abstract

The power system state estimation (SE) algorithm estimates the complex bus voltages based on the available set of measurements. Because phasor measurement units (PMUs) are becoming more widely employed in transmission power systems, a fast SE solver capable of exploiting PMUs' high sample rates is required. To accomplish this, we present a method for training a model based on graph neural networks (GNNs) to learn estimates from PMU voltage and current measurements, which, once it is trained, has a linear computational complexity with respect to the number of nodes in the power system. We propose an original GNN implementation over the power system's factor graph to simplify the incorporation of various types and numbers of measurements both on power system buses and branches. Furthermore, we augment the factor graph to improve the robustness of GNN predictions. Training and test examples were generated by randomly sampling sets of power system measurements and annotated with the exact solutions of linear SE with PMUs. The numerical results demonstrate that the GNN model provides an accurate approximation of the SE solutions. Additionally, errors caused by PMU malfunctions or the communication failures that make the SE problem unobservable have a local effect and do not deteriorate the results in the rest of the power system.

Distributed Nonlinear State Estimation in Electric Power Systems using Graph Neural Networks

“…The power system augmented factor graph has bounded node degree, prediction for a single node requires only information from the node's K-hop neighborhood, and the number of GNN layers is small. Analogously to [18], we can conclude that the computational complexity of the proposed model's inference for a single state variable is constant, which implies that the overall complexity of the proposed GNN-based SE is O(n). The implementation of the GNN model in large-scale networks can be further improved by distributing the inference computation among local processors in the power system, avoiding the communication delays between the measurement devices and the central processing unit in the centralised SE implementation.…”

“…Contributions: In this work, we propose a data-driven nonlinear state estimator based on graph attention networks [15] operating on the factor-graph-like structure [16] obtained by transforming the bus/branch power system model. The proposed approach is an extension of our previous work on linear SE with PMUs [17] and linear SE with PMUs considering covariances of measurement phasors represented in rectangular coordinates [18]. The proposed method takes into account all of the legacy measurements, as well as bus voltage and branch current phasor measurements, and provides a trivial way to remove or add additional measurements by altering the corresponding factor nodes in the graph.…”

“…We designed the GNN-based state estimator to be fast, robust to ill-conditioned scenarios, and an accurate nonlinear SE approximator, once trained on the relevant dataset, labelled by the Gauss-Newton SE solver. Similar to the linear SE with PMUs model [18], the proposed model has O(n) computational complexity during the inference and can be distributed among geographically separated processing units. We evaluated the performance of the proposed method by testing on various data samples, including unobservable cases caused by communication errors or measurement device failures and scenarios corrupted by malicious data injections.…”