Gaussian Processes in Power Systems: Techniques, Applications, and Future Works

TL;DR

The paper tackles the challenge of increasing uncertainty in modern power systems by surveying Gaussian Process (GP) methods for uncertainty-aware analysis. It covers GP-based modeling across forecasting, steady-state/dynamic power-flow learning, and probabilistic optimization, emphasizing closed-form posteriors and principled uncertainty quantification. It also discusses risk assessment, optimization/control, co-optimization, and broader applications, while outlining challenges in interpretability, scalability, robustness, and online adaptiveness along with future directions like physics-informed kernels and integration with other data-driven approaches. The work provides a comprehensive reference for deploying GP-driven decision support in contemporary grids, enabling more reliable, data-informed grid operation and planning.

Abstract

The increasing integration of renewable energy sources (RESs) and distributed energy resources (DERs) has significantly heightened operational complexity and uncertainty in modern power systems. Concurrently, the widespread deployment of smart meters, phasor measurement units (PMUs) and other sensors has generated vast spatiotemporal data streams, enabling advanced data-driven analytics and decision-making in grid operations. In this context, Gaussian processes (GPs) have emerged as a powerful probabilistic framework, offering uncertainty quantification, non-parametric modeling, and predictive capabilities to enhance power system analysis and control. This paper presents a comprehensive review of GP techniques and their applications in power system operation and control. GP applications are reviewed across three key domains: GP-based modeling, risk assessment, and optimization and control. These areas serve as representative examples of how GP can be utilized in power systems. Furthermore, critical challenges in GP applications are discussed, and potential research directions are outlined to facilitate future power system operations.

Figures (3)

• Figure 1: Illustration of GP applications across various domains of power systems. GP-based learning techniques have been used to support load and renewable energy forecasting, static and dynamic modeling of power flows, and are integrated into power system optimization and control. These methods further contribute to power system risk assessment through improved forecasting, risk quantification, and mitigation strategies, enhancing secure and reliable grid operation. See Table \ref{['forecasting_summary']} and Table \ref{['Table_Optimization']} for specific works related to specific problems.
• Figure 2: Comparison of linear, polynomial, and GP regressions (using Square Exponential kernel with $\nu^2=1$) on $y = \sin(x) + 0.05 \cdot \varepsilon$, where $\varepsilon \sim \mathcal{N}(0, 1)$. Importantly, confidence interval information is only available with GP regression.
• Figure 3: Load forecasting example using GP with squared exponential kernel.