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

Bendong Tan, Tong Su, Yu Weng, Ketian Ye, Parikshit Pareek, Petr Vorobev, Hung Nguyen, Junbo Zhao, Deepjyoti Deka · 2025

Read Gaussian Processes in Power Systems: Techniques, Applications, and Future Works by Bendong Tan, Tong Su, Yu Weng, Ketian Ye, Parikshit Pareek, Petr Vorobev, Hung Nguyen, Junbo Zhao, Deepjyoti Deka (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-15950.txt.

Introduction

Bendong Tan and co-authors review Gaussian processes in power systems in a 2025 paper. Cleaner energy sources and smart sensors flood modern grids with data and doubt, and grid work needs tools that state their own uncertainty.

Gaussian processes give a probability-based frame for modeling, forecasting, and control without fixed formulas. The review sorts uses into modeling, risk assessment, and optimization with control. Each area shows how the method sharpens grid analysis and decisions. Hurdles include scale, speed, and messy real data. The authors map open problems and next steps for wider grid use. Power engineers get a full guide to a flexible statistical tool.

Catalog record

CREATORBendong Tan, Tong Su, Yu Weng, Ketian Ye, Parikshit Pareek, Petr Vorobev, Hung Nguyen, Junbo Zhao, Deepjyoti Deka
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE63 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-15950

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Gaussian Processes in Power Systems: Techniques, Applications, and Future Works Bendong Tan, Tong Su, Yu Weng, Ketian Ye, Parikshit Pareek, Petr Vorobev, Hung Nguyen, Junbo Zhao, Deepjyoti Deka 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. I Introduction The extensive reliance on fossil fuels has drastically elevated carbon emissions, presenting a substantial threat to global climate stability and sustainable development. As the principal medium for energy production and transmission, power systems are positioned at the forefront of the energy transition, playing an indispensable role in global carbon reduction efforts [ 1 ] . This transition has catalyzed the widespread adoption of renewable energy sources (RESs), distributed energy resources (DERs), and flexible loads, such as electric vehicles (EVs) [ 2 ] . While these components contribute to decarbonization, they simultaneously introduce considerable uncertainties and volatilities, posing challenges to the secure operation of power systems. Furthermore, as these elements are primarily integrated into the grid through power electronic devices, the overall system complexity is substantially increased, complicating both structural and parametric aspects of system modeling. Consequently, physics-based power system analyses may become either time-intensive or imprecise in the face of such uncertainties and complexities. The proliferation of advanced sensing technologies, such as phasor measurement units (PMUs), smart meters, Internet of Things-related sensors, etc, has resulted in power systems generating more comprehensive and valuable data than ever before. This evolution is propelling power systems into a new era of digitalization and intelligence, offering a promising avenue for addressing the aforementioned challenges. As a result, there is a growing demand for data-driven analytical methods that not only extract critical steady-state and dynamic insights from data, but also improve the accuracy of physical system modeling. A variety of machine learning (ML) techniques have been introduced for power system analysis, including shallow learning methods such as support vector machines (SVM) [ 3 ] , as well as newer deep learning models like Transformer [ 4 ] . These approaches offer the advantage of being model-free and capable of generalizing to new power system scenarios, providing a degree of adaptability. However, learning models solely on data makes them prone to overfitting or incorporating unwanted biases, especially when the dataset is small [ 5 ] . To address these limitations, Bayesian methods have been developed to integrate prior knowledge into model learning. Of particular note is the framework of GP, an important Bayesian learning paradigm, that generalizes multivariate Gaussian distributions by placing distributions over functions [ 6 , 7 ] . This enables GP to uncover hidden functional relationships in power system tasks when sufficient observations are collected. GP offers several key advantages that make it feasible for use in the analysis of critical infrastructure such as power systems, where safety is a priority. • Flexible expressivity. The nonparametric nature of GP allows it to model highly complex functions [ 8 ] , making it ideal for capturing the intricate dynamics of modern power systems, which may include complex devices with proprietary internal control structures, such as inverter-based resources (IBRs). • Well-calibrated uncertainty quantification. A key advantage of GP is its ability to provide closed-form uncertainty estimates, accounting for both data noise and discrepancies from new scenarios that differ from the training data distribution [ 8 ] . This is essential for use within safety-critical systems such as power grids. • Efficient with small datasets. By using Bayesian learning principles [ 7 ] and with only a few hyperparameters to optimize, GP models can generalize effectively even with limited data [ 9 ] , which is particularly beneficial for applications where training data is scarce. Additionally, GP training can be interlaced with active learning [ 10 ] to overcome data scarcity if generating data is expensive and time-consuming. Due to these advantages, GP has garnered increasing attention in the field of power systems in recent years, particularly for applications such as forecasting, risk analysis, and decision-making. This article presents the fundamentals of GP modeling and then explores its use cases in power grid analysis. To the best of our knowledge, this paper is the first to provide a comprehensive survey of GP applications in power systems. In addition, we discuss the key challenges and future directions for the use of GP in this field, as well as their integration with other data-driven methods. The main contributions of this paper are summarized as follows: • This paper presents a comprehensive and structured overview of the fundamental concepts of GP, along with an overview of the supporting software ecosystem. • This paper highlights key applications of GP in power systems, including load and renewable energy forecasting, power system static and dynamic modeling, risk assessment, and decision-making. The application procedures and prior work are discussed in detail, demonstrating how GP effectively addresses challenges in these domains. • This paper investigates the major challenges in applying GP to power systems, such as scalability and robustness, and discusses applications where GP may have greater applicability. Lessons learned are discussed and finally, future research directions, particularly in integrating GP with other advanced data-driven methods, are outlined. The remainder of this paper is organized as follows. Section II offers an overview of the fundamental principles and ecosystem of GP. Section III provides a comprehensive review of GP-based learning in power system applications: forecasting, static and dynamic estimation and solutions to optimization problems. Next, we show how such learned models can be used for static and dynamic risk assessment in power grids in Section IV , and inside optimization and control modules in Section V . Finally, Section VI discusses the challenges and proposes future research directions for the application of GP in power systems. Section VII concludes the paper. Fig. 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 IV and Table V for specific works related to specific problems. II Preliminaries of Gaussian Process II-A Gaussian Process Fundamentals This section presents the basic details of GP from a functional space perspective. GP is a non-parametric and probabilistic modeling method. Consider a latent function f(\cdot) that

Read the complete text