{"title":"Machine learning-based inertia estimation in power systems: a review of methods and challenges","authors":"Santosh Diggikar, Arunkumar Patil, Siddhant Satyapal Katkar, Kunal Samad","doi":"10.1186/s42162-025-00496-7","DOIUrl":null,"url":null,"abstract":"<div><p>The transformation of power systems is accelerating due to the widespread integration of renewable energy sources (RES) and the growing role of inverter-based generations (IBGs). This shift has significantly reduced rotational inertia, increasing the system’s vulnerability to frequency fluctuations during disturbances. Consequently, the accurate and adaptive estimation of inertia has become crucial for maintaining frequency stability and grid reliability. Traditional estimation methods, though effective in certain scenarios, struggle to capture the non-linear and dynamic behaviors of modern power systems, necessitating the adoption of advanced solutions. This review comprehensively explores machine learning (ML)-based methodologies for inertia estimation, emphasizing their adaptability, scalability, and real-time capabilities compared to conventional approaches. The study categorizes ML techniques into supervised learning (SL), unsupervised learning (USL), semi-supervised learning (SSL), and reinforcement learning (RL), highlighting their applications, advantages, and limitations. Advanced methodologies, such as hybrid and ensemble models, are examined for their effectiveness in overcoming challenges posed by noisy data, dynamic behaviors, and complex grid configurations. Some advanced techniques demonstrate proficiency in analyzing complex datasets and providing real-time insights into the evolving dynamics of inertia. In addition to evaluating existing approaches, the review identifies key research gaps and emerging trends, offering strategic guidance and important considerations for the development of innovative ML-driven inertia estimation methods. By addressing these challenges, this study aims to support the creation of adaptive and reliable tools that ensure effective grid management in an energy ecosystem increasingly dominated by RES. </p><h3>Graphical abstract</h3><div><figure><div><div><picture><source><img></source></picture></div></div></figure></div></div>","PeriodicalId":538,"journal":{"name":"Energy Informatics","volume":"8 1","pages":""},"PeriodicalIF":0.0000,"publicationDate":"2025-04-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://energyinformatics.springeropen.com/counter/pdf/10.1186/s42162-025-00496-7","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Energy Informatics","FirstCategoryId":"1085","ListUrlMain":"https://link.springer.com/article/10.1186/s42162-025-00496-7","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"Energy","Score":null,"Total":0}
引用次数: 0
Abstract
The transformation of power systems is accelerating due to the widespread integration of renewable energy sources (RES) and the growing role of inverter-based generations (IBGs). This shift has significantly reduced rotational inertia, increasing the system’s vulnerability to frequency fluctuations during disturbances. Consequently, the accurate and adaptive estimation of inertia has become crucial for maintaining frequency stability and grid reliability. Traditional estimation methods, though effective in certain scenarios, struggle to capture the non-linear and dynamic behaviors of modern power systems, necessitating the adoption of advanced solutions. This review comprehensively explores machine learning (ML)-based methodologies for inertia estimation, emphasizing their adaptability, scalability, and real-time capabilities compared to conventional approaches. The study categorizes ML techniques into supervised learning (SL), unsupervised learning (USL), semi-supervised learning (SSL), and reinforcement learning (RL), highlighting their applications, advantages, and limitations. Advanced methodologies, such as hybrid and ensemble models, are examined for their effectiveness in overcoming challenges posed by noisy data, dynamic behaviors, and complex grid configurations. Some advanced techniques demonstrate proficiency in analyzing complex datasets and providing real-time insights into the evolving dynamics of inertia. In addition to evaluating existing approaches, the review identifies key research gaps and emerging trends, offering strategic guidance and important considerations for the development of innovative ML-driven inertia estimation methods. By addressing these challenges, this study aims to support the creation of adaptive and reliable tools that ensure effective grid management in an energy ecosystem increasingly dominated by RES.