ПРИМЕНЕНИЕ ТЕЛЕМЕТРИИ И ЭКСПЛУАТАЦИОННОЙ АНАЛИТИКИ ДЛЯ ВЫЯВЛЕНИЯ АНОМАЛИЙ РАБОТЫ ОБОРУДОВАНИЯ НА ОБЪЕКТАХ ВОЗОБНОВЛЯЕМОЙ ЭНЕРГЕТИКИ

УДК 620.92:621.398

Abstract

The article examines the role of telemetry data and operational analytics in detecting equipment anomalies at renewable energy facilities. The aim of the study is to systematize and structure approaches to identifying abnormal operating conditions in solar and wind power plants based on monitoring data, diagnostic parameters, and analytical models. The research is based on a structured review of recent peer-reviewed scientific publications, industry reports, regulatory materials, and applied case studies related to condition monitoring, SCADA data analysis, predictive diagnostics, and intelligent maintenance in renewable energy systems.

The study classifies the main groups of anomalies, including operational, electrical, mechanical, thermal, and information-communication deviations, and identifies the telemetry and operational parameters used for their detection. Particular attention is paid to threshold-based methods, statistical models, machine learning approaches, deep learning algorithms, explainable artificial intelligence, and digital twin-based monitoring systems. It is concluded that telemetry should be considered primarily as a source of operational data, while the practical value of anomaly detection depends on data quality, model selection, interpretability, and integration of analytical results into asset management and maintenance processes.

Аннотация

В статье рассматривается роль телеметрических данных и эксплуатационной аналитики в выявлении аномалий работы оборудования на объектах возобновляемой энергетики. Цель исследования заключается в систематизации и структурировании подходов к выявлению аномальных режимов работы солнечных и ветровых электростанций на основе данных мониторинга, диагностических параметров и аналитических моделей. Исследование основано на структурированном обзоре современных рецензируемых научных публикаций, отраслевых отчетов, нормативных материалов и прикладных кейсов, связанных с мониторингом технического состояния, анализом SCADA-данных, предиктивной диагностикой и интеллектуальным техническим обслуживанием в системах возобновляемой энергетики.

В статье классифицированы основные группы аномалий, включая эксплуатационные, электрические, механические, тепловые и информационно-коммуникационные отклонения, а также определены телеметрические и эксплуатационные параметры, используемые для их выявления. Особое внимание уделено пороговым методам, статистическим моделям, алгоритмам машинного и глубокого обучения, объяснимому искусственному интеллекту и системам мониторинга на основе цифровых двойников. Сделан вывод о том, что телеметрию следует рассматривать прежде всего как источник эксплуатационных данных, тогда как практическая ценность выявления аномалий зависит от качества данных, выбора моделей, интерпретируемости и интеграции аналитических результатов в процессы управления активами и технического обслуживания.

Keywords: renewable energy, telemetry, operational analytics, equipment diagnostics, SCADA data, intelligent monitoring, predictive analytics.

Ключевые слова: возобновляемая энергетика, телеметрия, эксплуатационная аналитика, диагностика оборудования, SCADA-данные, интеллектуальный мониторинг, предиктивная аналитика.

Introduction

The expansion of renewable energy facilities has significantly changed the requirements for equipment monitoring, operational reliability, and data-based maintenance. Solar and wind power plants operate under variable environmental conditions, geographically distributed asset structures, and fluctuating generation regimes. As a result, the technical condition of individual components increasingly affects not only local equipment performance, but also the stability of power output, the efficiency of maintenance planning, and the reliability of renewable energy integration into power systems. Under these conditions, the detection of equipment anomalies becomes one of the key tasks of operational asset management.

The relevance of this problem is reinforced by the continued growth of renewable energy capacity and the increasing role of inverter-based resources in modern power systems. According to the U.S. Energy Information Administration, 53 GW of new capacity were added to the U.S. power grid in 2025, which became the largest annual addition since 2002. In 2026, capacity additions are projected to reach 86 GW, with solar energy, battery storage, and wind power accounting for 51%, 28%, and 14% of this growth, respectively. At the global level, the International Energy Agency also projects further expansion of renewable electricity capacity in 2025–2030, with solar photovoltaic and wind generation remaining the main drivers of this process (fig. 1).

Figure 1. Growth of renewable energy capacity by technology segment, baseline scenario, 2013–2030 [1].

At the same time, the North American Electric Reliability Corporation emphasizes that the growing share of inverter-based resources creates additional requirements for monitoring voltage control, reactive power limits, frequency ride-through, protection settings, and active power recovery [2]. Therefore, equipment anomalies at renewable energy facilities should be considered not only as isolated technical failures, but also as deviations related to measurement quality, control logic, protection behavior, communication stability, and changing operating regimes.

From an engineering perspective, anomaly detection at renewable energy facilities is a multiparametric task. Solar power plants require continuous monitoring of power output, irradiance, inverter status, string-level parameters, module temperature, connector condition, and thermal deviations [3]. Wind power plants require the analysis of SCADA data, vibration signals, generator and gearbox parameters, pitch and yaw system behavior, rotor speed, component temperature, and meteorological conditions. In both cases, telemetry data serve as the primary source of operational information, while the diagnostic value of this information depends on the quality of data structuring, preprocessing, interpretation, and integration with analytical models.

Recent scientific studies confirm the importance of operational data analytics for early anomaly detection in renewable energy systems. Nogueira et al. used SCADA data from wind turbines and proposed an approach combining an autoencoder-based neural network with Failure Mode and Symptoms Analysis. The model achieved high classification accuracy on simulated data and made it possible to identify abnormal equipment behavior before recorded failures [4]. Miraftabzadeh et al. demonstrated the applicability of an unsupervised autoencoder model for detecting anomalies in photovoltaic power time series under conditions of incomplete information. Their results showed that deep learning models can support the identification of hidden deviations in photovoltaic plant operation when labeled fault data are limited or unavailable [5].

At the same time, existing studies often focus on individual algorithms, specific components, or separate fault types. However, practical intelligent monitoring at renewable energy facilities requires a broader systematization of anomaly types, diagnostic parameters, operational data sources, analytical models, and applicability criteria. This creates the need for a structured approach that links engineering failure characteristics with data availability, model selection, interpretability, and operational decision support.

The aim of this study is to systematize and structure information on the detection of equipment operation anomalies at renewable energy facilities based on telemetry data, diagnostic parameters, and operational analytics models. To achieve this aim, the study addresses the following objectives: to identify the main operational and technological prerequisites for anomaly monitoring in solar and wind power plants; to classify the principal types of equipment anomalies and the diagnostic parameters used for their detection; to compare the main classes of analytical models applied to telemetry and SCADA data streams; to determine the limitations, performance criteria, and practical requirements for intelligent monitoring systems; and to integrate the obtained results into a generalized framework that links anomaly types, diagnostic parameters, analytical models, and operational decision support.

Materials and methods

The study was conducted as a structured analytical review of scientific and applied sources on anomaly detection, condition monitoring, SCADA data analysis, predictive diagnostics, and intelligent maintenance at renewable energy facilities. The source base included peer-reviewed publications from the last five years, industry and regulatory reports, and applied cases on solar, wind, battery storage, and hybrid renewable assets. Priority was given to sources addressing equipment operating parameters, anomaly types, telemetry and SCADA data structures, analytical models, diagnostic indicators, and implementation constraints.

The reviewed materials were analyzed using comparative, classification, and structural synthesis methods. Classification was carried out according to three main criteria: the engineering nature of the anomaly, the type of diagnostic parameter used for its detection, and the analytical approach applied to operational data processing. Based on these criteria, equipment deviations were grouped into operational, electrical, mechanical, thermal, and information-communication anomalies. Analytical models were grouped according to their detection logic, including threshold-based methods, statistical models, clustering and preprocessing approaches, machine learning and deep learning models, explainable artificial intelligence, and digital twin-based monitoring systems.

The logic of structuring the results in tables was determined by the sequence of the research objectives. Table 1 systematizes anomaly types and informative diagnostic parameters. Table 2 compares analytical models used for processing telemetry and SCADA data streams. Table 3 summarizes implementation limitations, diagnostic and operational performance criteria, and prospects for intelligent monitoring. Table 4 combines these results into a generalized monitoring framework that links anomaly types, diagnostic parameters, analytical models, and decision-support functions. This structure makes it possible to move from a descriptive overview of individual studies to a systematic representation of anomaly detection as an operational analytics process in renewable energy asset management.

Results and Discussion

Operational prerequisites for anomaly monitoring at renewable energy facilities

The expansion of renewable energy facilities reflects a transition toward a more distributed, weather-dependent, and dynamic generation structure. In such systems, output reliability increasingly depends on the quality of equipment condition monitoring and the speed of detecting deviations in operating parameters. Fluctuations in irradiance, wind speed, temperature, and grid conditions directly affect the load on inverters, converter units, power electronics, switching components, storage systems, and auxiliary subsystems [6, 7].

In this study, telemetry is understood as a source of continuously collected operational data that describe equipment condition, operating regimes, and environmental parameters. The interpretation of these data is performed through operational analytics, which relates changes in monitored parameters to possible signs of degradation, abnormal behavior, or reduced equipment performance. Therefore, anomaly detection at renewable energy facilities includes not only the identification of equipment failures, but also the distinction between physical deviations, measurement errors, communication failures, incorrect protection settings, and changes in operating regimes [8, 9].

At the component level, the importance of early anomaly detection is associated with the technical and economic consequences of late fault identification. Recent studies on photovoltaic and wind energy systems show that fault diagnosis may rely on inverter data, SCADA measurements, temperature indicators, vibration signals, image-based inspection, and machine learning models, depending on the type of asset and the monitored component [10, 11]. This demonstrates that even local equipment defects can produce measurable operational and economic losses if they are not detected in time.

Thus, the operation of renewable energy facilities is shaped by high weather variability, increasing asset complexity due to power electronics and storage, and the rising cost of late anomaly detection. Under these conditions, telemetry should be viewed as a structured source of operational data, while the central analytical role belongs to operational analytics and intelligent monitoring systems that link diagnostic parameters with signs of abnormal behavior and support proactive maintenance decisions.

Typology of equipment operation anomalies and parameters for their detection

The development of an anomaly monitoring system at renewable energy facilities requires not only the accumulation of telemetry data, but also their structuring according to the types of deviations. From a theoretical perspective, an anomaly is understood as a change in equipment operating conditions that deviates statistically, physically, or functionally from the expected behavior of the system [12, 13].

For solar and wind power plants, such deviations are heterogeneous and require the use of different categories of diagnostic data. A 2025 study on photovoltaic system reliability shows that anomaly analysis may rely on production data, I–V characteristics, meteorological measurements, and visual or infrared inspection data [14]. This confirms that a single operating signal is usually insufficient for reliable diagnosis, while anomaly detection requires the combined analysis of several technical and environmental parameters.

In the most general sense, anomalies at renewable energy facilities can be divided into five groups: operational, electrical, mechanical, thermal, and information-communication anomalies (table 1).

Table 1. Classification of anomaly types and informative parameters for their detection at renewable energy facilities

In wind energy, SCADA data preprocessing is crucial for the reliability of condition monitoring and fault detection. A 2025 study on cluster-based filtering showed that the treatment of negative power values and the choice of outlier rejection thresholds directly affect the reliability of early generator fault detection [15]. This is also consistent with recent findings on SCADA data imbalance, which show that data structure and class distribution can influence the effectiveness of condition monitoring, diagnosis, and prognosis in wind turbines [16]. Therefore, an informative parameter should not be interpreted in isolation, but as part of a coordinated feature set whose diagnostic value depends on operating context, data quality, and model architecture.

Thus, anomaly typology provides a basis for moving from separate signal monitoring to the integrated analysis of operational, electrical, mechanical, thermal, and information-related indicators. This creates the methodological transition to the next stage of the study, where structured telemetry and SCADA parameters are linked to analytical models for anomaly detection, classification, and interpretation.

Analytical models and practices for anomaly detection in telemetry data streams

Anomaly detection at renewable energy facilities involves a range of analytical approaches that differ in data type, detection logic, and suitability for real operating conditions. Telemetry and SCADA data are now analyzed not only for fault detection, but also for fault diagnosis and prognosis, making anomaly monitoring part of a broader operational framework where interpretability, noise tolerance, early sensitivity, and scalability matter alongside accuracy. Recent studies show that AI-based condition monitoring in wind energy includes multiple diagnostic tasks, while deep learning models are increasingly adapted to extract fault-related features from noisy and heterogeneous operating data [17, 18]. The main classes of models used for processing telemetry and SCADA data streams at renewable energy facilities are summarized in table 2.

Table 2. Main analytical models for anomaly detection in telemetry data from renewable energy facilities

In applied corporate practice, the significance of analytical models is confirmed by their use in predictive equipment monitoring systems. For example, GE Vernova reported a 2025 case in which a predictive analytics system detected an abnormal trend in a turbine subsystem before serious damage occurred. After verification and component replacement, the deviation disappeared, while avoided losses and downtime-related costs were estimated at more than $500,000 [19]. This example shows that analytical models can support early detection of abnormal conditions, reduce the risk of unplanned outages, and guide proactive maintenance decisions.

A similar logic is observed in energy storage systems operating alongside renewable energy assets. Fluence reported that its State of Balance algorithm had been deployed across about 80% of its LFP systems, representing more than 10 GWh, and processed billions of cell data points each month [20]. At one 60 MW ERCOT site, charge imbalance analytics and rolling recalibration recovered 25% of stranded capacity within eight weeks; across the broader fleet, the company reported at least a 50% reduction in stranded energy within three months. These cases indicate that analytical models are used not only to detect anomalies, but also to improve asset availability, operational efficiency, and economic performance.

Thus, anomaly detection in renewable energy telemetry and SCADA data streams is evolving from isolated signal control to integrated operational analytics. The practical value of these models depends on their ability to identify abnormal patterns, improve observability, support timely maintenance actions, and operate within broader digital management frameworks.

Limitations, performance criteria, and prospects for implementing intelligent monitoring at renewable energy facilities

The implementation of intelligent monitoring at renewable energy facilities depends not only on telemetry data processing algorithms, but also on operational, organizational, and infrastructural constraints. Its effectiveness is determined by detection accuracy, data completeness, equipment heterogeneity, variable generation conditions, and integration with dispatch and maintenance systems. Therefore, intelligent monitoring should be assessed through implementation limitations, performance criteria, and development prospects within the operational reliability architecture of renewable energy assets (table 3).

Table 3. Limitations, performance criteria, and prospects for implementing intelligent monitoring at renewable energy facilities

Thus, the effectiveness of intelligent monitoring at renewable energy facilities should be assessed not only by model performance, but also by its applicability in real operating conditions. Practical effectiveness depends on telemetry data quality, interpretability of analytical outputs, robustness to heterogeneous regimes, and integration with maintenance and dispatch processes [21, 22]. Further development of this field is associated with a more mature digital infrastructure in which operational data, analytical models, and decision-support systems are combined to improve reliability, reduce generation losses, and support stable renewable asset operation.

These conditions also indicate the need to summarize the results of the classification and comparative analysis in an integrated framework. Such a framework links anomaly types, diagnostic parameters, analytical models, and decision-support functions, making it possible to present anomaly detection as a coordinated monitoring system rather than as a set of isolated diagnostic procedures (table 4).

Table 4. Integrated framework for anomaly detection and monitoring at renewable energy facilities

The integrated framework presented in table 4 shows that the effectiveness of anomaly detection depends on the alignment between the type of deviation, the available diagnostic parameters, and the analytical model used for interpretation. In this context, telemetry should be regarded primarily as a source of structured operational data, while the main analytical value is created through classification, model selection, and the integration of diagnostic results into maintenance, dispatch, and asset management processes. Therefore, intelligent monitoring at renewable energy facilities should be designed as a comprehensive decision-support system rather than as a narrow tool for recording abnormal signals.

Conclusion

The study showed that telemetry data constitute an important source of operational information for anomaly detection and reliability management at renewable energy facilities, while operational analytics provides the methods for processing, interpreting, and using these data in anomaly detection. It was established that anomalies at such facilities are multidimensional in nature and include operational, electrical, mechanical, thermal, and information-communication deviations. Such an analysis should include an integrated approach towards assessing structured telemetry and SCADA data through approaches beyond the use of thresholds, such as statistical techniques, machine learning, deep learning, and explainable artificial intelligence for identifying, classifying, and interpreting anomalies.

The key advantage of implementing such solutions would be that they help mitigate the occurrence of unplanned downtime, minimize the generation losses associated with this phenomenon, and increase the reliability of renewable energy sources. However, the successful integration of intelligent monitoring is contingent upon having high-quality data, explainable models, and advanced digital infrastructures. The future development of this field is associated with the standardization of telemetry, the advancement of digital twins, and the deeper integration of analytical models into maintenance and operational management processes.

References:

1. International Energy Agency. Renewable electricity [Electronic resource] – URL: https://www.iea.org/reports/renewables-2025/renewable-electricity (accessed 01.05.2026).
2. North American Electric Reliability Corporation. Essential Actions to Industry [Electronic resource] – URL: https://www.nerc.com/globalassets/programs/bpsa/alerts/2023/level-3-alert-essential-actions-ibr-performance-and-modeling.pdf (accessed 01.05.2026).
3. Baltacı Ö., Kıncay Z., Dalkılınç K., Karaman O. Thermal Image and Inverter Data Analysis for Fault Detection and Diagnosis of PV Systems // Applied Sciences. 2024. Vol. 14. No. 9. Art. 3671. DOI: 10.3390/app14093671.
4. Nogueira W.F., Melani A.H.A., de Souza G.F.M. Wind turbine fault detection through autoencoder-based neural network and FMSA // Sensors. 2025. Vol. 25. No. 14. Art. 4499. DOI: 10.3390/s25144499.
5. Miraftabzadeh S.M., Longo M., Leva S., Matera N. Data anomaly detection in photovoltaic power time-series via unsupervised deep learning with insufficient information // Sustainable Energy, Grids and Networks. 2025. Vol. 43. Art. 101769. DOI: 10.1016/j.segan.2025.101769.
6. Ejuh Che E., Roland Abeng K., Iweh C.D., Tsekouras G.J., Fopah-Lele A. The impact of integrating variable renewable energy sources into grid-connected power systems: Challenges, mitigation strategies, and prospects // Energies. 2025. Vol. 18. No. 3. Art. 689. DOI: 10.3390/en18030689.
7. Amiri A.F., Oudira H., Alshenaifi R., Alshammari A., Alenezi M. Fault Detection and Diagnosis of a Photovoltaic System Based on Deep Learning Using the Combined InceptionV3–XGBoost Model // Sustainability. 2024. Vol. 16. No. 3. Art. 1012. DOI: 10.3390/su16031012.
8. Dao P.B., Barszcz T., Staszewski W.J. Anomaly detection of wind turbines based on stationarity analysis of SCADA data // Renewable Energy. 2024. Vol. 232. Art. 121076. DOI: 10.1016/j.renene.2024.121076.
9. Fazli A., Poshtan J. Wind turbine fault detection and isolation robust against data imbalance using KNN // Energy Science & Engineering. 2024. Vol. 12. No. 3. P. 1174–1186. DOI: 10.1002/ese3.1706.
10. Bougoffa M., Benmoussa S., Djeziri M., Palais O. Hybrid Deep Learning for Fault Diagnosis in Photovoltaic Systems // Machines. 2025. Vol. 13. No. 5. Art. 378. DOI: 10.3390/machines13050378.
11. Alagha N., Othman M.L., Hannan M.A., Ker P.J., Hussain A., Mansor M. Artificial intelligence in wind turbine fault detection and diagnosis: A systematic review // Energies. 2025. Vol. 18. No. 7. Art. 1680. DOI: 10.3390/en18071680.
12. Liu D., Wang S., Li Y., Du J., Li J. Research on new energy power plant network traffic anomaly detection method based on EMD // Energy Informatics. 2025. Vol. 8. No. 1. Art. 14. DOI: 10.1186/s42162-025-00474-z.
13. Campoverde-Vilela L., Vidal Y., Acho L. Anomaly-based fault detection in wind turbine main bearings // Wind Energy Science. 2023. Vol. 8. No. 4. P. 557–574. DOI: 10.5194/wes-8-557-2023.
14. Chen X., Li B., Braid J.L., Byford B., Colvin D.J., Glaws A., Jost N., Pierce B., Rabade S., Springer M., Jain A. Open data sets for assessing photovoltaic system reliability // Applied Energy. 2025. Vol. 395. Art. 126132. DOI: 10.1016/j.apenergy.2025.126132.
15. Kijanowski K., Barszcz T., Dao P.B. A Cluster-Based Filtering Approach to SCADA Data Preprocessing for Wind Turbine Condition Monitoring and Fault Detection // Energies. 2025. Vol. 18. No. 22. P. 5954. DOI: 10.3390/en18225954.
16. Oliveira-Filho A., Comeau M., Cave J., Nasr C., Côté P., Tahan A. Wind Turbine SCADA Data Imbalance: A Review of Its Impact on Health Condition Analyses and Mitigation Strategies // Energies. 2025. Vol. 18. No. 1. Art. 59. DOI: 10.3390/en18010059.
17. Sasinthiran A., Gnanasekaran S., Ragala R. A review of artificial intelligence applications in wind turbine health monitoring // International Journal of Sustainable Energy. 2024. Vol. 43. No. 1. P. 1–35. DOI: 10.1080/14786451.2024.2326296.
18. Chen Q., Zhao D., Li M. Bearing fault diagnosis based on efficient cross space multiscale CNN transformer parallelism // Scientific Reports. 2025. Vol. 15. Art. 12197. DOI: 10.1038/s41598-025-95895-x.
19. GE Vernova. $500K+ in damage and downtime avoided by catching inlet filter obstruction [Electronic resource] – URL: https://www.gevernova.com/software/blog/predictive-analytics-prevents-turbine-damage-downtime (accessed 03.05.2026).
20. Fluence. Beyond 20 Years: Maximizing Battery Storage Lifespan and Value [Electronic resource] – URL: https://blog.fluenceenergy.com/maximizing-battery-storage-lifespan-value (accessed 03.05.2026).
21. Snytko A., Kroposki B., Hoke A., Palmintier B. Fault diagnosis for photovoltaic systems: A validated industrial SCADA framework // Applied Sciences. 2025. Vol. 15. No. 23. Art. 12656. DOI: 10.3390/app152312656.
22. Shadi M.R., Mirshekali H., Shaker H.R. Explainable artificial intelligence for energy systems maintenance: A review on concepts, current techniques, challenges, and prospects // Renewable and Sustainable Energy Reviews. 2025. Vol. 216. Art. 115668. DOI: 10.1016/j.rser.2025.115668.