Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{201203,
        author = {Shreyash Singh and Nithya S and Abishek KK and Praveenkumar S},
        title = {THEMIS: AN EXPLAINABLE MACHINE LEARNING FRAMEWORK FOR SOLARVOLTAIC DEGRADATION PREDICTION AND SUSTAINABLE ENERGY ANALYTICS},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {12},
        pages = {2891-2899},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=201203},
        abstract = {Solar photovoltaic (PV) systems are a key element in addressing the increasing global need for renewable energy in line with Sustainable Development Goal 7 (affordable and clean energy). While being installed at a fast pace, degradation over time is a key issue affecting energy yield, cost of investment and system performance. The performance of solar panels deteriorates over time due to environmental and operational stresses like thermal cycling, humidity, dust, moisture, and electrical overstress. Conventional monitoring solutions offer real-time dashboard and threshold-based notifications, but lack predictive capabilities to understand long-term performance trends. In this paper, we present THEMIS, an explainable machine learning model to predict degradation and estimate performance efficiency of solar panels, using structured environmental and operating data. We built a dataset of 10,000 instances with features for irradiance, temperature, humidity, dust index, voltage, current, power and efficiency. We applied GridSearchCV for hyper- parameter tuning, and shuffled-target testing for evaluation. Feature group importance was assessed through an ablation study, and SHAP explainability pinpointed the root cause of degradation. Our findings indicate that an optimized XGBoost model achieves better accuracy, stability and performance than baselines. THEMIS offers a transparent and validated approach for predictive maintenance, energy yield optimisation and sustainable management of solar parks.},
        keywords = {Sustainable Energy, Solar Energy, PV Degra- dation, Machine Learning, XGBoost, Explainable AI, SHAP, Predictive Maintenance, Sustainable Energy Analytics},
        month = {May},
        }