MUMBAI, India, Oct. 26 -- Intellectual Property India has published a patent application (202641123294 A) filed by Mahantesh P Mattada; Karuna Gangaram Choudhari; and Saba Fatima on October 14, 2026, for A Machine Learning Based System And Method For Electrical Fault Detection And Classification In Photovoltaic Panels.
Inventors include Karuna Gangaram Choudhari; and Saba Fatima.
The application for the patent was published on October 23, 2026, under issue no. 43/2026.
Abstract: A system and method are disclosed for machine learning based electrical fault detection and classification in photovoltaic panel arrays. Electrical parameters, including string current statistics from multiple ammeters, DC voltage, DC power, solar radiation and temperature, are acquired and normalized using a z-score standardization to reduce outlier influence and overfitting. The normalized feature vector is classified using a trained machine learning classifier, preferably a decision tree classifier configured with a minimum sample split of two and a Gini impurity criterion, to determine whether the array is in a normal condition or is affected by a string fault, a string-to-ground fault or a string-to-string fault. Experimentally, the preferred decision tree embodiment achieves an average recall of about 95.83%, precision of about 95%, F1-score of about 94.94% and overall accuracy of about 95%, outperforming k-nearest neighbors, support vector machine, naïve Bayes and random forest classifier embodiments. The invention enables reliable, computationally efficient, real-time fault monitoring for large photovoltaic installations, reducing dependence on manual inspection.
Disclaimer: Curated by HT Syndication.
