AI-Enabled Energy Forecasting and Fault Detection in Off-Grid Solar Networks for Rural Electrification
Year: 2025
Type: preprint
Abstract: Rural electrification remains a significant challenge globally, particularly in remote areas where access to centralized grids is limited. Off-grid solar systems have emerged as a viable solution to provide clean and sustainable energy to underserved communities. However, ensuring the efficient management and fault detection of these systems is crucial for their long-term success. This paper proposes an AI-enabled approach that combines energy forecasting and fault detection for off-grid solar networks. The proposed solution uses machine learning algorithms to predict solar energy generation and detect system faults, improving the reliability and performance of these systems. Energy forecasting is achieved through supervised learning models trained on historical data, including weather patterns, time of day, and energy production. Fault detection uses unsupervised learning models to identify anomalies in real-time data such as voltage and current levels. By integrating these two AI models, off-grid solar networks can optimize energy storage, improve system efficiency, and reduce downtime. The evaluation of this AI-enabled system shows promising results, with accurate energy predictions and effective fault detection. The system demonstrated a high true positive rate for fault identification and reduced prediction errors, leading to better energy management and consistent power supply. This approach can significantly enhance the sustainability and performance of off-grid solar networks, contributing to global rural electrification efforts and promoting clean energy adoption.
Author: Fozlur Rayhan
Institution: Lamar University
Language: English
Field: Energy
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W4413680989