Abstract

Conventional energy generating strategies, such as reactive and scheduled maintenance, often lead to increased downtime, energy waste, and inefficiencies. This study integrates edge analytics with machine learning-based predictive maintenance to boost the reliability and sustainability of off-grid energy generating systems. Using Long Short-Term Memory and regression models, the approach enables early anomaly detection and fault prediction, reducing unplanned outages and maintenance costs. A comparative analysis between standard edge analysis and integrated edge-predictive methods shows that the integrated system achieves an accuracy of 91.6%, compared to the edge analytics model with an accuracy of 86.2% effectively stabilizing short-term fluctuations, generating fewer and more stable alerts, with a coefficient of determination R² of 0.98. Results highlight that combining predictive models with edge analytics enhances reliability, supports timely interventions, and strengthens system robustness in off-grid energy generating applications.