Proactive Alerting Solution: An LSTM Model for Power Outage Forecasting in Chalki, Greece

In today's world, electrical power is essential for the majority of human activities. Power outages not only result in significant economic losses, but also pose risks to public health and safety. At present, public authorities frequently lack proactive tools for anticipating power outages and promptly notifying citizens who may be affected. This paper introduces a robust solution that predicts the likelihood of power outages within the subsequent 30-minutes window. Our system provides recommendations to public authorities on whether to issue an alert based on the predicted probability of an outage. The predictive analytics are powered by a Long Short-Term Memory (LSTM) model, which has been trained on actual energy consumption data from the island of Chalki, Greece. This solution has been designed and deployed in Chalki as part of the DECIDO project, demonstrating its practical application and utility in a real-world setting.