A CLOUD-BASED GRID COLLAPSE PREDICTION AND DISASTER RECOVERY MANAGEMENT SYSTEM: A CASE OF THE NIGERIAN POWER SECTOR
Abstract
Nigeria’s power grid experiences recurring partial and total collapses that cause widespread electricity outages and socioeconomic disruption. This study proposes a cloud-based system that combines grid-collapse prediction with coordinated disaster-recovery management. The predictive component uses a hybrid Convolutional Neural Network–Long Short-Term Memory model. The CNN identifies local patterns within multivariate grid data, while the LSTM captures temporal relationships that may indicate developing instability. Disaster recovery is organised through a three-horizon “Now, Next, and Later” framework. “Now” focuses on immediate stabilisation, situational awareness, communication, incident coordination, and restoration of critical loads. “Next” covers staged restoration, cause assessment, resource deployment, and a return to stable operation. “Later” addresses predictive maintenance, infrastructure reinforcement, preventive investment, model improvement, and organisational learning. It is designed to complement rather than replace existing protection and control-room systems. The framework was evaluated using simulated Nigerian power grid data. The CNN-LSTM achieved 95.5% accuracy, 96.0% precision, 95.0% recall, an F1-score of 95.0%, and an ROC-AUC of 0.982. Median inference latency was 184 milliseconds, and the false-alarm rate was 2.4%. The recovery module correctly classified 98.6% of 500 simulated actions into Now, Next, and Later. A regional service-failure exercise restored essential service in 7 minutes 42 seconds with a 46-second recovery point. These results show that the prototype provides a scalable, economically viable solution to enhance grid resilience in developing power systems with limited resources and infrastructure constraints. The framework also shows how artificial intelligence, cloud computing, and structured disaster-recovery planning can be integrated into a unified decision-support approach for improving Nigerian grid resilience.
