IoT-based Intelligent Power Supply Management Using Ensemble Learning for Seismic Observation Stations

Abstract

Seismic observation stations perform a vital part in monitoring and analyzing seismic activity for early warning and disaster preparedness. This paper investigates the integration of an IoT-based intelligent power supply management model to improve station reliability and effectiveness. Traditional systems often suffer from reliability issues and inadequate monitoring, impacting timely seismic data delivery during critical events. The study employs IoT sensors for real-time monitoring of voltage, current, battery status, and environmental conditions. Data are centralized for analysis, leveraging the SeismoGuard Ensemble classifier—a novel machine learning model combining Random Forest, SVM, and KNN models with a Logistic Regression meta-classifier. The novelty lies in its distinctive blend of Random Forest, SVM, KNN, and Logistic Regression improves predictive accuracy and robustness in power supply handling for seismic observation stations. This approach improves forecasting accuracy and robustness in preventing power failures, achieving high prediction measurements like accuracy (90%), precision (88%), recall (91%), and F1-score (89%). Implementation leads to enhanced data transmission throughput and packet delivery ratio, ensuring reduced downtime and increased resilience during seismic events. Integrating IoT technologies in power supply management offers substantial benefits, including enhanced reliability and operational continuity, vital for effective seismic monitoring and early warning systems.