Granular Edge Intelligence for Early Earthquake Detection Using Sand-Based Sensing and Hybrid Deep Learning with Human Impact Analysis

Earthquakes are among the most devastating natural disasters, causing widespread destruction to infrastructure and posing significant threats to human life. Early warning systems are essential to reduce fatalities and enable effective disaster management. Traditional seismic monitoring systems, while valuable, are often limited by their reliance on fixed sensor placements, high costs, and inability to integrate with real-time human identification in affected areas. Additionally, environmental factors such as heavy air pollution can impair visibility and facial recognition capabilities, hindering post-disaster rescue operations. Research presents an advanced Granular Edge Intelligence Framework that integrates sand-based geotechnical sensing, IoT-enabled real-time data acquisition, and hybrid deep learning models for early earthquake detection. The framework further incorporates facial recognition techniques to identify trapped or injured individuals within collapsed buildings, even under conditions of dense particulate matter. Utilizing CNN-LSTM architectures, the system processes multivariate sensor data and video streams Simultaneously, providing high-accuracy earthquake prediction and rapid human identification. Extensive simulations and experimental evaluations demonstrate 95.2% accuracy in earthquake detection and robust human identification performance under heavy pollution, indicating the framework’s applicability for deployment in urban disaster-prone environments