A Machine Learning-Based Early Landslide Warning System Using IoT

Abstract

Landslides are ubiquitous than any other geological event and can betide anywhere in the world, making its effect on human lives devastating. The development of a predictive model for landslides and their early warning in real-time based on machine learning and IoT techniques is delineated here. A predictive model was trained using data of various geotechnical parameters like soil moisture, shear strength of the soil, severity of the rain, the slope of the terrain, etc. The hardware consists of a set of sensors that obtains the required soil and terrain parameters in real-time. The model was validated using standard validation techniques, obtaining an accuracy of 98% and zero false negatives. This paper discusses the deployment and data acquisition from the geophysical sensors, the algorithms utilized by the predictive model, the communication between the models and the sensor modules.