Solar-Powered IoT-Based River Flood Monitoring and Prediction Using Random Forest Regression

Abstract

The design and implementation details of the solar power-based IoT system for river flood monitoring and prediction for water level measurement are provided in this paper. The main objective of the system is to measure the river water level in real-time and perform predictive analysis to reduce flood risk. The system predicts the water level 15 minutes in advance using a sliding window-based Random Forest regression model. A mechanical float and a timing pulley system, along with an incremental rotary encoder, are used for precise water level displacement measurement. The pulses from the incremental rotary encoder represent the pulley movement, and are processed by the ESP32 microcontroller to measure the water level by considering the pulley circumference and pulses per revolution. The information gathered by the system is transmitted to the Firebase Realtime Database using a 4G communication module for remote monitoring using a web-based application. A solar power system with a battery backup is used for power supply to ensure uninterrupted operation in adverse weather conditions. A vertical enclosure is provided for the mechanical system to increase system durability by protecting the system from possible damage. The system design is cost-effective, viable, and reliable for use in flood-prone and remote areas.