An effective real-time flood disaster management with automated GSM warning system

Abstract

Flood is one most devastating calamity of Bangladesh. A project is introduced that emphasizes management strategies throughout the crisis and issues alerts before and after a flood to reduce any concerns. To our study, flash flood is identified as a major concern of disaster. The flood warning system is triggered by two systems: sensors based and machine learning. Sensors with send signal to the main processing unit; Arduino. The Six distinct machine learning algorithms were developed in this study to estimate the river’s water level on a daily basis using data that was gathered from 2017 to 2021 and utilized to train and test the proposed model. To determine the model’s correctness, various design strategies were investigated. The method that produced the best prediction result had the lowest error rate. Machine leaning is taking place in a slave processing unit; Raspberry Pi. Both processing unit before and after flood can communicate via GSM module. Voice over is a new termed established in this project, where authority can directly connect and send announcement to the people of affected area.