Flood Prediction and Alert System using ML and Sensor Networks

Floods and Landslides are the most common disasters that affect the people who dwell across the rivers and hilly areas. The impact of such instances decimates the normal lives of people who are made to either displace from the proximity or permanently relocate themselves to some other regions which in turn collapses the house hold and the country’s economy. An early prediction and detection system can act as a warning system to make the concerned authority and the people to prepare themselves in advance to counter the forthcoming devastation. This paper aims at incorporating the environmental based sensors and the Machine Learning algorithms to predict the possible flooding and raise an alert alarm in advance. The system is developed using several data points to predict the possibility of flooding. The parameters to be included to feed, as input, to the machine learning algorithm include the meteorological rainfall data of past years which acts as an empirical data, past data of the flooding which acts as a forehand information of flooding, the forecasted rainfall data of the present year and the actual rainfall data of the present year and the monthly average rainfall. Feeding all these data into the ML system will enable the system to get trained. The system is then capable to analyse and predictions are made. This system to predict in real time demands an array of machine learning algorithms. In this system, we have use an Arduino Uno to design the sensor system which includes sensors like ultrasonic sensor for measuring water levels, a float sensor to detect the level of water in the body, a flow sensor to determine the speed of water, and a humidity sensor to determine the humidity. With the aid of an IOT system, these sensor combinations are used to anticipate floods and warn the appropriate authorities.