IoT based real-time monitoring system of rainfall and water level for flood prediction using LSTM network

Abstract

This project outlines the design of a flood monitoring system to obtain accurate data on river overflows. Additionally, it provides the machine learning technique, to predict the arrival of floods, by considering the rainfall data and water level from previously available data to predict the rainfall and water level for the next hours. The problem is the shortage of flood information in areas that are constantly flooded leads to malfunction in analysing the flood reasons. In addition, the fuzzy and unpredicted situation of the flood. Moreover, there is no flood data analysis so action can be taken based. Finally, data is not visualized in a Dashboard, so they can have a deeper look at the situation. The object of this study is to design an IoT flood monitoring system based on two water level sensors and a rain gauge sensor. In addition, to forecast the flood based on Long Short-Term Memory (LSTM) networks for historical data and the data collected from the monitoring system. The monitoring system utilizes a submersible water level sensor that measures the water level. Additionally, the tipping bucket rainfall sensor measures the rain gauge and tests the rainfall in the natural environment. The system is based on IoT to provide real-time data. The recorded data is transmitted to the cloud via a GSM network and displayed on an online platform. The flood forecasting model used Long Short-Term Memory (LSTM) networks to predict future floods. The aim of this case study is to contribute to the reduction of casualties and flood damage in streams, as well as to the development of more accurate flood forecasting in typical urban flood risk locations. The result was experimented with using historical data since the current data is insufficient yet to make an accurate prediction. The main findings of the research are the predicted values of streamflow and rainfall for historical data, also water level and rain gauge for new data. The forecasting method that applied LSTM showed high accuracy of the result reaching more than 90% with evaluation errors for historical data MAE, RMSE and MSE are 0.93, 1.7 and 3.025 respectively. Also, 0.0055, 0.3325 and 0.1175 for new data respectively. The developed monitoring system and flood forecasting can be used efficiently as a non-structural solution to alleviate the damage caused by urban floods.