Iot Based Flood Detection, Alarm and Monitoring System Using Multilayer Perceptron and Regression

Abstract

Flooding is a common natural event caused by heavy rainfall and high tides in the Philippines. While this disaster cannot be prevented, people can prepare themselves to face it. The City of Ilagan in the province of Isabela is one of the highly prone areas to flooding caused by the welling of the Cagayan River. Many communities are living in low-lying areas which most likely experiencing floods. An analysis of the location is conducted including the people living in the area. The analysis resulted in the development of an IoT-based technology for detection and early warning signals to people using sms notifications that can help lessen the difficulty of evacuation. The system uses Arduino UNO as a microcontroller where sensors are attached. These sensors are Light Detection and Ranging (LiDAR) for flood level measurement in feet (ft), Rain Gauge to measure the precipitation rate (mm/hr), and Flow Rate Meter to measure the fluctuation flow of the river in (L/hr). Data gathered from these sensors are processed and sent immediately to people living nearby to monitor the flood level in real-time. The predictive models are developed using the Pinacanauan River dataset taken from the river stations. The multilayer perceptron is used to develop the predictive model with 99% accuracy. The data from sensors are used also and processed using linear regression and calculated as 88% accurate and significant for prediction.