ESP32-Based Flood Monitoring and Early Warning System Using the Sugeno Fuzzy Logic Method

Abstract

Floods remain one of the most frequent natural disasters in Indonesia, highlighting the need for reliable real-time monitoring and early warning systems. This study aims to develop an ESP32-based flood monitoring and early warning system integrated with a web-based monitoring platform using the Sugeno Fuzzy Logic method. The proposed system employs three environmental parameters, namely water level measured by the JSN-SR04T ultrasonic sensor, rainfall intensity measured by a tipping bucket rain gauge, and soil moisture measured by a capacitive soil moisture sensor. Sensor data are processed using the Sugeno Fuzzy Logic algorithm to classify flood conditions into Safe, Alert, and Danger, while the monitoring results are transmitted to Firebase Realtime Database for real-time website visualization and notification delivery through Telegram and Webpushr. Experimental results show that the ultrasonic, soil moisture, and tipping bucket rain gauge sensors achieved average accuracies of 99.35%, 98.01%, and 94.64%, respectively. The Sugeno implementation demonstrated excellent agreement with manual calculations, achieving an MSE of 0.00069, RMSE of 0.02627, and R² of 0.999998, while the flood status classification achieved 100% accuracy, precision, recall, and F1-score. In addition, the average notification response times were 1.63 s for Telegram and 2.35 s for Webpushr. These results demonstrate that the proposed system provides accurate flood monitoring, reliable flood status classification, and timely warning dissemination, making it suitable for real-time flood early warning applications.