Smart flood early warning system with AI prediction and LoRa communication (FloodGuard AI)

Mohd Bakhtiar, Muhamad Akmal Harith and Sarnin, Suzi Seroja and Saipul, Nur Syahirah and Ishak, Norlela and Naim, Nani Fadzlina and Dohad, Mohd Rizal and S Abdul Kadir, Ros Shilawani (2026) Smart flood early warning system with AI prediction and LoRa communication (FloodGuard AI). Journal of Electrical and Electronic Systems Research (JEESR), 29 (1): 27. pp. 238-244. ISSN e-ISSN : 3030-640X
Identification Number (DOI): 10.24191/jeesr.v29i1.027
Abstract

FloodGuard AI is a prototype flood monitoring and risk-classification platform integrating an HC-SR04 ultrasonic water-level sensor, YL-83 rain sensor, Random Forest classifier, SX1262 LoRa communication, solar-battery power, and ESP32-CAM visual monitoring. Unlike a stand-alone threshold alarm, the prototype combines sensor trends with three-class risk inference and returns the classification to a local LED/siren node via a LoRa acknowledgment. Evaluation comprised controlled calibration, indoor link verification, and monitoring records collected between 30 January and 24 June 2026. The water-level sensor achieved 1.6 cm mean absolute error in 30 controlled measurements. The classifier attained 99.0% accuracy on a 200-sample held-out set and 92.9% on 420 valid monitoring records. Indoor bidirectional LoRa testing delivered 93 acknowledgments for 93 packets (100% packet delivery); outdoor 1 km performance was not validated with distance-specific packet delivery measurements. These results demonstrate prototype feasibility, while flood-event validation, independent risk labels, and outdoor link characterization remain necessary before operational deployment.

Item Details
Edit Item
Edit Item
Downloads & Files
[thumbnail of 149177.pdf]
Text
149177.pdf
Download (1MB)
Indexing & Metrics