Date of Publication

3-19-2025

Document Type

Master's Thesis

Degree Name

Master of Science in Electronics and Communications Engineering

Subject Categories

Electrical and Computer Engineering

College

Gokongwei College of Engineering

Department/Unit

Electronics And Communications Engg

Thesis Advisor

Edwin Sybingco

Defense Panel Chair

Maria Antonette C. Roque

Defense Panel Member

Leonard U. Ambata
Alvin Y. Chua

Abstract (English)

This thesis aims to provide a flood prediction system in the Philippines to increase flood awareness, which may help reduce property damage and save lives. Real-time flood status can significantly increase community awareness and preparedness by developing a secured IoT-based flood monitoring and prediction system using a Genetic Algorithm-based Neuro-Fuzzy network.

A flood model will simulate the flood level of a community. Moreover, the researcher developed a flood monitoring system with secured data flow from the sensor to the cloud and then to the end user. The algorithms embedded in the flood predicting model include Fuzzy Logic, LSTM neural network, and Genetic Algorithm. On the other hand, the microcontroller and sensors used in the project are Infineon CY8CPROTO-062-4343W PSoC™ 6 Wi-Fi BT Prototyping Kit, Infineon DPS310, rain sensor circuit, and DHT22 Precision Humidity sensor. In addition, the project uses the Infineon Shield2Go Cloud Security OPTIGA™ Trust M module to create a secure connection from the setup to Amazon Web Services (AWS). All data transmitted are encrypted when sending it to AWS IoT Core, Timestream, and Grafana.

After training and testing one month of data, the Neuro-Fuzzy LSTM Network with Genetic Algorithm solution has improved flood prediction accuracy by 47.4% compared to the 1000 Hidden Layered Stochastic Gradient Descent with Momentum (SGDM) solver that predicts every 24 hours trained with a 0.01 initial learning rate for 1000 epochs. The best solution suggests forecasting flood every 2 hours using an Adaptive Moment Estimation (ADAM) solver with 216 LSTM Hidden Layers trained with a 0.02 initial learning rate for 150 epochs.

Abstract Format

html

Abstract (Filipino)

None

Abstract Format

html

Language

English

Format

Electronic

Keywords

Flood forecasting

Upload Full Text

wf_yes

Embargo Period

3-19-2026

Download

Share

COinS