Smart Fire Detection and Alert System

Ang, Jun Xuan (2026) Smart Fire Detection and Alert System. Final Year Project (Diploma), Tunku Abdul Rahman University of Management and Technology.

Text Ang Jun Xuan_Thesis.pdf Restricted to Registered users only Download (10MB)

Abstract

Although there are many technologies in this modern era, the fire detection and alert systems also may suffer from false alarms or delayed responses. Fire alarm systems are a critical line of emergency avoidance, 3986 fire detection systems were quality tested and it was determined that these detectors had a failure rate of 0.32%. The thesis introduces an Iot-based fire detection system which uses multiple sensors for its operation and performance testing. Fire detection systems based to a single sensor system at present create false alarms while they take extended time to respond to actual fires. The implemented system will be using three types of sensors which include temperature sensors, gas sensors and flame sensors to enhance both detection performance and operational system reliability. The ESP32 microcontroller will be used to control the whole works flow of other components which have the LCD screen to display the real time data, GPS and GSM modules with LED and buzzer to alert users away from the fire occurred. The Firebase platform will used to display the real time data while the MIT App Inventor and Telegram platforms are used to make multiple alerts for users. The simulation environment provides a complete testing platform which recreate all aspects of real-world testing from standard operations through early smoke detection to fire emergencies and equipment failures and temporary operational disruptions. The testing process evaluates how the system operates under standard conditions and handles unexpected events such as smoke only or flicker scenarios. The system performs tests by using different sensor fusion methods which include OR, AND, Weighted, Time Sequence, Two Stage Confirmation, Fuzzy, Adaptive Baseline, Trend Predict, Frequency Analysis and AI Decision Tree Logics. The analysis results show that using multiple sensors together with sensor fusion technology improves fire detection capabilities by decreasing false alarms and increasing response speed. Advanced logic methods demonstrate superior ability to handle noise and sensor faults and short-term disturbances when compared to traditional methods. The analysis also demonstrates that reliable fire detection systems require integration of multiple sensor systems and advanced decision-making algorithms. The main logic will be adapted in this system will be AI Decision Tree logic which provides an accuracy of 100% and the response time is about ~ 0.5s to instant ideally but in real world case may response with a little delay about 5 ~ 7.5s.