Home / Current Issue / Paper 1712252
Design And Development of An Artificial Neural Network-Based Fire Detection System
Subject area: Science,Engineering and Technology · Area of research: Embedded Systems and Artificial Intelligence
Abstract
Fire outbreaks are often catastrophic, destroying lives and property. It is essential to develop an accurate and reliable fire detection system to safeguard lives and protect assets. Traditional fire detection approaches, such as using common sensors, are inaccurate and usually trigger false alarms. To address this issue and significantly improve the precision, accuracy, response time, and dependability of fire detection systems, we designed and developed an Artificial Neural Network (ANN)-based system. Our ANN-based fire detection system integrates six (6) sensors, including temperature, humidity, smoke, gas, flame, and light sensors, with an ESP32s microcontroller into a sensing node to maximize the properties of each sensor and reduce false alarms. Four sensing nodes were developed to capture environmental dimensions during data acquisition. A Central data Station (mainly comprising Raspberry Pi 3 B+ microcomputer, real-time clock, display screen, buzzer, and indicators) was also developed to serve as the processing device and the central hub for decision making. The ground truth was established using a manual switch attached to the Central data station (i.e., 'ON' or a '1' for fire scenarios and 'OFF' or a '0' for no fire situations), and data were collected on a Google sheet. The collected data was processed and used to train ANN models of different architectures and hyperparameters on the Central Data Station. The best model was selected using the F1-score evaluation metric. The trained model was deployed to make predictions in real-time. Compared to many conventional systems, the system demonstrated exceptional accuracy of 95% with a false alert rate of less than 3%. Additionally, the system's relative response time is good; on average, fires were detected within 10 seconds of their start. By providing enhanced security and ensuring a prompt response in the event of a fire, this state-of-the-art fire detection system offers a competitive alternative to traditional methods.
Keywords
Fire Detection, Artificial Neural Networks, Sensors, Microcontroller, Accuracy, Response Time.
References
[1] R. Dastres and M. Soori, “Artificial neural network systems,” International Journal of Imaging and Robotics (IJIR), vol. 21, no. 2, pp. 13–25, 2021.
[2] T. Tyagi, H. Sharma, K. Jain, and C. Mittal, “Handwritten Digit Recognition System using Machine Learning,” International Journal for Research in Applied Science & Engineering Technology (IJRASET), vol. 12, no. 5, pp. 3150–3154, May 2024. doi: 10.22214/ijraset.2024.62254
[3] S. A. Abdullah and A. Al-Ashoor, “An artificial deep neural network for the binary classification of network traffic,” International Journal of Advanced Computer Science and Applications, vol. 11, no. 1, pp. 402–408, 2020.
[4] C. Joshi, R. K. Ranjan, and V. Bharti, “ANN based multi-Class classification of P2P Botnet,” pp. 1319–1325, 2021.
[5] W. F. Schmidt, M. A. Kraaijveld, and R. P. W. Duin, “Feed forward neural networks with random weights,” in Proc. Int. Conf. on Pattern Recognition, IEEE Computer Society Press, 1992, pp. 1–1.
[6] F. Günther and S. Fritsch, “neuralnet: Training of neural networks,” 2010.
[7] A. Krenker, J. Bešter, and A. Kos, “Introduction to the artificial neural networks,” in Artificial Neural Networks: Methodological Advances and Biomedical Applications, InTech, 2011, pp. 1–18.
[8] J. Lederer, “Activation functions in artificial neural networks: A systematic overview,” arXiv preprint arXiv:2101.09957, 2021.
[9] A. Lichtner-Bajjaoui, “A mathematical introduction to neural networks,” 2021.
[10] M. Cilimkovic, “Neural networks and back propagation algorithm,” Institute of Technology Blanchardstown, no. 1, pp. 18, 2015.
[11] N. Zhang, S.-L. Shen, A. Zhou, and Y.-S. Xu, “Investigation on performance of neural networks using quadratic relative error cost function,” IEEE Access, vol. 7, pp. 106642–106652, 2019.
[12] A. Buja, W. Stuetzle, and Y. Shen, “Loss functions for binary class probability estimation and classification: Structure and applications,” Working Draft, Nov. 2005, pp. 13.
[13] M. H. Sazlı, “A brief review of feed-forward neural networks,” Communications Faculty of Sciences University of Ankara Series A2-A3 Physical Sciences and Engineering, vol. 50, no. 1, 2006.
[14] D. Soni and A. Makwana, “A survey on MQTT: a protocol of Internet of Things (IoT),” in Proc. Int. Conf. on Telecommunication, Power Analysis and Computing Techniques (ICTPACT-2017), vol. 20, Apr. 2017.
[15] R. Ponomarenko et al., “Review of the environmental characteristics of fire extinguishing substances of different composition used for fires extinguishing of various classes,” 2019.
How to cite this paper
@article{1712252,
author = {Boluwatife Olayiwola Demokun, Akindele Segun Afolabi, Kehinde Usman Adewuyi; Timileyin Daniel Ajayi, Abdul-Quadri Hujatullahi Aliyu; Abdulmalik Adewale Ajayi, Daniel Oluwatobi Ajibola; Azeez Opeyemi Azeez},
title = {Design And Development of An Artificial Neural Network-Based Fire Detection System},
journal = {Iconic Research And Engineering Journals},
year = {2025},
volume = {9},
number = {5},
pages = {2160-2168},
issn = {2456-8880},
url = {https://www.irejournals.com/formatedpaper/1712252.pdf},
abstract = {Fire outbreaks are often catastrophic, destroying lives and property. It is essential to develop an accurate and reliable fire detection system to safeguard lives and protect assets. Traditional fire detection approaches, such as using common sensors, are inaccurate and usually trigger false alarms. To address this issue and significantly improve the precision, accuracy, response time, and dependability of fire detection systems, we designed and developed an Artificial Neural Network (ANN)-based system. Our ANN-based fire detection system integrates six (6) sensors, including temperature, humidity, smoke, gas, flame, and light sensors, with an ESP32s microcontroller into a sensing node to maximize the properties of each sensor and reduce false alarms. Four sensing nodes were developed to capture environmental dimensions during data acquisition. A Central data Station (mainly comprising Raspberry Pi 3 B+ microcomputer, real-time clock, display screen, buzzer, and indicators) was also developed to serve as the processing device and the central hub for decision making. The ground truth was established using a manual switch attached to the Central data station (i.e., 'ON' or a '1' for fire scenarios and 'OFF' or a '0' for no fire situations), and data were collected on a Google sheet. The collected data was processed and used to train ANN models of different architectures and hyperparameters on the Central Data Station. The best model was selected using the F1-score evaluation metric. The trained model was deployed to make predictions in real-time. Compared to many conventional systems, the system demonstrated exceptional accuracy of 95% with a false alert rate of less than 3%. Additionally, the system's relative response time is good; on average, fires were detected within 10 seconds of their start. By providing enhanced security and ensuring a prompt response in the event of a fire, this state-of-the-art fire detection system offers a competitive alternative to traditional methods.},
keywords = {Fire Detection, Artificial Neural Networks, Sensors, Microcontroller, Accuracy, Response Time.},
month = {November},
doi = {https://doi.org/10.64388/IREV9I5-1712252}
}