Journal of the Korean Society of Marine Environment & Safety (해양환경안전학회지)

The Korean Society of Marine Environment and safety (해양환경안전학회)
권호 표지
DOI QR코드

DOI QR Code

Fundamental Study on Algorithm Development for Prediction of Smoke Spread Distance Based on Deep Learning

딥러닝 기반의 연기 확산거리 예측을 위한 알고리즘 개발 기초연구

  • Kim, Byeol (Graduate school of Korea Maritime and Ocean University) ;
  • Hwang, Kwang-Il (Division of Mechanical Engineering, Korea Maritime and Ocean University)
  • 김별 (한국해양대학교 대학원) ;
  • 황광일 (한국해양대학교 기계공학부)
  • Received : 2020.11.02
  • Accepted : 2021.02.25
  • Published : 2021.02.28
Download PDF
⟨ Previous Next ⟩

Abstract

This is a basic study on the development of deep learning-based algorithms to detect smoke before the smoke detector operates in the event of a ship fire, analyze and utilize the detected data, and support fire suppression and evacuation activities by predicting the spread of smoke before it spreads to remote areas. Proposed algorithms were reviewed in accordance with the following procedures. As a first step, smoke images obtained through fire simulation were applied to the YOLO (You Only Look Once) model, which is a deep learning-based object detection algorithm. The mean average precision (mAP) of the trained YOLO model was measured to be 98.71%, and smoke was detected at a processing speed of 9 frames per second (FPS). The second step was to estimate the spread of smoke using the coordinates of the boundary box, from which was utilized to extract the smoke geometry from YOLO. This smoke geometry was then applied to the time series prediction algorithm, long short-term memory (LSTM). As a result, smoke spread data obtained from the coordinates of the boundary box between the estimated fire occurrence and 30 s were entered into the LSTM learning model to predict smoke spread data from 31 s to 90 s in the smoke image of a fast fire obtained from fire simulation. The average square root error between the estimated spread of smoke and its predicted value was 2.74.

본 연구는 화재진압 및 피난활동을 지원하는 딥러닝 기반의 알고리즘 개발에 관한 기초 연구로 선박 화재 시 연기감지기가 작동하기 전에 검출된 연기 데이터를 분석 및 활용하여 원격지까지 연기가 확산 되기 전에 연기 확산거리를 예측하는 것이 목적이다. 다음과 같은 절차에 따라 제안 알고리즘을 검토하였다. 첫 번째 단계로, 딥러닝 기반 객체 검출 알고리즘인 YOLO(You Only Look Once)모델에 화재시뮬레이션을 통하여 얻은 연기 영상을 적용하여 학습을 진행하였다. 학습된 YOLO모델의 mAP(mean Average Precision)은 98.71%로 측정되었으며, 9 FPS(Frames Per Second)의 처리 속도로 연기를 검출하였다. 두 번째 단계로 YOLO로부터 연기 형상이 추출된 경계 상자의 좌표값을 통해 연기 확산거리를 추정하였으며 이를 시계열 예측 알고리즘인 LSTM(Long Short-Term Memory)에 적용하여 학습을 진행하였다. 그 결과, 화재시뮬레이션으로부터 얻은 Fast 화재의 연기영상에서 경계 상자의 좌표값으로부터 추정한 화재발생~30초까지의 연기 확산거리 데이터를 LSTM 학습모델에 입력하여 31초~90초까지의 연기 확산거리 데이터를 예측하였다. 그리고 추정한 연기 확산거리와 예측한 연기 확산거리의 평균제곱근 오차는 2.74로 나타났다.

Keywords

References

  1. Arvidson, M., J. Axelsson, and T. Hertzberg(2008) Large-scale fire tests in a passenger Cabin, SP Technical Research Institute of Sweden, Report, pp. 1-100.
  2. Christopher, O.(2015), Understanding LSTM Networks, https://colah.github.io/posts/2015-08-Understanding-LSTMs/
  3. Hochreiter, S. and J. Schmidhuber(1997), Long short-term memory, Neural computation, Vol. 9, No. 8, pp. 1735-1780. https://doi.org/10.1162/neco.1997.9.8.1735
  4. Hwang, K. I., I. K Cho., G. H. Yun, and B. Kim(2018), A comparison of the trainees' evacuation characteristics according to the indoor smoke-fullfill during the safety training on ship, Journal of the Kore Society of Marine Environment & Safety, Vol. 24, No. 4, pp. 422-429. https://doi.org/10.7837/kosomes.2018.24.4.422
  5. IMO(2000), International Maritime Organization, MSC.98(73), Adoption of the International Code for Fire Safety Sysem, p. 19.
  6. Kim, B. and K. I. Hwang(2016), Smoke exhaust performance prediction according to air supply and exhaust conditions for shipboard fires from a human safety point of view, Journal of the Kore Society of Marine Environment & Safety, Vol. 22, No. 7, pp. 782-790. https://doi.org/10.7837/kosomes.2016.22.7.782
  7. Kim, B. and K. I. Hwang(2017), A Study on Crid Size and Generation Method for Fire Simulations for Ship Accommodation Areas, Journal of the Kore Society of Marine Environment & Safety, Vol. 23, No. 7, pp. 791-800. https://doi.org/10.7837/kosomes.2017.23.7.791
  8. Kim, Y. J. and H. C. Cho(2020), Detecting location of fire in video stream environment using deep learning, The transactions of the Korean Institute of Electrical Engineers, Vol. 69, No. 3, pp. 474-479. https://doi.org/10.5370/kiee.2020.69.3.474
  9. KMST(2019), Korea Maritime Safety Tribunal, Statistics for the marine accidents (in Korean), https://www.kmst.go.kr/kmst/statistics/annualReport/selectAnnualReportList.do.
  10. McGrattan, K., S. Hostikka, R. McDermott, J. Floyd, C. Weinschenk, and K. Overholt(2014), Fire Dynamics Simulator FDS User's Guide, 6th ed., National Institute of Standards and Technology, p. 35.
  11. Nazilly, M. L., B. Rahmat, and E. Y. Puspaningrum(2020), Implementation of YOLO Algorithm for Fire Detection, Jurnal Informatika dan Sistem Informasi, Vol. 1, No. 1, pp. 81-91.
  12. NFPA(2005), National Fire Protection Association, Standard for smoke management systems in malls, atria and large area, NFPA 92B.
  13. Park, K. M. and C. O. Bae(2019), A study on fire detection in ship engine rooms using convolutional neural network, Journal of the Kore Society of Marine Environment & Safety, Vol. 25, No. 4, pp. 476-481. https://doi.org/10.7837/kosomes.2019.25.4.476
  14. Redmon, J., S. Divvala, R. Girshick, and A. Farhadi(2016), You Only Look Onece: Unified, Real-Time Object Detection, In: CVPR.
  15. Shim, S. B. and S. I. Choi(2019), Development on Identification Algorithm of Risk Situation around Construction Vechicle using YOLO-v3, Journal of Korea AcademiaIndustrial cooperation Society, Vol. 20, No. 7, pp. 622-629.
  16. SOLAS(2015a), Regulation 3 Definitions, Chapter II-2 Construction-Fire Detection and Fire Extinction.
  17. SOLAS(2015b), Regulation 13 Means of Escape, Chapter II-2 Construction-Fire Detection and Fire Extinction.