Skip to main content
SpringerLink
Log in

Generating Synthetic Sensor Data to Facilitate Machine Learning Paradigm for Prediction of Building Fire Hazard

  • Published:
Fire Technology Aims and scope Submit manuscript

Abstract

Using the zone fire model CFAST as the simulation engine, time series data for building sensors, such as heat detectors, smoke detectors, and other targets at any arbitrary locations in multi-room compartments with different geometric configurations, can be obtained. An automated process for creating inputs files and summarizing model results, CData, is being developed as a companion to CFAST. An example case is presented to demonstrate the use of CData where synthetic data is generated for a wide range of fire scenarios. Three machine learning algorithms: support vector machine (SVM), decision tree (DT), and random forest (RF), are used to develop classification models that can predict the location of a fire based on temperature data within a compartment. Results show that DT and RF have excellent performance on the prediction of fire location and achieve model accuracy in between 93% and 96%. For SVM, model performance is sensitive to the size of training data. Additional study shows that results obtained from DT and RT can be used to examine the importance of each input feature. This paper contributes a learning-by-synthesis approach to facilitate the utilization of a machine learning paradigm to enhance situational awareness for fire fighting in buildings.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9
Figure 10
Figure 11
Figure 12

Similar content being viewed by others

Notes

  1. CData is under active development. In the future version of CData, its fundamental elements, statistics of input parameters for residential buildings, and available distribution functions, can be varied to enhance data generation capacity and numerical efficiency.

  2. Hyperparameters can be thought of as the “dials” or “knobs” of a machine learning model.

  3. There are 24 sets of temperature profiles. Each temperature profile has 1000 s of data. With the minimum window size of 10 s, the maximum instances for this dataset will be 23,760.

  4. Accuracy is defined as the number of correct classified instances over the total number of instances.

  5. Feature importance is calculated based on the decrease in node impurity weighted by the probability of reaching that node.

References

  1. Hamins AP, Bryner NP, Jones AW, Koepke GH (2015) Research roadmap for smart fire fighting (No. Special Publication (NIST SP)-1191)

  2. Qolomany B, Al-Fuqaha A, Gupta A, Benhaddou D, Alwajidi S, Qadir J, Fong AC (2019) Machine learning, big data, and smart buildings: a comprehensive survey. arXiv:1904.01460

  3. Overholt KJ, Ezekoye OA (2012) Characterizing heat release rates using an inverse fire modeling technique. Fire Technol 48(4):893–909

    Article  Google Scholar 

  4. Lin CC, Wang LL (2017) Real-time forecasting of building fire growth and smoke transport via ensemble Kalman filter. Fire Technol 53(3):1101–1121

    Article  Google Scholar 

  5. Mahdavinejad MS, Mohammadreza R, Mohammadamin B, Peyman A, Payam B, Sheth AP (2018) Machine learning for Internet of Things data analysis: a survey. Digit Commun Netw 4(3):161–175

    Article  Google Scholar 

  6. Bishop CM (2006) Pattern recognition and machine learning. Springer, Berlin

    MATH  Google Scholar 

  7. Chenebert A, Breckon TP, Gaszczak A (2011) A non-temporal texture driven approach to real-time fire detection. In: 2011 18th IEEE international conference on image processing. IEEE, pp 1741–1744

  8. Yin Z, Wan B, Yuan F, Xia X, Shi J (2017) A deep normalization and convolutional neural network for image smoke detection. IEEE Access 5:18429–18438

    Article  Google Scholar 

  9. Shen D, Chen X, Nguyen M, Yan WQ (2018) Flame detection using deep learning. In: 2018 4th international conference on control, automation and robotics (ICCAR). IEEE, pp 416–420

  10. Aslan S, Güdükbay U, Töreyin BU, Çetin AE (2019) Deep convolutional generative adversarial networks based flame detection in video. arXiv:1902.01824

  11. Zhao Y, Ma J, Li X, Zhang J (2018) Saliency detection and deep learning-based wildfire identification in UAV imagery. Sensors 18(3):712

    Article  Google Scholar 

  12. Celik T, Demirel H, Ozkaramanli H, Uyguroglu M (2007) Fire detection using statistical color model in video sequences. J Vis Commun Image Represent 18(2):176–185

    Article  Google Scholar 

  13. McGrattan K, Hostikka S, McDermott R, Floyd J, Weinschenk C, Overholt K (2013) Fire dynamics simulator user’s guide, vol 1019, no 6. NIST Special Publication

  14. Yuen RK, Lee EW, Lo SM, Yeoh GH (2006) Prediction of temperature and velocity profiles in a single compartment fire by an improved neural network analysis. Fire Saf J 41(6):478–485

    Article  Google Scholar 

  15. Hodges JL, Lattimer BY, Luxbacher KD (2019) Compartment fire predictions using transpose convolutional neural networks. Fire Saf J 108:102854

    Article  Google Scholar 

  16. Lattimer BY, Hodges JL, Lattimer AM (2020) Using machine learning in physics-based simulation of fire. Fire Saf J 114: 102991

    Article  Google Scholar 

  17. Sharma J, Granmo OC, Goodwin M, Fidje JT (2017) Deep convolutional neural networks for fire detection in images. In: International conference on engineering applications of neural networks. Springer, Cham, pp. 183–193

  18. Lichman M (2013) UCI machine learning repository

  19. Bruns MC (2018) Estimating the flashover probability of residential fires using Monte Carlo simulations of the MQH correlation. Fire Technol 54(1):187–210

    Article  Google Scholar 

  20. https://www.eia.gov/consumption/commercial/data/2012/. Online, Accessed 26 Aug 2019

  21. Jones E, Oliphant E, Peterson P (2001) SciPy: open source scientific tools for python. http://www.scipy.org/. Online, Accessed 26 Aug 26

  22. Peacock RD, Reneke PA, Forney GP (2017) CFAST—consolidated model of fire growth and smoke transport (version 7) volume 2: user’s guide. NIST Technical Note 1889v2

  23. Mishra M, Rout PK (2017) Detection and classification of micro-grid faults based on HHT and machine learning techniques. IET Gen Transm Distribut 12(2):388–397

    Article  Google Scholar 

  24. Kazem HA, Yousif JH, Chaichan MT (2016) Modeling of daily solar energy system prediction using support vector machine for Oman. Int J Appl Eng Res 11(20): 10166–10172

    Google Scholar 

  25. Moutis P, Skarvelis-Kazakos S, Brucoli M (2016) Decision tree aided planning and energy balancing of planned community microgrids. Applied Energy 161: 197–205

    Article  Google Scholar 

  26. Jiang H, Li Y, Zhang Y, Zhang JJ, Gao DW, Muljadi E, Gu Y (2017) Big data-based approach to detect, locate, and enhance the stability of an unplanned microgrid islanding. J Energy Eng 143(5):04017045

    Article  Google Scholar 

  27. Vapnik V (1998) The support vector method of function estimation. In: Nonlinear modeling. Springer, Boston, pp 55–85

  28. Breiman L (2017) Classification and regression trees. Routledge, Abingdon

    Book  Google Scholar 

  29. Breiman L (2001) Random forests. Mach Learn 45(1):5–32

    Article  MATH  Google Scholar 

  30. Schittenkopf C, Deco G, Brauer W (1997) Two strategies to avoid overfitting in feedforward networks. Neural Netw 10(3):505–516

    Article  Google Scholar 

  31. Pedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O, Blondel M, et al. (2011) Scikit-learn: machine learning in python. J Mach Learn Res 12(Oct): 2825–2830

    MathSciNet  MATH  Google Scholar 

  32. Oliphant TE (2007) Python for scientific computing. Comput Sci Eng 9(3):10–20

    Article  Google Scholar 

  33. Staelin C (2003) Parameter selection for support vector machines. Hewlett-Packard Company, Tech. Rep. HPL-2002-354R1

  34. Jebara T (2004) Multi-task feature and kernel selection for SVMs. In: Proceedings of the twenty-first international conference on Machine learning. ACM, p. 55

Download references

Author information

Authors and Affiliations

  1. National Institute of Standards and Technology, Gaithersburg, MD, USA

    Wai Cheong Tam, Richard Peacock, Paul Reneke & Thomas Cleary

  2. Department of Computing, The Hong Kong Polytechnic University, Hung Hom, Hong Kong

    Eugene Yujun Fu & Jun Wang

  3. Department of Industrial Design, Guangdong University of Technology, Guangzhou, China

    Jiajia Li

Authors
  1. Wai Cheong Tam
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Eugene Yujun Fu
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Richard Peacock
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Paul Reneke
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Jun Wang
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Jiajia Li
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Thomas Cleary
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Wai Cheong Tam.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendices

Appendix 1

See Table 5.

Table 5 Summary of Important Experimental Parameters [22]
Full size table

Appendix 2

See Table 6.

Table 6 Summary of Statistics for all Quantities of Interest [22]
Full size table

Note that the term δ is a calculated bias factor representing the degree to which the model over-predicted or under-predicted experimental data, the term σM is a measure of model uncertainty, and the term σE is a measure of experimental uncertainty. The expression δ > 0 means the model over-predicted the observations, and σM < σ°E means that the model uncertainty is within experimental uncertainty.

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Tam, W.C., Fu, E.Y., Peacock, R. et al. Generating Synthetic Sensor Data to Facilitate Machine Learning Paradigm for Prediction of Building Fire Hazard. Fire Technol 59, 3027–3048 (2023). https://doi.org/10.1007/s10694-020-01022-9

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s10694-020-01022-9

Keywords

Search

Navigation