Advertisement

Neural Network Based Algorithm for the Measurements of Fire Factors Processing

  • A. I. Guseva
  • G. F. Malykhina
  • A. S. Nevelskiy
Conference paper
Part of the Studies in Computational Intelligence book series (SCI, volume 799)

Abstract

The objective of study is to develop a neural network algorithm for early warning of fire on a ship, using the results of measuring a multitude of sensors. The sensors are designed to measure the main fire factors such as temperature, concentration of carbon dioxide and the presence of smoke. Optimal coordinates of the sensors of the multi-sensor system were solved using suggested evolutionary algorithm. The study is based on the results of modeling fires on a supercomputer. The measurement of many fire factors allowed to develop an algorithm for determining the type of ignition. The algorithm uses a probabilistic neural network with delays at the input. Knowledge of the ignition type makes it possible to increase the speed of reaction to fire and to reduce damage on the ship. The suggested enhanced genetic algorithm fulfills optimization of the sensors position in the cabin. The algorithm allows to reduce the mean time of fire detection by of 15% in average. This effect can be achieved by increasing of the number of sensors in the cabin from 3 up to 11.

Keywords

Neural network Evolutionary algorithm Fire-fighting system Modeling 

Notes

Acknowledgment

The authors are gratitude the administration of the supercomputer center of Saint Petersburg State Polytechnic University for the opportunity to perform simulations to develop a neural network algorithm for early warning of fire on a ship, using simulation of sensors measurement results.

References

  1. 1.
    Guseva, A.I., Malykhina, G.F., Militsyn, A.V.: Algorithms of the early warning of fire in the premises of the vessel. In the collection: Complex protection of objects of information - 2016 Collection of scientific works of the all-Russian scientific and practical conference with international participation, p. 39–4 (Rus) (2016)Google Scholar
  2. 2.
    Malykhina, G.F., Guseva, A.I., Militsyn, A.V.: Early fire prevention in the plant. In: International Conference on Industrial Engineering, Applications and Manufacturing (ICIEAM), 19 May 2017 (2017)Google Scholar
  3. 3.
    Echegoyen, C., Mendiburu, A., Santana, R., Lozano, J.A.: On the taxonomy of optimization problems under estimation of distribution algorithms. Evol. Comput. 21(3), 471–495 (2012)CrossRefGoogle Scholar
  4. 4.
    Pernin, C.G., Comanor, K., Menthe, M., Moore, L.R., Anderson, T.: Allocation of Forces, Fires, and Effects Using Genetic Algorithms (Technical Report (RAND)) (2008). ISBN-13: 978-0833044792Google Scholar
  5. 5.
    Akbari, Z.: A multilevel evolutionary algorithm for optimizing numerical functions. J. Ind. Eng. Comput. 2, 419–430 (2011)CrossRefGoogle Scholar
  6. 6.
    Goldberg, D.E.: Genetic Algorithms in Search, Optimization, and Machine Learning, 1st edn., 403 pp. The University of Alabama, Addison-Wesley (1989). ISBN-13: 978-0201157673Google Scholar
  7. 7.
    Simon, H.: Neural Networks: A Comprehensive Foundation, 2nd edn. Prentice Hall, Upper Saddle River (1999). 842 pGoogle Scholar

Copyright information

© Springer Nature Switzerland AG 2019

Authors and Affiliations

  • A. I. Guseva
    • 1
  • G. F. Malykhina
    • 1
  • A. S. Nevelskiy
    • 1
  1. 1.Peter the Great Saint-Petersburg Polytechnic UniversitySaint PetersburgRussia

Personalised recommendations