DOI: https://doi.org/10.33099/2311-7249/2019-34-1-69-75

Method of forecasting of unfavorable aircraft accidents in the flight based on hybrid neural networks

Evhenii Gryshmanov, Iryna Zakharchenko, Iryna Novikova

Abstract


The paper describes the structure of a method for forecasting of unfavorable aircraft accidents in the flight that uses convolutional neural networks (CNN) and recurrent neural networks (RNN) based on LSTM modules. In this work, forecasting means solving the problem of analyzing text messages. They are presented as а structured and unstructured text and are formed based on data obtained from various sources of information in the process of air traffic control. The procedures for determining hyperactive parameters and training a hybrid neural network model for forecasting of unfavorable aircraft accidents in the flight using CNN networks and LSTM modules are considered in detail.

Keywords


flight safety; forecasting; convolutional neural network; recurrent neural network; LSTM module; activation function; tensor

References


1. Fact sheet. Status of civil aviation safety of the States Parties to the Agreement on Civil Aviation and the Use of Air in the first half of 2018. URL: https://mak-iac.org/upload/iblock/5b0/bp-18-1.pdf.

2. Statistical Summary of Commercial Jet Airplane Accidents. Worldwide Operations 19592008, Boeing Commercial Airplanes, Seattle, Washington 98124-2207, USA.

3. V.D.Sharov. Application of the Bayesian approach for specifying the probabilities of events in the automated system of forecasting and prevention of aviation accidents / V.D.Sharov // Management of large systems. – 2013. №43. – С.240 – 253.

4. V.P.Makarov. Method of forecasting and prevention of aviation accidents on the basis of the “Factors hazard tree” analysis: Dissertation Candidate of Technical Sciences / V.P.Makarov. – М., 2013. – P.137.

5. Y. Kim. Convolutional neural networks for sentence classification. arXiv:1408.5882 [cs.CL], 2014.

6. C. Olah. Neural networks, recurrent neural networks, convolutional neural networks. http://colah.github.io/.

7. Dos Santos C.N., Gatti M. Deep Convolutional Neural Networks for Sentiment Analysis of Short Texts // COLING, 2014. P. 69 – 70.

8. S.Hochreiter, J.Schmidhuber. Long short-term memory // Neural computation 9, 1997. Issue 8. P. 1735 – 1780.


GOST Style Citations


1. Справка. Состояние безопасности полетов
в гражданской авиации государств-участников соглашения о гражданской авиации и об использовании воздушного пространства в первом полугодии 2018 г. URL:  https://mak-iac.org/upload/iblock/5b0/bp-18-1.pdf.

2. Statistical Summary of Commercial Jet Airplane Accidents. Worldwide Operations 19592008, Boeing Commercial Airplanes, Seattle, Washington 98124-2207, USA.

3. Шаров В.Д. Применение байесовского подхода для уточнения вероятностей событий в автоматизированной системе прогнозирования и предотвращения авиационных происшествий / В. Д. Шаров // Управление большими системами. –2013. №43. – С.240 – 253.

4. Макаров В.П. Метод прогнозирования и предупреждения авиационных происшествий на основе анализа “дерева факторов опасности”: дисер. канд. техн. наук / В.П. Макаров. – М., 2013. – 137с.

5. Y. Kim. Convolutional neural networks for sentence classification. arXiv:1408.5882 [cs.CL], 2014.

6. C. Olah. Neural networks, recurrent neural networks, convolutional neural networks. http://colah.github.io/.

7. dos Santos C.N., Gatti M. Deep Convolutional Neural Networks for Sentiment Analysis of Short Texts // COLING, 2014. P. 69 – 70.

8. S.Hochreiter, J.Schmidhuber. Long short-term memory // Neural computation 9, 1997. Issue 8. P. 1735 – 1780.




ISSN 2410-7336 (Online)

ISSN 2311-7249 (Print)