Computing safe parameters of maneuver commencing of aerobatics aircraft using artificial neural network

Aeronautical and Space-Rocket Engineering

Dynamics, ballistics, movement control of flying vehicles


Аuthors

Ied K. 1*, Maslennikova G. E.2**, Tiumentsev Y. V.1***

1. Moscow Aviation Institute (National Research University), 4, Volokolamskoe shosse, Moscow, А-80, GSP-3, 125993, Russia
2. ,

*e-mail: Kaissied@mai.ru
**e-mail: maslennikova@gosniiga.ru
***e-mail: yutium@gmail.com

Abstract

The article considers artificial neural networks employed for sporting aircraft maneuvers computing method developing. System approach, describing it in the form of the MPL neuron network, is used for representation of such network. As long as initial training data represent complex functional dependencies with the number of variables greater than two, conventional approximation methods application is complicated. Thus, neural network modelling was employed for the problem solution. The concept of neuron represents the basis of neuron representation of aircraft flight trajectories (in the context of movement determining for an AIRCRAFT, and in the context of detecting and tracking devices). Correction of the MPL network architecture structure means the number of hidden layers and neurons (nodes) in each layer. Activation functions for each level are selected at this stage as well, i.e. they are assumed to be known. Weights and deflections are the unknown parameters with should be evaluated. Whereas excitations from the other neurons are fed to the input. For practical implementation of this approach a mathematical model of the Yak-55M sporting aircraft kind was developed on the X-Plane flight simulator using an algorithm of the training cycle of the network of multi-layer perceptron. The article presents also simulation results of the set problem on computing the safe parameters of a sporting aircraft maneuver starting. The study demonstrates that the neural network properties, such as non-linearity and good generalization ability, enhance its ability for complex tasks learning and can produce correct result for new initial data. The aircraft under analysis is out of effective system for collisions with ground prevention based on the predicted course of evasive manoeuver. However, the problem can be solved by developing relationship between the piloting errors and flight safety, and employing neuron network modelling for a number of maneuvers, which associate velocity and altitude parameters and automatically compared with the preset values. The model demonstrated the results of the sporting aircraft maneuvering starting parameters computing. With this, the probability of reliability of a great number of maneuvers should correspond to the reality. The results obtained while mathematical modelling should be loaded to the warning system to warn the pilot on the maneuver performing at the inappropriate altitude, and offer the recovery from the manoeuver allowing secure the flight and minimize hazardous situation.

Keywords:

aerobatics aircraft, prevention of flight accidents, delay of recovery from a maneuver, safe parameters of a maneuver commencing, warning system, multi-layer perceptron, neural network

References

  1. Ied K. Developing a technique for hazardous situations warning system design while piloting errors occurrence. Aerospace MAI Journal, 2019, vol. 26, no. 3, pp. 201-209.

  2. Korovin A.E., Novikov Yu.F. Prakticheskaya aerodinamika i dinamika poleta samoletov Yak-52 i Yak-55 (Practical aerodynamics and flight dynamics of Yak-52 and Yak-55), Moscow, DOSAAF SSSR, 1989, 357 p.

  3. Komartsova L.G., Maksimov A.V. Neirokomp’yutery (Neurocomputers), 2nd edition, Moscow, MGTU im. N.E Baumana, 2004, 400 p.

  4. Certification Specifications CS-23 Normal, Utility, Aerobatic, and Commuter Aeroplanes. European Aviation Safety Agency (EASA), 2010, 404 p. URL: https://www.easa.europa.eu/certification-specifications/cs-23-normal-utility-aerobatic-and-commuter-aeroplanes

  5. Airplane flying Handbook: ASA FAA-H-8083-3B. Aviation Supplies & Academics, Inc., 2016, 352 p.

  6. Aviatsionnye pravila. Ch. 23. Normy letnoi godnosti grazhdanskikh legkikh samoletov (Civil Aviation Safety Regulations. Part 23 Airworthiness standards for civil light aircraft), Moscow, Aviaizdat, 2013, 210 p.

  7. Kuznetsov Yu.V., Baev A.B., Konovaluk M.A. Multi-point scatterer target identification using radar image spectrum. Aerospace MAI Journal, 2010, vol. 17, no. 3, pp. 193-198.

  8. Haykin S. Neural Networks: A Comprehensive Foundation. 2nd Edition, Prentice Hall, New Jersey, 1998, 842 p.

  9. Kondratiev A.I., Tiumentsev Yu.V. Neural network modeling of controlled motion for aircraft. Aerospace MAI Journal, 2010, vol. 17, no. 5, pp. 5-11.

  10. Tan W., Efremov A.V., Tiumentsev Yu.V. An evaluation of aircraft handling qualities by means of neural network based prediction model for characteristics of pilot control actions. Aerospace MAI Journal, 2008, vol. 15, no. 1, pp. 92-108.

  11. Efremov A.V., Ogloblin A.V., Tan W., Tiumentsev Yu.V. A neural network model of pilot control behavior. Aerospace MAI Journal, 2007, vol. 14, no. 2, pp. 53-66.

  12. Evdokimenkov V.N., Kim R.V., Vekshina A.B., Yakimenko V.A. Study of pilot’s control actions personality during landing based on neural network models. Aerospace MAI Journal, 2015, vol. 22, no. 3, pp. 17-29.

  13. Goncharenko V.I. Informatsionno-izmeritel’nye i upravlyayushchie sistemy, 2012, vol. 10, no. 4, pp. 54-60.

  14. Soldatkin V.M. Metody i sredstva postroeniya bortovykh informatsionno-upravlyayushchikh sistem obespecheniya bezopasnosti poleta (Methods and means of onboard information and control systems constructing for flight safety), Kazan, Kazanskii gosudarstvennyi tekhnicheskii universitet, 2004, 350 p.

  15. Berestov L.M., Kharin E.G., Yakushev A.F. et al. Patent RU 2128854 C1, 10.04.1999.

  16. Petunin V.I., Neugodnikova L.M. Vestnik UGATU, 2014, vol. 18, no. 4(65), pp. 90-104.

  17. Enhanced Ground Proximity Warning System (EGPWS). URL: https://aerospace.honeywell.com/en/pages/enhanced-ground-proximity-warning-system

  18. Rutkowska D. Sieci neuronowe, algorytmy genetyczne isystemy rozmyte. Warszawa, Јodї: Wydawnictwo Naukowe PWN, 2004. cop. 83-01-12304-4

  19. Skoog M.A., Less J.L. Development and Flight Demonstration of a Variable Autonomy Ground Collision Avoidance System. American Institute of Aeronautics and Astronautics, 2014, 22 p. URL: https://pdfs.semanticscholar.org/eb7a/bebfaa354054c7f5497d1e765ec58578f8c7.pdf?_ga=2.169496804.455545469.1584376893-204996624.1582911794

  20. Kuznetsov Yu.V., Baev A.B., Konovaluk M.A. Multi-point scatterer target identification using radar image spectrum. Aerospace MAI Journal, 2010, vol. 17, no. 3, pp. 193-198.

  21. Kussul’ E.M. Assotsiativnye neiropodobnye struktury (Associative neuro-like structures), Kiev, Naukova dumka, 1992, 140 p.

  22. Piegat A. Fuzzy modeling and control. Physica-Verlag, Springer Science & Business Media, 2001, 728 p.

  23. Khlopkov Yu.I., Dorofeev E.A., Zeya M’o, Polyakov M.S., Khlopkov A.Yu., Agaeva I. Fundamental’nye issledovaniya. 2013, no. 11-9, pp. 1834-1840.

  24. Samarasinghe S. Neural Networks for Applied Sciences and Engineering: From Fundamentals to Complex Pattern Recognition. 1st edition. Auerbach Publications, 2006, 570 p.

  25. Hampton T.W. The quest for air dominance: F-22 Cost versus capability. Doctoral thesis, Air command and staff college, April, 1998.

  26. Prostov Yu.S., Tyumentsev Yu.V. Trudy MAI, 2014, no. 78. URL: http://trudymai.ru/eng/published.php?ID=53557

  27. Egorchev M.V., Kozlov D.S., Tiumentsev Yu.V. Aircraft aerodynamic model identification: a semi-empirical neural network based approach. Aerospace MAI Journal, 2014, vol. 21, no. 4, pp. 13-24.

mai.ru — informational site of MAI

Copyright © 1994-2020 by MAI