S.A. Kurbatsky1, E.I. Minakov2, A.N. Grachev3, M.S. Basherov4, А.А. Panova5

1,3,4,5 JSC “Central Design Bureau of Apparatus Engineering” (Tula, Russia)

2,4 Tula State University (Tula, Russia)

1 qrbs@rambler.ru, 2 eminakov@bk.ru, 3 ga150161@mail.ru, 4 basherov666@yandex.ru, 5 panova005@gmail.com

One of the significant problems in the application of neural network methods for classification of aerial targets in surveillance pulse-Doppler radar systems (radars) using trajectory features and radar cross-section (RCS) is the limited volume of experimental radar data required to form representative training datasets. Conducting full-scale tests involves significant time, organizational and financial costs, which leads to insufficient representation of individual classes of aerial objects and complicates the training of neural network classifiers with high generalization capability. The use of synthetic data alone also does not fully ensure adequate reproduction of the features of real radar information.

Development of a methodology for integrated formation of training datasets for neural network classification of aerial targets using trajectory features and radar cross-section based on combined use of experimental and synthetic radar data.

A methodology for integrated formation of training datasets for neural network classification of aerial targets using trajectory features and radar cross-section is proposed, based on combined use of experimental and synthetic radar data. The methodology includes generation of synthetic trajectories based on physically sound models of motion of aerial objects, taking into account constraints on kinematic parameters and maneuverability, as well as statistical augmentation of training data using a priori information on ranges of variation of trajectory features and RCS of various target classes. It is shown that application of the proposed approach ensures expansion of the space of possible states of objects, improvement of feature distribution uniformity, and increase in training data representativeness.

The proposed methodology can be used for formation of training datasets for intelligent automatic recognition systems of aerial targets in surveillance pulse-Doppler radars and allows increasing data representativeness under conditions of limited volume of experimental radar measurements.

The work was carried out with the financial support of a Grant in the form of a research grant from the Committee on Science and Innovation of the Tula Region No. 15 dated June 21, 2024.

Kurbatsky S.A., Minakov E.I., Grachev A.N., Basherov M.S., Panova А.А. Methodology for generating datasets for training neural network models for classification of aerial targets // Radioengineering. 2026. V. 90. № 9. P. 64−76. DOI: https://doi.org/10.18127/j00338486-202609-07

1. Dale H., et al. SNR-dependent drone classification using convolutional neural networks. IET Radar, Sonar & Navigation. 2022. V. 16. № 1. P. 22–33.
2. Mohajerin N., Histon J., Dizaji R., Waslander S.L. Feature extraction and radar track classification for detecting UAVs in civilian airspace. 2014 IEEE Radar Conference (RadarCon). Cincinnati. USA. 2014. P. 674–679. DOI: 10.1109/RADAR.2014.6875676.
3. Chen W., Liu J., Li J. Classification of UAV and bird target in low-altitude airspace with surveillance radar data. The Aeronautical Journal. 2019. V. 123. P. 191–211. DOI: 10.1017/aer.2018.158.
4. Zyuzin A.V., Minakov E.I., Basherov M.S., Filonovich A.V. Raspoznavanie malorazmernyh ob"ektov s ispol'zovaniem iskusstvennyh nejronnyh setej v zadachah radiolokacii. Elektromagnitnye volny i elektronnye sistemy. 2026. T. 31. № 1. S. 14–27. DOI: 10.18127/j15604128-202601-02 (in Russian).
5. Gridin V.N., Kurbatskij S.A., Minakov E.I., Evdokimov I.A., Polynkin A.V., Basherov M.S. Matematicheskaya model' formirovaniya mikrodoplerovskoj sostavlyayushchej radiolokacionnogo signala dlya razlichnyh klassov bespilotnyh letatel'nyh apparatov. Radiotekhnika i elektronika. 2026. T. 71. № 4. S. 465–475 (in Russian).
6. Minakov E.I., Makareckij E.A., Basherov M.S. Radiolokacionnaya sistema dlya uluchsheniya situacionnoj osvedomlyonnosti vozdushnyh sudov. Elektromagnitnye volny i elektronnye sistemy. 2025. T. 30. № 4. S.76–85. DOI: 10.18127/j15604128-202504-07 (in Russian).
7. Zhang X., Mehta V., Bolic M., Mantegh I. Hybrid AI-enabled Method for UAS and Bird Detection and Classification. 2020 IEEE International Conference on Systems, Man, and Cybernetics (SMC). Toronto. Canada. 2020. P. 2803–2807. DOI: 10.1109/SMC42975.2020.9282965.
8. Liu J., Xu Q., Chen W. Classification of Bird and Drone Targets Based on Motion Characteristics and Random Forest Model Using Surveillance Radar Data. IEEE Access. 2021. V. 9. P. 160135–160144. DOI: 10.1109/ACCESS.2021.3130231.
9. Kostromickij S.M., Nefyodov D.S. Radiolokacionnye harakteristiki mikro-BPLA. Vestnik Koncerna VKO «Almaz – Antej». 2023. № 3. S. 12–23. DOI: 10.38013/2542-0542-2023-3-12-23 (in Russian).
10. Kostromickij S.M., Nefyodov D.S., Hramenkov A.S., Chigryaj V.G. Statisticheskie modeli fluktuacij effektivnoj poverh-nosti rasseyaniya malorazmernyh bespilotnyh letatel'nyh apparatov. Vestnik Koncerna VKO «Almaz – Antej». 2023. № 3. S. 24–36. DOI: 10.38013/2542-0542-2023-3-24-36 (in Russian).
11. Pal'guev D.A., Parhachyov V.V., Piunov K.N., Vasil'ev D.A. Raspoznavanie vozdushnyh ob"ektov tipa «pticy» po traektornym priznakam. Radiotekhnicheskie i telekommunikacionnye sistemy. 2023. № 4. S.39–51. DOI: 10.24412/2221-2574-2022-4-39-50 (in Russian).
12. Fioranelli F., Ram S. S., Kernec J. L., Gurbuz S. Foreword to the Special Section on AI Approaches for Radar Processing and Applications. IEEE Transactions on Radar Systems. 2025. V. 3. P. 1243–1256. DOI: 10.1109/TRS.2025.360130.
13. Minakov E.I., Grachev A.N., Basherov M.S., Shprekher D.M. Klassifikaciya vozdushnyh celej v RLS na osnove nejrosetevoj modeli s rasshirennym priznakovym prostranstvom. Programmnye produkty i sistemy. 2025. № 4. S. 682–693 (in Russian).
14. Macyura A.V. Radiolokacionnye ornitologicheskie nablyudeniya: kratkij obzor stacionarnyh i mobil'nyh kompleksov. Acta Biologica Sibirica. 2015. T. 1, № 3−4. S. 118–147. DOI: 10.14258/abs.v1i3-4.917 (in Russian).
15. Dinevich L.A. Raspoznavanie radioekha ptic na baze radiolokatora MRL-5. Nauka. Innovacii. Tekhnologii. 2014. № 1(5). S. 7–37 (in Russian).
16. Dinevich L., Leshem Y. Radar monitoring of seasonal bird migration over central Israel. The Ring. 2010. V. 32. № 1−2. P. 31–53. DOI: 10.2478/v10050-010-0003-z.
17. Maltsev K., Yaroslavsky L., Leshem Y. Processing of Weather Radar Images for Bird Detection and Tracking. Proceedings of the 7th IASTED International Conference on Visualization, Imaging and Image Processing (VIIP 2007). Palma de Mallorca, Spain. 2007.
18. Osadchyi V.V., Eremeev V.S., Matsyura A.V. Mathematical model of bird species identifying: implication of radar data processing. Biological Bulletin of Bogdan Chmelnitsky Melitopol State Pedagogical University. 2016. V. 6, № 3. P. 463–471. DOI: 10.15421/2016119.
19. Bayandina T.A. Dinamika polyota samoleta. Raschet traektorij i letnyh harakteristik: Uchebnoe posobie. Samara: Izd-vo Samarskogo universiteta. 2023. 152 s. (in Russian).
20. Chumadin A.S., Ershov V.I., Barvinok V.A. i dr. Teoreticheskie osnovy avia- i raketostroeniya v konspektah lekcij: Uchebnoe posobie dlya vuzov. M.: Drofa. 2005. 784 s. (Ser. Proektirovanie, konstrukciya i proizvodstvo letatel'nyh apparatov) (in Russian).
21. Onushkin Yu.P., Kochish G.G. Aerodinamika, dinamika poleta i prakticheskaya aerodinamika. Ch. II. Dinamika poleta: kurs lekcij. Syzran': Syzranskij voennyj aviacionnyj institut. 2004. 162 s. (in Russian).
22. Petrovskij S.P., Dzhaksbaev V.A. Prakticheskaya aerodinamika vertoleta Mi-8T(MT): Uchebnoe posobie. V 2-h chastyah. Ch. 1. Syzran': VUNC VVS «Voenno-vozdushnaya akademiya imeni professora N.E. Zhukovskogo i Yu.A. Gagarina». 2016. 122 s. (in Russian).
23. Tereshchenkov E.V., Matveev A.M., Meshcheryakov V.A. Osnovy aviacionnoj tekhniki i prakticheskoj aerodinamiki: uchebnoe posobie. Pod obshch. red. V.A. Meshcheryakova. Ul'yanovsk: UVAU GA. 2007. 109 s. (in Russian).
24. Onushkin Yu.P., Mihajlov S.A., Nosov A.A. Prakticheskaya aerodinamika vertoleta ANSAT-U: uchebno-metodicheskoe posobie dlya letnogo sostava. Kazan': Izd-vo Kazanskogo gosudarstvennogo tekhnicheskogo universiteta. 2010. 225 s. (in Russian).
25. Fomichev P.A., Zaruckij A.V., Mandzyuk S.F. Raschet na prochnost' samoleta: Konspekt lekcij. Ch. 1. Har'kov: Nacional'nyj aerokosmicheskij universitet im. N.E. Zhukovskogo «Har'kovskij aviacionnyj institut». 2017. 165 s. (in Russian).
26. Chepurnyh I.V. Prochnost' konstrukcij letatel'nyh apparatov: uchebnoe posobie. Komsomol'sk-na-Amure: FGBOU VPO «KnAGTU». 2013. 137 s. (in Russian).
27. Chepurnyh I.V. Dinamika poleta samoletov: Uchebnoe posobie. Komsomol'sk-na-Amure: FGBOU VPO «KnAGTU». 2014. 112 s. (in Russian).
28. Moiseev G.V., Moiseev V.S. Osnovy teorii sozdaniya i primeneniya imitacionnyh bespilotnyh aviacionnyh kompleksov: monografiya. Kazan': Redakcionno-izdatel'skij centr. 2013. 208 s. (in Russian).
29. Nel'son H. Bazovaya matematika dlya iskusstvennogo intellekta: Per. s angl. Astana: ALIST. 2024. 592 s. (in Russian).
30. H'yuen Ch. Proektirovanie sistem mashinnogo obucheniya: Per. s angl. Astana: Foliant. 2023. 368 s. (in Russian).
31. Prikaz Federal'nogo agentstva vozdushnogo transporta ot 28.12.2022 № 970-P «Ob utverzhdenii Norm letnoj godnosti grazhdanskih legkih samoletov NLG-23». URL: https://legalacts.ru/doc/prikaz-rosaviatsii-ot-28122022-n-970-p-ob-utverzhdenii-norm/ (data obrashcheniya: 07.08.2026) (in Russian).
32. Federal'noe agentstvo vozdushnogo transporta. Normy letnoj godnosti bespilotnyh aviacionnyh sistem s bespilotnym vozdushnym sudnom samoletnogo tipa s maksimal'noj vzletnoj massoj do 5400 kg. NLG BAS-ST. Izdanie 1. M. 2023. 315 s. (in Russian).
33. Aviacionnye pravila. Ch. 29. Normy letnoj godnosti vintokrylyh apparatov transportnoj kategorii NLG 29: utv. prikazom Rosaviacii ot 07.12.2022 № 890-P. M.: Mezhgosudarstvennyj aviacionnyj komitet. 2023. 128 s. (in Russian).
34. Swerling P. Probability of detection for fluctuating targets. IRE Transactions on Information Theory. 1960. V. 6. № 2. P. 269–308.
35. Azim A. W., Bazzi A., Bomfin R., Giakoumidis N., Rappaport T. S., Chafii M. Indoor Statistical and Deterministic RCS Characterization for ISAC Channel Modeling. IEEE Transactions on Wireless Communications. 2026. DOI: 10.48550/arXiv.2411.03206.
36. Jaynes E.T. Information Theory and Statistical Mechanics. Physical Review. 1957. V. 106, № 4. P. 620–630.
37. Velishchanskij M.A. Sintez kvazioptimal'noj traektorii dvizheniya bespilotnogo letatel'nogo apparata. Nauka i obrazovanie. MGTU im. N.E. Baumana. Elektronnyj zhurnal. 2013. № 12. DOI: 10.7463/1213.0646471 (in Russian).
38. Kanatnikov A.N., Krishchenko A.P., Tkachev S.B. Dopustimye prostranstvennye traektorii bespilotnogo letatel'nogo apparata v vertikal'noj ploskosti. Nauka i obrazovanie. MGTU im. N.E. Baumana. Elektronnyj zhurnal. 2012. № 3 (in Russian).
39. Kanatnikov A.N. Postroenie traektorij letatel'nyh apparatov s nemonotonnym izmeneniem energii. Nauka i obrazovanie. MGTU im. N.E. Baumana. Elektronnyj zhurnal. 2013. № 4. DOI: 10.7463/0413.0554666 (in Russian).