Abstract. A methodology for conducting aerial radiation reconnaissance using a rotary-wing unmanned aerial vehicle (UAV) is presented. Aerial radiation reconnaissance is an effective monitoring tool enabling rapid assessment of contamination extent, determination of the volumetric concentration of radioactive materials in the atmosphere, and tracking of the spatial distribution of radionuclides. The developed methodology automates parameterizing the spatiotemporal distribution of radioactive materials in the atmosphere, minimizing time requirements while maintaining a specified confidence level in determining contamination boundaries and total released activity. Identifying the boundaries of a dispersing radionuclide plume via dynamic route optimization at waypoints involves the integrated use of a priori, a posteriori, and predictive data when solving the assimilative-adaptive problem of refining the route at UAV turning points (decision nodes). Determining atmospheric radioactive contamination parameters involves sequentially intersecting the released plume, measuring radiation intensity along the track, and constructing an atmospheric radioactive contamination profile with dynamic assimilative adaptation. Utilizing a pre-trained model enables rapid generation of a predicted concentration field of radioactive materials at each decision node, including uncertainty quantification of the system and reconnaissance route optimization, while preserving the specified confidence level in determining contamination parameters. Integrating a pre-trained predictive model with a bootstrap ensemble significantly improves the quality of the a priori assessment of the state of the contamination area and reduces computational costs when solving the problem under time constraints and limited measurement data. The bootstrap ensemble ensures robust reconstruction of a hypothetical contamination area with insufficient actual measurement data. The proposed approach provides a statistically sound determination of atmospheric radioactive contamination parameters.

Keywords: radioactive contamination of terrain, aerial radiation reconnaissance, decision node, dynamic route optimization, three-dimensional dynamic contamination model, radionuclide detection, pre-trained model

For citation: Ponomarenko A. V., Ustyuzhantsev A. M., Starkov V. D., Dryagin V. A. Method for determining the total amount of radioactive materials emitted into the atmosphere during aerial radiation reconnaissance via dynamic route optimization at waypoints using a pre-trained model. Aerospace Engineering and Technology. 2026. Vol. 4, no. 1, pp. 105–123. (In Russ.) https://doi.org/10.52467/2949-401X-2026-4-1-105-123; https://elibrary.ru/DKXCBF

References
1. Kozhemyakin V. A. Experience in the Development of Airborne Radiation Reconnaissance Equipment. Tezisy dokladov XV Mezhdunarodnogo soveshchaniya «Problemy prikladnoj spektrometrii i radiometrii» [Abstracts of the XV International Meeting “Problems of Applied Spectrometry and Radiometry”, October 7–11, 2019, Kazan, Russia]. (In Russ.). Kazan, 2019. pp. 66–67. https://elibrary.ru/YNHPIK
2. Bakin E. N., Petrikin A. N., Kolesov D. G. The helicopter type unmanned aerial vehicles application in the aerial nuclear and chemical reconnaissance establishing. Vozdushno-kosmicheskie sily. Teoriya i praktika [Aerospace Forces. Theory and Practice]. 2017. No. 3, pp. 7–14. (In Russ.) https://elibrary.ru/YNHPIK
3. Kharchenko E. V. Ispolʹzovanie matematicheskikh modelej perenosa i rasseyaniya radionuklidov v atmosfere dlya upravleniya riskami na stadii proektirovaniya atomnykh elektrostantsij: Dissertatsiya na soiskanie uchenoj stepeni kandidata fiziko-matematicheskikh nauk. [Using Mathematical Models of Radionuclide Transport and Dispersion in the Atmosphere for Risk Management at the Nuclear Power Plant Design Stage: Dissertation for the Degree of Candidate of Physical and Mathematical Sciences] St. Petersburg: RSUH, 2015. 117 p. (In Russ.)
4. Blashchinskaya O. N., Patrushev K. Yu. Modeling of dispersion of pollutants emission taking into account the effect of adsorption on the change of concentration in the atmosphere of populated areas. Scientific Papers Collection of the Angarsk State Technical University. 2023. No. 1, pp. 252–257. (In Russ.) https://doi.org/10.36629/2686-7788-2023-1-252-257
5. Tory K. Models of buoyant plume rise. Report № 451.2018. Bureau of Meteorology. Australian Government, Department of Industry, Innovation and Science. Bushfire and Natural Hazards CRC, 2018. 26 p.
6. Malinin V. N., Gordeeva S. M. About modern changes of global temperature of air. Society. Environment. Development (Terra Humana). 2011. Vol. 2, no. 19, pp. 215–221 (In Russ.) https://elibrary.ru/NYCNON
7. Sportisse B. A review of parametrization for modeling dry deposition and scavenging of radionuclides. Atmospheric Environment. 2007. Vol. 41, no. 13, pp. 2683–2698. https://doi.org/10.1016/j.atmosenv.2006.11.057
8. Bakels L., Tatsii D., Tipka A., Thompson R. et al. FLEXPART version 11: improved accuracy, efficiency, and flexibility. Geoscientific Model Development. 2024, Vol. 17, no 21, pp. 7595–7627. https://doi.org/10.5194/gmd-17-7595-2024
9. Kurbanshoev S. Z., Yakubov N. S. The averaging method for linear differential equations with periodic coefficients. Reports of the Academy of Sciences of the Republic of Tajikistan. 2017. Vol. 60, no. 7–8. Pp. 286–290. (In Russ.) https://elibrary.ru/ZHJOHH
10. Bellasio R., Bianconi R., Mosca S., Zannetti P. Formulation of the Lagrangian particle model LAPMOD and its evaluation against Kincaid SF6 and SO2 datasets. Atmospheric Environment. 2017. Vol. 163, pp. 87–98. https://doi.org/10.1016/j.atmosenv.2017.05.039
11. Klimova E. G. A stochastic ensemble kalman filter with perturbation ensemble transformation. Numerical analysis and applications. 2019. Vol. 12, no. 1, pp. 26–36. https://doi.org/10.1134/S1995423919010038 , https://elibrary.ru/MWQPVX
12. Fine T. L. Feedforward Neural Network Methodology, Springer-Verlag. New York, 1999. 340 p.
13. Panchenko A. A., Rakhman P. A., Safarov A. M. Neural network models for prediction of air pollution index in an industrial city. Modern Science: actual problems of theory and practice. Series: Natural and Technical Sciences. 2018. No. 5, pp. 121–126. (In Russ.) https://elibrary.ru/XWASST
14. Bühlmann P. Bootstrap schemes for time series. Quantile. 2007. No. 3, pp. 37–56. (In Russ.)
15. Barbashova E. V., Stroganova M. V. Hybrid decision support systems: integration of symbolic ai and machine learning. Ekonomika i kachestvo sistem svyazi [Economics and Quality of Communication Systems]. 2025. Vol. 4, no. 38, pp. 96–106. (In Russ.)