PREDICTION OF ROCKET STAGE IMPACT ZONES BASED ON A DIGITAL TWIN AND DEEP LEARNING
DOI:
https://doi.org/10.56132/2791-3368-2026-3-67-39-59Keywords:
digital twin, trajectory prediction, impact zone, separated rocket stage, Deep Q Network, extended Kalman filter, particle filter, Bayesian updating, Monte Carlo method, GIS-based risk analysisAbstract
An intelligent system is proposed for predicting the impact locations of separated launch vehicle stages. The system integrates digital twin technology, mathematical trajectory modeling, probabilistic state estimation, deep reinforcement learning, and spatial GIS-based risk analysis. The developed framework enables continuous collection and analysis of telemetry, meteorological, and geospatial data, synchronization of the virtual model with the real object, and dynamic correction of the calculated flight trajectory.To improve the estimation of the current process state, an extended Kalman filter, a particle filter, and Bayesian updating are employed. The selection of optimal actions aimed at reducing prediction error is performed using a Deep Q Network neural architecture. The system performance was evaluated through numerical simulation and Monte Carlo analysis under varying wind speeds, stage-separation parameters, and levels of telemetry data loss.The experimental results demonstrate that the combined use of digital twin technology and the DQN algorithm reduces the RMSE and MAE to 19.7 m and 15.3 m, respectively, while achieving a prediction accuracy of 99.1%. In addition, a GIS-based risk map was developed for the Ulytau Region by integrating data on the probability of impact events, population density, critical infrastructure, and potential economic damage.The proposed method can be applied to improve the safety of space launches, support real-time monitoring, and assist decision-making processes.
Downloads
References
1. R. Wei, R. Yang, S. Liu, C. Fan, R. Zhou, Z. Wu, H. Wang, Y. Cai, and Z. Jiang.(2024). Towards an Extensible Model-Based Digital Twin Framework for Space Launch Vehicles. Journal of Industrial Information Integration.Vol. 41, article 100641. DOI: 10.1016/j.jii.2024.100641.
2. L. Schena, P. A. Marques, R. Poletti, S. Ahizi, J. Van den Berghe, and M. A. Mendez.(2024).Reinforcement Twinning: From Digital Twins to Model-Based Reinforcement Learning. Journal of Computational Science.Vol.82, article 102421. DOI: 10.1016/j.jocs.2024.102421
3. N. Gratius, Z. Wang, M. Y. Hwang, Y. Hou, A. Rollock, C. George, M. Bergés, and B. Akinci.(2024). Digital Twin Technologies for Autonomous Environmental Control and Life Support Systems. Journal of Aerospace Information Systems. Vol. 21, no. 4, pp. 332–347. DOI: 10.2514/1.I011320.
4. S. Henao-Garcia, M. Kapteyn, K. E. Willcox, M. Tezzele, M. Castroviejo-Fernandez, T. Kim, I. Kolmanovsky, H. Basu, P. Jirwankar, and R. G. Sanfelice. (2025). Digital-Twin-Enabled Multi-Spacecraft On-Orbit Operations. In AIAA SCITECH 2025 Forum, Orlando, FL, USA, AIAA Paper 2025-1432. DOI: 10.2514/6.2025-1432
5. S. Leon and H. Moncayo. (2025). Machine-Learning-Based Validation of a Digital Twin Spacecraft for Health-Management Systems. In AIAA SCITECH 2025 Forum, Orlando, FL, USA, AIAA Paper 2025-0561. DOI: 10.2514/6.2025-0561.
6. M. E. Iliuţă et al. (2024). Digital Twin—A Review of the Evolution from Concept to Technology and Its Analytical Perspectives on Applications in Various Fields.Vol. 14, no. 13, article 5454. DOI: 10.3390/app14135454
7. C. Bae, E. Choi, and S. Lee.(2025).Technologies, Applications, and Challenges of Digital Twin Across Industries: A Systematic Review of the State-of-the-Art Literature. IEEE Access. Vol. 13, pp. 152843–152869. DOI: 10.1109/ACCESS.2025.3601615
8. B. A. Adu-Gyamfi.(2026).The Role of Digital Twin Technology in Enhancing Sustainable Aviation Transition: A State-of-the-Art Review and Future Direction. Journal of Open Innovation: Technology, Market, and Complexity. Vol. 12, no. 1, article 100693. DOI: 10.1016/j.joitmc.2025.100693.
9. P. Razzaghi, A. Tabrizian, W. Guo, S. Chen, et al. (2024).A Survey on Reinforcement Learning in Aviation Applications. Engineering Applications of Artificial Intelligence. Vol. 136, Part A, article 108911. DOI: 10.1016/j.engappai.2024.108911.
10. G. Peng, B. Wang, L. Liu, H. Fan, and Z. Cheng. (2023). Real-Time Adaptive Entry Trajectory Generation with Modular Policy and Deep Reinforcement Learning. Aerospace Science and Technology.Vol. 142, Part A, article 108594. DOI: 10.1016/j.ast.2023.108594.
