View in EDS

Machine learning and numerical simulation based prediction of the penetration efficiency of depleted uranium long-rod projectiles: a multi-factor analysis.

Saved in:
Bibliographic Details
Title: Machine learning and numerical simulation based prediction of the penetration efficiency of depleted uranium long-rod projectiles: a multi-factor analysis.
Authors: Wang, Ji-rui1 (AUTHOR), Tang, Kui1 (AUTHOR) tkui2014@sina.com, Wang, Jin-xiang1 (AUTHOR) wjx@njust.edu.cn, Gu, Min-hui1 (AUTHOR), Li, Yuan-bo1 (AUTHOR)
Source: International Journal of Impact Engineering. Mar2026, Vol. 209, pN.PAG-N.PAG. 1p.
Subjects: Machine learning, Artificial neural networks, Computer simulation, Uranium, Factor analysis, Mechanical efficiency
Abstract: • Application of artificial neural network on long-rod penetration. • Equivalent strength – hardness relationship of target was summarised. • L / D effect is more significant with low initial velocity and high target hardness. • Change in penetration mode was found with low initial velocity and large L / D ratio. Long-rod projectile is the predominant type of modern kinetic energy (KE) penetrator, and depleted uranium alloy (DU) serves as one of its primary materials. Although DU penetrators exhibit excellent penetration performance, their radioactive nature limits the availability of experimental results. Moreover, it is challenging for theoretical or semi-empirical models to accurately estimate the penetration efficiency (P / L) of DU penetrator under the influence of multiple factors. In this study, a novel data-driven machine learning framework based on artificial neural network (ANN) was developed to predict the penetration efficiency of DU long-rod projectiles with length-to-diameter (L / D) ratios ranging from 10 to 35 and initial velocities between 1200 and 2200 m/s when impacting semi-infinite armour steel targets with hardness levels between 270 and 579 BHN. A dataset comprising 180 examples derived from validated numerical simulations with LS-DYNA was utilized to train and test the neural network, achieving high accuracy while effectively avoiding overfitting. The neural network model revealed that the relationship between the equivalent strength and the target hardness is monotonically increasing and concave, exhibiting a nearly linear trend within the hardness range of 250 to 600 BHN. Additionally, the L / D effect has a negative correlation with initial velocity but a positive correlation with target hardness. Furthermore, when the initial velocity is low and the L / D ratio is high, subsidiary radial penetration occurs, leading to a significant reduction in penetration efficiency. [ABSTRACT FROM AUTHOR]
Copyright of International Journal of Impact Engineering is the property of Pergamon Press - An Imprint of Elsevier Science and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Engineering Source
Description
Abstract:• Application of artificial neural network on long-rod penetration. • Equivalent strength – hardness relationship of target was summarised. • L / D effect is more significant with low initial velocity and high target hardness. • Change in penetration mode was found with low initial velocity and large L / D ratio. Long-rod projectile is the predominant type of modern kinetic energy (KE) penetrator, and depleted uranium alloy (DU) serves as one of its primary materials. Although DU penetrators exhibit excellent penetration performance, their radioactive nature limits the availability of experimental results. Moreover, it is challenging for theoretical or semi-empirical models to accurately estimate the penetration efficiency (P / L) of DU penetrator under the influence of multiple factors. In this study, a novel data-driven machine learning framework based on artificial neural network (ANN) was developed to predict the penetration efficiency of DU long-rod projectiles with length-to-diameter (L / D) ratios ranging from 10 to 35 and initial velocities between 1200 and 2200 m/s when impacting semi-infinite armour steel targets with hardness levels between 270 and 579 BHN. A dataset comprising 180 examples derived from validated numerical simulations with LS-DYNA was utilized to train and test the neural network, achieving high accuracy while effectively avoiding overfitting. The neural network model revealed that the relationship between the equivalent strength and the target hardness is monotonically increasing and concave, exhibiting a nearly linear trend within the hardness range of 250 to 600 BHN. Additionally, the L / D effect has a negative correlation with initial velocity but a positive correlation with target hardness. Furthermore, when the initial velocity is low and the L / D ratio is high, subsidiary radial penetration occurs, leading to a significant reduction in penetration efficiency. [ABSTRACT FROM AUTHOR]
ISSN:0734743X
DOI:10.1016/j.ijimpeng.2025.105534