城市建筑外爆威力场与毁伤效应数智仿真模型及应用

彭江舟; 潘刘娟; 高光发; 王祉乔; 胡杰; 吴威涛; 王明洋; 何勇

doi:10.11883/bzycj-2024-0471

城市建筑外爆威力场与毁伤效应数智仿真模型及应用

doi: 10.11883/bzycj-2024-0471

1.
南京理工大学机械工程学院，江苏南京 210094
2.
南京理工大学安全科学与工程学院，江苏南京 210094
3.
陆军工程大学爆炸冲击与防灾减灾国家重点实验室，江苏南京 210007

基金项目: 国家重点研发计划项目课题（Grant No. 2021YFC3100705）

详细信息

作者简介:
彭江舟（1996－　），男，博士，博士后，pengjz@njust.edu.cn

通讯作者:
何　勇（1964－　），男，博士，教授，he1964@mail.njust.edu.cn

中图分类号: O389
计量
- 文章访问数: 274
- HTML全文浏览量: 38
- PDF下载量: 86
- 被引次数: 0
出版历程
- 收稿日期: 2024-12-02
- 修回日期: 2025-03-19
- 网络出版日期: 2025-03-26

A digital intelligence simulation model for explosion power field and urban building damage effect and its application

1.
School of Mechanical Engineering, Nanjing University of Science and Technology, Nanjing 210094, Jiangsu, China
2.
School of Safety Science and Engineering, Nanjing University of Science and Technology, Nanjing 210094, Jiangsu, China
3.
State Key Laboratory of Explosion & Impact and Disaster Prevention & Mitigation, Army Engineering University of PLA, Nanjing 210007, Jiangsu, China

摘要

摘要: 为准确预测建筑外爆威力场，解决传统经验公式中未能充分考虑环境因素的复杂性而导致的精度受限、数值仿真在处理大规模城市场景时效率低下的难题，构建了一种基于图神经网络（graph neural network, GNN）的爆炸威力场预测模型，直接利用建筑的几何特征，对其表面的爆炸峰值超压、峰值冲量及冲击波到达时间等三维物理场的进行预测。与数值仿真结果的对比验证表明，本文模型展现出了卓越的预测性能：对不同几何结构的单体建筑表面超压参数的预测均方误差为0.97%；对复杂几何建筑、建筑群落建筑表面超压参数的平均预测误差为3.17%；当应用于实际城市区域时，平均预测误差为1.29%；物理场单次预测耗时不超过0.6 s，与数值仿真相比速度提升3~4个数量级。基于模型的高精度预测，不仅可以重构建筑表面任意位置的超压时程曲线，还能准确评估结构的毁伤程度。
- 爆炸威力场 /
- 毁伤评估 /
- 爆炸冲击 /
- 数智仿真 /
- 图神经网络
Abstract: To accurately predict the explosion power fields in buildings, solving the failure of traditional empirical formulas often failing to account for complex environmental factor due to their inability to account for complex environmental factors, and that of numerical simulations inefficient for large-scale urban scenarios and do not meet the needs of rapid damage assessment. Addressing this challenge, an innovative prediction model for explosion power fields based on Graph Neural Networks (GNN) was constructed using an end-to-end strategy. This model enabled rapid and precise forecasting of three-dimensional physical fields, including peak overpressure, peak impulse, and shock-wave arrival times on building surfaces. Compared to numerical simulations, the proposed GNN model demonstrated excellent predictive performance: it achieved a mean square error of 0.97% for predicting surface overpressure parameters of single buildings with varying geometries, and an average prediction error of 3.17% for complex geometric buildings and building communities. When applied to real-world urban settings, the model maintains an average prediction error of 1.29%, completing individual physical field predictions in under 0.6 seconds—three to four orders of magnitude faster than numerical simulations. Furthermore, the model's high-precision predictions allow for the reconstruction of overpressure time history curves at any building surface location and the accurate assessment of structural damage. The proposed GNN model offers a novel approach for rapidly and accurately predicting explosion power fields in urban buildings during blast events. This advancement significantly enhances the capabilities for explosion damage assessment and anti-explosion design in ultra-large-scale complex engineering scenarios, providing substantial engineering value.
- explosion power fields /
- damage assessment /
- explosion shock /
- data-driven intelligent simulation model /
- graph neural networks

HTML全文

图 1 方法流程图

Figure 1. Methodology flow chart

下载: 全尺寸图片幻灯片

图 2 实验与仿真的爆炸场景

Figure 2. Experimental and simulated explosion scenarios

下载: 全尺寸图片幻灯片

图 3 仿真方法的有效性验证

Figure 3. Validation of simulation methods

下载: 全尺寸图片幻灯片

图 4 建筑外爆数值仿真

Figure 4. Numerical simulation of building implosion

下载: 全尺寸图片幻灯片

图 5 200 kg当量爆炸在城市区域的冲击波耗散与传播特性

Figure 5. Dissipation and propagation characteristics of shock waves from a 200kg TNT equivalent explosion in urban areas.

下载: 全尺寸图片幻灯片

图 6 训练测试集空间分布示例

Figure 6. Spatial distribution example of the training test sets

下载: 全尺寸图片幻灯片

图 7 爆炸超压载荷参数预测模型网络架构

Figure 7. Network architecture of the blast overpressure load parameter prediction model

下载: 全尺寸图片幻灯片

图 8 训练损失收敛情况（1 epoch=3200 steps）

Figure 8. Training loss convergence (1 epoch=3200 steps)

下载: 全尺寸图片幻灯片

图 9 单体建筑测试集R²分布

Figure 9. R² distribution on the single building test sets

下载: 全尺寸图片幻灯片

图 10 单体建筑样本预测结果

Figure 10. Prediction results of single building samples

下载: 全尺寸图片幻灯片

图 11 一随机测试数据y=37 m截面处预测结果

Figure 11. Prediction results at y=37 m cross-section of a random test data

下载: 全尺寸图片幻灯片

图 12 复杂几何建筑与建筑群落样本预测结果

Figure 12. Prediction results of multiple buildings and complex geometric building samples

下载: 全尺寸图片幻灯片

图 13 复杂建筑构型一随机数据 x=55 m截面处预测结果

Figure 13. Prediction results at x=55 m cross-section of a random test data in complex geometric building samples

下载: 全尺寸图片幻灯片

图 14 建筑群落构型一随机数据 x=74 m截面处预测结果

Figure 14. Prediction results at x=74 m cross-section of a random test data in multiple buildings samples

下载: 全尺寸图片幻灯片

图 15 城市数字孪生建模和爆炸事件分布

Figure 15. Modeling of urban digital twins and distribution of explosion cases

下载: 全尺寸图片幻灯片

图 16 城市区域爆炸预测结果

Figure 16. Prediction results of explosion in urban areas

下载: 全尺寸图片幻灯片

图 17 超压时程曲线重构

Figure 17. Reconstruction of the overpressure time history curve

下载: 全尺寸图片幻灯片

图 18 峰值压力-峰值冲量曲线

Figure 18. Curves of peak pressure vs. peak impulse

下载: 全尺寸图片幻灯片

图 19 毁伤等级划分结果

Figure 19. Damage classification results

下载: 全尺寸图片幻灯片

表 1 GNN和OpenFOAM计算硬件配置对比表格

Table 1. 1 Computing Hardware Configuration Comparison Table of GNN and OpenFOAM

计算方法	操作系统	开发语言	CPU	GPU
GNN	Ubuntu 20.04	Python	Intel Xeon Gold 5317	GeForce RTX 4090ke
OpenFOAM	Ubuntu 20.04	C++	AMD EPYC 7H12	无显卡

下载: 导出CSV

表 2 复杂几何建筑与建筑群落测试集模型预测性能

Table 2. Predictive performance of GNN models on the test sets of the complex geometric building and multiple buildings

建筑类型	ε_RSE/%			计算时间/s
建筑类型	峰值超压	峰值冲量	到时	OpenFOM	GNN
复杂建筑—十二边形	1.15	2.34	0.85	991	0.15
复杂建筑—扇形	1.51	3.68	1.33	857	0.14
复杂建筑—十字形	1.16	6.69	2.74	984	0.35
建筑群落—2栋	1.95	8.16	2.04	991	0.27
建筑群落—3栋	1.67	7.57	3.07	1450	0.36
建筑群落—4栋	2.61	6.52	1.99	2045	0.51

下载: 导出CSV

表 3 城市区域爆炸事件GNN模型预测性能

Table 3. Predictive performance of the GNN model for the blasts in urban areas

爆炸事件	ε_RSE/%			计算时间/s
爆炸事件	峰值超压	峰值冲量	到时	OpenFOAM	GNN
1	2.74	6.06	0.89	1143	0.37
2	0.92	5.10	0.85	875	0.34
3	1.19	4.72	1.22	984	0.20
4	0.74	5.32	1.48	928	0.21
5	0.98	4.24	2.63	961	0.13
6	1.36	3.01	0.80	1105	0.23

下载: 导出CSV

参考文献(34)

[1]	殷文骏, 程帅, 刘文祥, 等. 远场爆炸冲击波作用下高层建筑上部结构动态响应试验研究 [J]. 兵工学报, 2024, 45(11): 4039–4051. DOI: 10.12382/bgxb.2023.0960. YIN W J, CHENG S, LIU W X, et al. Experimental study on dynamic response of upper structure of high-rise building under far-field explosion shock wave loading [J]. Acta Armamentarii, 2024, 45(11): 4039–4051. DOI: 10.12382/bgxb.2023.0960.
[2]	王明涛, 程月华, 吴昊. 柱形装药空中爆炸冲击波荷载研究 [J]. 爆炸与冲击, 2024, 44(4): 043201. DOI: 10.11883/bzycj-2023-0197. WANG M T, CHENG Y H, WU H. Study on blast loadings of cylindrical charges air explosion [J]. Explosion and Shock Waves, 2024, 44(4): 043201. DOI: 10.11883/bzycj-2023-0197.
[3]	童晓. 爆炸场冲击波压力测量及数据处理方法研究 [D]. 南京: 南京理工大学, 2015. TONG X. Research on shock wave pressure measurement and data processing method in explosion field [D]. Nanjing: Nanjing University of Science and Technology, 2015.
[4]	都浩, 李忠献, 郝洪. 建筑物外部爆炸超压荷载的数值模拟 [J]. 解放军理工大学学报(自然科学版), 2007, 8(5): 413–418. DOI: 10.3969/j.issn.1009-3443.2007.05.002. DU H, LI Z X, HAO H. Numerical simulation on blast overpressure loading outside buildings [J]. Journal of PLA University of Science and Technology, 2007, 8(5): 413–418. DOI: 10.3969/j.issn.1009-3443.2007.05.002.
[5]	李忠献, 师燕超, 周浩璋, 等. 城市复杂环境中爆炸波的传播规律与超压荷载 [J]. 工程力学, 2009, 26(6): 178–183. LI Z X, SHI Y C, ZHOU H Z, et al. Propagation law and overpressure load of blast wave in urban complex environment [J]. Engineering Mechanics, 2009, 26(6): 178–183.
[6]	王栋. 拱形地下结构抗爆性能试验研究 [D]. 北京: 清华大学, 2022. DOI: 10.27266/d.cnki.gqhau.2022.000098. WANG D. Experimental study on anti-explosion performance of arched underground structure [D]. Beijing: Tsinghua University, 2022. DOI: 10.27266/d.cnki.gqhau.2022.000098.
[7]	ZHOU X S, ZHANG X M, REN T H, et al. Waveform characteristics of tunnel blast waves and a wave-blocking method [J]. Shock and Vibration, 2022, 2022: 3013130. DOI: 10.1155/2022/3013130.
[8]	KRÁLIK E J, BARAN M. Numerical analysis of the exterior explosion effects on the buildings with barriers [J]. Applied Mechanics and Materials, 2013, 390: 230–234. DOI: 10.4028/www.scientific.net/AMM.390.230.
[9]	ZHANG M T, PEI Y, YAO X, et al. Damage assessment of aircraft wing subjected to blast wave with finite element method and artificial neural network tool [J]. Defence Technology, 2023, 25: 203–219. DOI: 10.1016/j.dt.2022.05.010.
[10]	严国建, 周明安, 余轮, 等. 空气中爆炸冲击波超压峰值的预测 [J]. 采矿技术, 2011, 11(5): 89–90,112. DOI: 10.3969/j.issn.1671-2900.2011.05.035. YAN G J, ZHOU M A, YU L, et al. Prediction of peak overpressure of explosion shock wave in air [J]. Mining Technology, 2011, 11(5): 89–90,112. DOI: 10.3969/j.issn.1671-2900.2011.05.035.
[11]	ZHANG K, ZHANG K, BAO R. Prediction of gas explosion pressures: a machine learning algorithm based on KPCA and an optimized LSSVM [J]. Journal of Loss Prevention in the Process Industries, 2023, 83: 105082. DOI: 10.1016/j.jlp.2023.105082.
[12]	潘美霖, 彭卫文, 冷春江, 等. 基于贝叶斯深度学习的复杂结构爆炸载荷的快速估计 [J/OL]. 爆炸与冲击, [2024-12-02]. https://www.bzycj.cn/cn/article/doi/ 10.11883/bzycj-2024-0191. DOI: 10.11883/bzycj-2024-0191. PAN M L, PENG W W, LENG C J, et al. Fast estimation of blast loading in complex structures based on Bayesian deep learning [J/OL]. Explosion and Shock Waves, [2024-12-02]. https://www.bzycj.cn/cn/article/doi/ 10.11883/bzycj-2024-0191. DOI: 10.11883/bzycj-2024-0191.
[13]	陈皓, 郭明明, 田野, 等. 卷积神经网络在流场重构研究中的进展 [J]. 力学学报, 2022, 54(9): 2343–2360. DOI: 10.6052/0459-1879-22-130. CHEN H, GUO M M, TIAN Y, et al. Progress of convolution neural networks in flow field reconstruction [J]. Chinese Journal of Theoretical and Applied Mechanics, 2022, 54(9): 2343–2360. DOI: 10.6052/0459-1879-22-130.
[14]	WANG Z Q, HUA Y, AUBRY N, et al. Fast optimization of multichip modules using deep learning coupled with Bayesian method [J]. International Communications in Heat and Mass Transfer, 2023, 141: 106592. DOI: 10.1016/j.icheatmasstransfer.2022.106592.
[15]	HUA Y, WANG Z Q, YUAN X Y, et al. Estimation of steady-state temperature field in Multichip Modules using deep convolutional neural network [J]. Thermal Science and Engineering Progress, 2023, 40: 101755. DOI: 10.1016/j.tsep.2023.101755.
[16]	闫盼盼, 牛青林, 高文强, 等. 基于卷积神经网络的地面尾喷焰流场预测 [J]. 力学学报, 2024, 56(4): 980–990. DOI: 10.6052/0459-1879-23-412. YAN P P, NIU Q L, GAO W Q, et al. Prediction of ground rocket exhaust plume flow field based on convolutional neural network [J]. Chinese Journal of Theoretical and Applied Mechanics, 2024, 56(4): 980–990. DOI: 10.6052/0459-1879-23-412.
[17]	吴昊恺, 陈耀然, 周岱, 等. 基于CNN与GAN深度学习模型近壁面湍流场超分辨率重构的精细化研究 [J]. 力学学报, 2024, 56(8): 2231–2242. DOI: 10.6052/0459-1879-24-019. WU H K, CHEN Y R, ZHOU D, et al. Refined study of super-resolution reconstruction of near-wall turbulence field based on CNN and GAN deep learning model [J]. Chinese Journal of Theoretical and Applied Mechanics, 2024, 56(8): 2231–2242. DOI: 10.6052/0459-1879-24-019.
[18]	PFAFF T, FORTUNATO M, SANCHEZ-GONZALEZ A, et al. Learning mesh-based simulation with graph networks [C]//International Conference on Learning Representations. Vienna: OpenReview. net, 2021. DOI: 10.48550/arXiv.2010.03409.
[19]	KANG M A, PARK C H. Prediction of peak pressure by blast wave propagation between buildings using a conditional 3D convolutional neural network [J]. IEEE Access, 2023, 11: 26114–26124. DOI: 10.1109/ACCESS.2023.3257345.
[20]	PENG J Z, AUBRY N, LI Y B, et al. Physics-informed graph convolutional neural network for modeling geometry-adaptive steady-state natural convection [J]. International Journal of Heat and Mass Transfer, 2023, 216: 124593. DOI: 10.1016/j.ijheatmasstransfer.2023.124593.
[21]	SHAO X Q, LIU Z J, ZHANG S Q, et al. PIGNN-CFD: a physics-informed graph neural network for rapid predicting urban wind field defined on unstructured mesh [J]. Building and Environment, 2023, 232: 110056. DOI: 10.1016/j.buildenv.2023.110056.
[22]	郝祎琛, 谢心喻, 丁家琦, 等. 瞬态多相流场图神经网络时空预测方法研究 [J]. 哈尔滨工程大学学报, 2024, 45(9): 1761–1769. DOI: 10.11990/jheu.202207004. HAO Y C, XIE X Y, DING J Q, et al. Spatiotemporal prediction method for the transient multiphase flow field via graph neural network [J]. Journal of Harbin Engineering University, 2024, 45(9): 1761–1769. DOI: 10.11990/jheu.202207004.
[23]	LI Q L, WANG Y, CHEN W S, et al. Machine learning prediction of BLEVE loading with graph neural networks [J]. Reliability Engineering & System Safety, 2024, 241: 109639. DOI: 10.1016/j.ress.2023.109639.
[24]	CULLIS I G, NIKIFORAKIS N, FRANKL P, et al. Simulating geometrically complex blast scenarios [J]. Defence Technology, 2016, 12(2): 134–146. DOI: 10.1016/j.dt.2016.01.005.
[25]	IOFFE S, SZEGEDY C. Batch normalization: accelerating deep network training by reducing internal covariate shift [C]//Proceedings of the 32nd International Conference on International Conference on Machine Learning. Lille: JMLR. org, 2015: 448–456. DOI: 10.48550/arXiv.1502.03167.
[26]	LI T Y, ZOU S F, CHANG X H, et al. Predicting unsteady incompressible fluid dynamics with finite volume informed neural network [J]. Physics of Fluids, 2024, 36(4): 043601. DOI: 10.1063/5.0197425.
[27]	FEY M, LENSSEN J E. Fast graph representation learning with PyTorch geometric [EB/OL]. arXiv: 1903.02428, 2019. [2024-12-02]. https://arxiv.org/abs/1903.02428. DOI: 10.48550/arXiv.1903.02428.
[28]	CHANG D T. Tiered graph autoencoders with PyTorch geometric for molecular graphs [EB/OL]. arXiv: 1908.08612, 2019. [2024-12-02]. https://arxiv.org/abs/1908.08612. DOI: 10.48550/arXiv.1908.08612.
[29]	LI S, ZHAO Y L, VARMA R, et al. PyTorch distributed: experiences on accelerating data parallel training [J]. Proceedings of the VLDB Endowment, 2020, 13(12): 3005–3018. DOI: 10.14778/3415478.3415530.
[30]	姚成宝, 王宏亮, 浦锡锋, 等. 空中强爆炸冲击波地面反射规律数值模拟研究 [J]. 爆炸与冲击, 2019, 39(11): 112201. DOI: 10.11883/bzycj-2018-0287. YAO C B, WANG H L, PU X F, et al. Numerical simulation of intense blast wave reflected on rigid ground [J]. Explosion and Shock Waves, 2019, 39(11): 112201. DOI: 10.11883/bzycj-2018-0287.
[31]	FEJES P, HORKAI A. Creating city models in ArchiCAD software environment [J]. The International Journal of Engineering and Science, 2021, 10(1): 11–17. DOI: 10.9790/1813-1001011117.
[32]	阎石, 刘蕾, 齐宝欣, 等. 爆炸荷载作用下方钢管混凝土柱的动力响应及破坏机理 [J]. 防灾减灾工程学报, 2011, 31(5): 477–482. DOI: 10.3969/j.issn.1672-2132.2011.05.002. YAN S, LIU L, QI B X, et al. Dynamic response and failure mode analysis of concrete infilled rectangular steel tube columns under blasting loading [J]. Journal of Disaster Prevention and Mitigation Engineering, 2011, 31(5): 477–482. DOI: 10.3969/j.issn.1672-2132.2011.05.002.
[33]	焦燏烽, 赵果, 侯延利. 框架柱在爆炸冲击荷载作用下动力系数及破坏模式研究 [J]. 建筑科学, 2015, 31(9): 32–37. DOI: 10.13614/j.cnki.11-1962/tu.2015.09.006. JIAO Y F, ZHAO G, HOU Y L. Dynamic factor and failure modes research of the frame column under blast pressure [J]. Building Science, 2015, 31(9): 32–37. DOI: 10.13614/j.cnki.11-1962/tu.2015.09.006.
[34]	PRUGH R W. The effects of explosive blast on structures and personnel [J]. Process Safety Progress, 1999, 18(1): 5–16. DOI: 10.1002/prs.680180104.