Page 43 - 《爆炸与冲击》2026年第11期

第 46 卷 潘美霖，等： 基于贝叶斯深度主动学习的三维城市建筑群爆炸载荷快速预测 第 11 期

recorded at 12 points of interest (POIs) in the building cluster, resulting in a dataset of 7 776 samples. A BDAL framework was
then developed. Bayesian inference was integrated into a deep neural network to enable probabilistic modeling of parameters.
Monte Carlo (MC) dropout was adopted as an approximate variational inference method to estimate predictive uncertainty. An
uncertainty-driven active sampling strategy was designed: the predictive variance of each unlabeled sample was computed via
30 stochastic forward passes with dropout enabled. Samples with variance exceeding 85% of the maximum variance were
selected as candidates, and the top 28 cases (336 samples) with the highest variance were chosen in each active learning cycle.
These selected samples were labeled by the blastFoam simulator and added to the training set. The model was retrained
iteratively until the relative improvement in mean absolute percentage error (MAPE) fell below 1% or the labeled set reached
the full training size. On a test set of 780 unseen samples (65 cases), the proposed BDAL method achieved an MAPE of 13.1%
2
and an R of 0.972 for peak overpressure prediction. The 95% prediction interval covered the true values in 85.9% of the cases,
with a normalized mean prediction interval width (NMPIW) of 0.026. Single-point prediction response time was below 20 ms,
5
representing a speedup of more than 10 compared to high-fidelity numerical simulations. Compared to passive deep learning
models trained on the full dataset, the BDAL method required only about 50% of labeled data to reach comparable prediction
accuracy. In a comparative experiment with 50% training data, BDAL achieved a MAPE of 17.2%, while a conventional fully
connected neural network (FCNN) and a three-dimensional direction-encoded Bayesian neural network (3D-DeBNN) gave
MAPEs of 52.9% and 22.6%, respectively. The proposed Bayesian deep active learning method enables efficient and low-cost
blast loading prediction in typical regularized urban environments. It effectively reduces the dependency on large-scale
numerical simulation data, maintains high prediction accuracy and reliable uncertainty quantification, and meanwhile achieves
millisecond-level inference speed. The method shows strong potential for disaster prevention and mitigation applications, such
as pre-disaster anti-blast design and post-disaster emergency response.
Keywords: three-dimensional typical urban building cluster; blast loading; Bayesian deep active learning; fast prediction

爆炸事件在高人口密度和建筑密集的城市环境中具有极大的破坏性 [1-3] 。快速预测城市爆炸载荷，
是评估爆炸冲击波对现代城市建筑等目标毁伤效应 [4-5] 的核心需求，对指导灾前抗爆结构设计 [6-7] 、优化
灾后应急响应 [8] 具有重要意义。然而，传统基于经验的爆炸载荷计算方法难以应用在复杂城市环境，
高精度数值模拟需进行三维显式动力学瞬态计算，资源消耗大、计算时间长。深度学习具备强大的非线
性映射与高维特征建模能力，相比数值模拟有显著的速度优势，已成为城市爆炸载荷快速预测研究的
重要方向。
基于深度学习的爆炸载荷预测研究呈现从可行性验证到性能提升的演进轨迹。在验证深度学习在
爆炸载荷预测领域的可行性方面，Remennikov 等 [9] 基于数值模拟软件生成了涵盖多种街道与建筑尺寸
的数值模拟数据，采用四层全连接神经网络（fully connected neural network, FCNN）预测了城市关键位置
的压力峰值与冲量，发现 FCNN 预测值与 CFD (computational fluid dynamics) 模拟值高度一致（决定系数 R 2
分别为 0.993 与 0.995），推理耗时达毫秒级；黄沛吉等 [10] 在缩比城市场景中基于 FCNN 对超压峰值进行
高效预测，90% 样本相对误差不超过 10%、单点预测耗时不超过 3 ms；Si 等 [11] 构建 1D-2D-3D 多尺度 ALE
(arbitrary Lagrangian-Eulerian) 仿真体系并以约 19 万条数据训练 FCNN，实现城市街区建筑表面超压峰值
与冲量的快速预测，平均相对误差分别为 1.69% 和 1.28%。这些研究证明了深度学习模型能够高效预测
城市爆炸载荷，推理速度远快于数值模拟。
随后，学者们致力于提升深度学习的爆炸载荷预测性能，研究围绕泛化、效率、成本与精度持续推
进。为提高爆炸载荷预测方法在不同场景中的泛化性，Dennis 等 [12-13] 提出方向编码神经网络（direction
encoded neural network, DeNN），将波传播路径与障碍物方向信息等物理先验嵌入模型，在未见复杂场景
中平均绝对误差约 5 kPa，该方法通过将空间信息作为模型输入，成功突破传统神经网络对预设场景的依

赖，为不同场景下的快速抗爆评估提供了通用新范式；Wang 等 [14] 提出基于图神经网络的 BlastGraphNet，
可在复杂三维建筑外表面实时预测超压峰值、冲量峰值与波阵面到达时间，在几何复杂性变化下仍保持

111503-2