Reconstruction of Excitation Signals for Vibration Environment Simulation Tests of Projectile Road Transportation

doi:10.3969/j.issn.1004-132X.2026.08.025

Abstract

Abstract:

Vibration environment simulation tests for missile road transportation are essential for equipment development and reliability verification. Conventional full-vehicle simulation tests， however， suffer from long development cycles and difficulties in obtaining realistic excitation signals， which often leads to inadequate reproduction of the complex coupled vibration environment. To address this issue， this study proposes a data-driven excitation signal reconstruction method and compares its accuracy with that of a model-driven approach as the benchmark. First， a long short-term memory （LSTM） neural network model optimized by the quantum-behaved particle swarm optimization （QPSO） algorithm （QPSO-LSTM） is constructed to achieve high precision signal reconstruction. Second， a rigid-flexible coupled model of the “transport vehicle-packaging box-missile body-missile components” is established based on multi-body dynamics theory to obtain the model-driven reconstructed signal. Finally， the reconstruction accuracy of both methods is compared through simulation tests under different road surfaces and vehicle speeds. The experimental results indicate that the data-driven method achieves an average root-mean-square relative error of 1.54%， which is significantly lower than the 5.15% obtained by the model-driven method. This performance advantage demonstrates the superiority of the data-driven method in engineering practice.

Key words: projectile transportation, signal reconstruction, data-driven, model-driven, neural network

摘要：

导弹道路运输振动环境模拟试验对装备研发及可靠性验证至关重要。传统整车模拟试验存在周期长、真实激励信号获取困难等问题，导致复杂耦合振动环境难以有效复现。为此，提出一种数据驱动的激励信号重构方法，并以模型驱动方法为基准框架进行精度对比验证。首先构建量子粒子群算法（QPSO）优化的长短期记忆（LSTM）神经网络模型（QPSO-LSTM），实现高精度激励信号重构。然后基于多体动力学理论构建“运输车辆-包装箱-弹体-导弹部件”刚柔耦合模型，获取模型驱动重构信号。最后通过不同路面、车速模拟试验对比分析重构精度。试验结果表明，数据驱动方法的均方根相对误差平均值（1.54%）低于模型驱动方法的相应值（5.15%）。这一性能优势凸显了数据驱动方法在工程实践中的优越性。

关键词: 弹体运输, 信号重构, 数据驱动, 模型驱动, 神经网络

CLC Number:

TJ06

LING Qihui, YAN Yongyong, LI Xinyu, DAI Juchuan, XU Xiaoqiang. Reconstruction of Excitation Signals for Vibration Environment Simulation Tests of Projectile Road Transportation[J]. China Mechanical Engineering, 2026, 37(8): 2058-2067.

凌启辉, 严勇勇, 李新宇, 戴巨川, 胥小强. 导弹道路运输振动环境模拟试验激励信号重构[J]. 中国机械工程, 2026, 37(8): 2058-2067.

Figures/Tables 17

Fig.1 Research route

Fig.2 Experimental platform

Tab.1 Test conditions

试验工况	试验项目	车速或激振频率
1	戈壁路运输模拟	10 km/h
2	铺面路运输模拟	10 km/h
3	起伏路运输模拟	10 km/h
4	砂石路运输模拟	10 km/h
5	铺面路运输模拟	40 km/h
6	随机路谱试验
7	正弦扫频激励试验	1～20 Hz

Fig.3 QPSO-LSTM network model

Fig.4 Particle fitness curve

Fig.5 Prediction results of training at different locations

Fig.6 Comparison of prediction results for different positions of various models

Tab.2 Comparison of prediction accuracy among various models

预测模型	预测目标	RMSE值	MAE值	R²
LSTM	y_d1	4.02	2.49	0.727
LSTM	y_d2	5.01	3.16	0.664
PSO-LSTM	y_d1	1.71	1.45	0.940
PSO-LSTM	y_d2	1.89	1.31	0.933
QPSO-LSTM	y_d1	1.69	1.16	0.951
QPSO-LSTM	y_d2	1.79	1.21	0.956

Fig.7 Coupled dynamic model of transportation vehicle packaging box missile body missile components

Tab.3 Parameters of rigid flexible coupling dynamic model

导弹部件和支撑	弹体和支撑	运输车辆
m_b1=5.35 kg	L₁=0.68 m	m_c=25.68 kg
m_b2=6.15 kg	L₂=0.37 m	I_c=9.53 kg · m²
m_b3=9.35 kg	L₃=0.71 m	m_d=500 kg
m_b4=4.23 kg	x_a11=0.09 m	I_d=2.9 $×$ 10³ kg · m²
k_b1=3.97 $×$ 10⁵ N/m	x_a12=0.43 m	m_ti=32.5 kg
k_b2=9.76 $×$ 10⁷ N/m	x_a21=0.16 m	k_cdi=1.9 $M$ N/m
k_b3=9.36 $×$ 10⁷ N/m	x_a31=0.19 m	c_cdi=2 $00$ N·s/m
k_b4=7.1 $0 ×$ 10⁷ N/m	x_a32=0.66 m	k_si=1.243 $×$ 10⁵ N/m
c_b1=1.5 $k$ N/（m·s）	c₁=0.73 m	c_si=3.6 $k$ N·s/m
c_b2=2.5 $k$ N/（m·s）	c₂=0.23 m	k_ti=6 $00 k$ N/m
c_b3=3 $k$ N/（m·s）	c₃=0.04 m	e₁=0.23 m
c_b4=1.7 $k$ N/（m·s）	c₄=0.38 m	e₂=0.38 m
x_b11=0.48 m	c₅=0.84 m	d₁=1.22 m
x_b21=0.22 m		d₂=1.83 m
x_b31=0.21 m		s₁=0.9 m
x_b32=0.56 m		s₂=1.8 m

Tab.3 Parameters of rigid flexible coupling dynamic model

导弹部件和支撑	弹体和支撑	运输车辆
m_b1=5.35 kg	L₁=0.68 m	m_c=25.68 kg
m_b2=6.15 kg	L₂=0.37 m	I_c=9.53 kg · m²
m_b3=9.35 kg	L₃=0.71 m	m_d=500 kg
m_b4=4.23 kg	x_a11=0.09 m	I_d=2.9 $×$ 10³ kg · m²
k_b1=3.97 $×$ 10⁵ N/m	x_a12=0.43 m	m_ti=32.5 kg
k_b2=9.76 $×$ 10⁷ N/m	x_a21=0.16 m	k_cdi=1.9 $M$ N/m
k_b3=9.36 $×$ 10⁷ N/m	x_a31=0.19 m	c_cdi=2 $00$ N·s/m
k_b4=7.1 $0 ×$ 10⁷ N/m	x_a32=0.66 m	k_si=1.243 $×$ 10⁵ N/m
c_b1=1.5 $k$ N/（m·s）	c₁=0.73 m	c_si=3.6 $k$ N·s/m
c_b2=2.5 $k$ N/（m·s）	c₂=0.23 m	k_ti=6 $00 k$ N/m
c_b3=3 $k$ N/（m·s）	c₃=0.04 m	e₁=0.23 m
c_b4=1.7 $k$ N/（m·s）	c₄=0.38 m	e₂=0.38 m
x_b11=0.48 m	c₅=0.84 m	d₁=1.22 m
x_b21=0.22 m		d₂=1.83 m
x_b31=0.21 m		s₁=0.9 m
x_b32=0.56 m		s₂=1.8 m

Fig.8 Comparison of yd1 amplitude frequency curve

Fig.9 Comparison of yd2 amplitude frequency curve

Fig.10 Comparison of yd1 at different frequencies

Fig.11 Comparison of yd2 at different frequencies

Fig.12 Comparison of displacement at different positions on 10km/h undulating roads

Fig.13 Comparison of power spectra at different positions on 10 km/h undulating roads

Tab.4 Comparison of root mean square displacement values under various operating conditions

试验工况	测点位置	实测数据	模型驱动	模型驱动相对误差/%	数据驱动	数据驱动相对误差/%
1	y_d1	3.40	3.49	2.65	3.35	1.47
1	y_d2	3.90	4.07	4.36	3.79	2.82
2	y_d1	1.81	1.80	0.55	1.79	1.10
2	y_d2	1.78	1.82	2.25	1.81	1.69
3	y_d1	4.76	5.02	5.46	4.81	1.05
3	y_d2	5.46	5.80	6.22	5.50	0.73
4	y_d1	6.29	6.92	10.02	6.44	2.38
4	y_d2	7.02	7.87	12.11	6.93	1.28
5	y_d1	4.51	4.59	1.77	4.59	1.77
5	y_d2	5.57	5.91	6.10	5.63	1.08
均值				5.15		1.54

References 25

[1]	商霖，张海瑞，李璞. 导弹前定向件公路机动运输疲劳载荷谱的编制及寿命预估［J］. 振动与冲击， 2019， 38（9）： 88-93.
	SHANG Lin， ZHANG Hairui， LI Pu. Compilation of Fatigue Load Spectrum and Life Estimation for Missile Front-sliders in Road Motorized Transport Environment［J］. Journal of Vibration and Shock， 2019， 38（9）： 88-93.
[2]	杨森，吴斌，郑旸. 机械式模拟运输台仿真研究［J］. 装备环境工程， 2012， 9（5）： 106-111.
	YANG Sen， WU Bin， ZHENG Yang. Simulation Research of the Mechanical Simulation of Transport Vibration Test Stand［J］. Equipment Environmental Engineering， 2012， 9（5）： 106-111.
[3]	刘新佳，郭迅. 空空导弹运输振动试验方法研究［J］. 航空科学技术， 2020， 31（4）： 54-58.
	LIU Xinjia， GUO Xun. Study on the Test Method of Airborne Missile Transportation Vibration［J］. Aeronautical Science and Technology， 2020， 31（4）： 54-58.
[4]	GÜMÜŞ M S， KALYONCU M. Fatigue Failure of High Strength Steels under Extreme Vibrations of Military Standards： a Comparative Study［J］. Journal of Mechanical Science and Technology， 2024， 38（3）： 1059-1068.
[5]	Tiegang LÜ， ZHANG Xiaowen， QIAO Junfu， et al. Theoretical Research on Dynamic Modeling of Rigid-flexible Coupling System with Double Joint Folded Wing［J］. AIP Advances， 2024， 14（4）： 045002.
[6]	于永利，张国伟，焦明亮，等. 发射架体带弹运输环境下的安全性评估及提升方法［J］. 兵器装备工程学报， 2023， 44（）： 182-186.
	YU Yongli， ZHANG Guowei， JIAO Mingliang， et al. Safety Assessment and Lifting Methods of a Launcher Body in a Projectile Transport Environment［J］. Journal of Ordnance Equipment Engineering， 2023， 44（S1）： 182-186.
[7]	TIEGANG L， GUOGUANG C， SHUAI L. Kinematic Characteristics of Longitudinal Double Folding Wings［J］. The Aeronautical Journal， 2021， 125（1293）： 1977-2001.
[8]	黄大伟，陈立昆，高亚东. 基于LSTM和响应分解的冲击载荷识别方法研究［J］. 振动与冲击， 2024， 43（3）： 69-76.
	HUANG Dawei， CHEN Likun， GAO Yadong. Impact Load Identification Method Based on LSTM and Response Decomposition［J］. Journal of Vibration and Shock， 2024， 43（3）： 69-76.
[9]	ZHOU J M， DONG Longlei， GUAN Wei， et al. Impact Load Identification of Nonlinear Structures Using Deep Recurrent Neural Network［J］. Mechanical Systems and Signal Processing， 2019， 133： 106292.
[10]	王光定，陶文斌，纵兆生，等. 某军用包装箱动力学仿真与试验研究［J］. 包装工程， 2019， 40（23）： 117-123.
	WANG Guangding， TAO Wenbin， ZONG Zhaosheng， et al. Dynamic Simulation and Experiment on Plastics Ammunition Package［J］. Packaging Engineering， 2019， 40（23）： 117-123.
[11]	WANG Lu， FANG Qiansheng， GAO Lifu， et al. Research on Load Excitation Identification Method of Multi-connected Air Conditioning Compressor Based on RBF Network with Multi-strategy Fusion SSA［J］. International Journal of Machine Learning and Cybernetics， 2024， 15（11）： 5185-5198.
[12]	韩庆，房强，范伟光，等. 特种车辆多维载荷谱加速方法及在振动环境模拟试验中的应用［J］. 装备环境工程， 2024， 21（6）： 45-52.
	HAN Qing， FANG Qiang， FAN Weiguang， et al. Multi-dimensional Load Signals Acceleration Method for Special Vehicles and Its Application in Vibration Environment Simulation Test［J］. Equipment Environmental Engineering， 2024， 21（6）： 45-52.
[13]	CHOU Y T， KUO P C， LI Kuangyao， et al. Inductive Graph-based Long Short-term Memory Network for the Prediction of Nonlinear Floor Responses and Member Forces of Steel Buildings Subjected to Orthogonal Horizontal Ground Motions［J］. Earthquake Engineering & Structural Dynamics， 2025， 54（2）： 491-507.
[14]	郭俊锋，胡婧怡，王智明. 基于压缩感知的缺失机械振动信号重构新方法［J］. 振动与冲击， 2024， 43（10）： 197-204.
	GUO Junfeng， HU Jingyi， WANG Zhiming. Novel Method for Missing Mechanical Vibration Signal Reconstruction Based on Compressed Sensing［J］. Journal of Vibration and Shock， 2024， 43（10）： 197-204.
[15]	ZHANG Ruiyang， CHEN Zhao， CHEN Su， et al. Deep Long Short-term Memory Networks for Nonlinear Structural Seismic Response Prediction［J］. Computers & Structures， 2019， 220： 55-68.
[16]	WEI Biao， ZHENG Xianglin， JIANG Lizhong， et al. Seismic Response Prediction and Fragility Assessment of High-speed Railway Bridges Using Machine Learning Technology［J］. Structures， 2024， 66： 106845.
[17]	WANG Haifeng， WU Teng. Knowledge-enhanced Deep Learning for Wind-induced Nonlinear Structural Dynamic Analysis［J］. Journal of Structural Engineering， 2020， 146（11）： 04020235.
[18]	ZHANG Bo， HU Ziang， WU Pin， et al. EPT： a Data-driven Transformer Model for Earthquake Prediction［J］. Engineering Applications of Artificial Intelligence， 2023， 123： 106176.
[19]	HANIFI S， LIU Xiaolei， LIN Zi， et al. A Critical Review of Wind Power Forecasting Methods—Past， Present and Future［J］. Energies， 2020， 13（15）： 3764.
[20]	LI H L， WANG T Y， YAN H. Real-time Prediction of Dynamic Irregularity and Acceleration of HSR Bridges Using Modified LSGAN and In-service Train［J］. Smart Structures and Systems， 2023， 31（5）： 501-516.
[21]	毛建锋，李铮，伍军，等. 基于PSO-LSTM的重载铁路车轨桥系统随机振动响应预测方法［J］. 铁道科学与工程学报， 2024， 21（9）： 3661-3671.
	MAO Jianfeng， LI Zheng， WU Jun， et al. Predicting Random Vibration Response of Heavy-haul Train-track-bridge System Based on PSO-LSTM［J］. Journal of Railway Science and Engineering， 2024， 21（9）： 3661-3671.
[22]	ZHOU Zhiyang， JIA Baoxin， YUAN Wenya， et al. Vibration Response Prediction and Vibration Safety Assessment Method for Building Structures around Viaducts under the Action of High-speed Railway Trains［J］. Construction and Building Materials， 2024， 422： 135877.
[23]	凌启辉，张维，赵前程，等. 轧机垂直系统动力学参数辨识［J］. 钢铁， 2019， 54（11）： 123-129.
	LING Qihui， ZHANG Wei， ZHAO Qiancheng， et al. Dynamic Parameter Identification of Rolling Mill Vertical System［J］. Iron and Steel， 2019， 54（11）： 123-129.
[24]	ZHANG Jinhua， ZHAO Zhengyang， YAN Jie， et al. Ultra-short-term Wind Power Forecasting Based on CGAN-CNN-LSTM Model Supported by Lidar［J］. Sensors， 2023， 23（9）： 4369.
[25]	DENG Wanyi， MA Xiaoxue， QIAO Weiliang. A Hybrid Intelligent Optimization Algorithm Based on a Learning Strategy［J］. Mathematics， 2024， 12（16）： 2570.