Adaptive surrogate model method for extreme value prediction of driving safety indexes of train-bridge coupled random system
ZHANG Xun1, HAN Yan1, WANG Lidong1, LIU Hanyun1, CAI Chunsheng2
1.School of Civil Engineering, Changsha University of Science & Technology, Changsha 410114, China;
2.School of Transportation, Southeast University, Nanjing 211189, China
Abstract：In order to improve the efficiency of the reliability evaluation of train running safety on bridge, a method of adaptive surrogate model for predicting the extreme value of the running safety indexes of the train-bridge stochastic system is proposed in this paper. Firstly, the initial sample set and candidate sample set are generated by minimizing the GF-discrepancy; Secondly, the theoretical model is used to calculate the extreme value of safety indexes of the train-bridge system corresponding to the initial sample set, and the radial basis function surrogate model is established; Finally, the learning function and established surrogate model are used to determine the new sample in the set of the candidate sample set to optimize the current surrogate model until the surrogate model meets the satisfactory prediction accuracy. In order to verify the effectiveness of the proposed method, taking the moving wheel and sprung mass passing through simply supported beam bridge and the train-bridge vertical coupling vibration model as examples, the distribution of training sample points of the adaptive surrogate model and the one-stage sampling surrogate model, as well as the prediction accuracy of the two surrogate models, are compared and analyzed. The results show that it can find the sparse regions of sample points in the current training set and the strong nonlinearity regions of the objective function, and make the sample points in these regions more densely. Thus, the prediction accuracy of the surrogate model in the whole sample space is significantly improved without increasing the number of training samples. Taking the vertical model of the train-bridge system as an example, the prediction accuracy of the adaptive surrogate model is 2.5 times higher than that of the one-time sampling surrogate model.
张迅1，韩艳1，王力东1，刘汉云1，蔡春声2. 车-桥随机系统行车安全指标极值预测的自适应代理模型方法[J]. 振动与冲击, 2023, 42(19): 70-78.
ZHANG Xun1, HAN Yan1, WANG Lidong1, LIU Hanyun1, CAI Chunsheng2. Adaptive surrogate model method for extreme value prediction of driving safety indexes of train-bridge coupled random system. JOURNAL OF VIBRATION AND SHOCK, 2023, 42(19): 70-78.
[1] HE Xuhui, WU Teng, ZOU Yunfeng, et al. Recent developments of high-speed railway bridges in China[J]. Structure and Infrastructure Engineering, 2017, 13(12): 1584-1595.
[2] YU Zhiwu, MAO Jianfeng. A stochastic dynamic model of train-track-bridge coupled system based on probability density evolution method[J]. Applied Mathematical Modelling, 2018, 59: 205-232.
[3] ZHAI Wanming, HAN Zhaoling, CHEN Zhaowei, et al. Train–track–bridge dynamic interaction: a state-of-the-art review[J]. Vehicle System Dynamics, 2019, 57(7): 984-1027.
[4] PAPADOPOULOS V, GIOVANIS D G, LAGAROS N D, et al. Accelerated subset simulation with neural networks for reliability analysis[J]. Computer Methods in Applied Mechanics and Engineering, 2012, 223: 70-80.
[5] NIE J, ELLINGWOOD B R. Directional methods for structural reliability analysis[J]. Structural Safety, 2000, 22(3): 233-249.
[6] AU S K, BECK J L. Important sampling in high dimensions[J]. Structural Safety, 2003, 25(2): 139-163.
[7] 周子骥, 张楠, 孙琪凯. 基于虚拟激励法列车脱轨系数概率统计研究[J]. 工程力学, 2022, 39(1): 219-227.
ZHOU Ziji, ZHANG Nan, SUN Qikai. The probabity statistics of train derailment coefficients based on the pseudo-excitation method [J]. Engineering Mechanics, 2022, 39(1): 219-227.
[8] WANG Lidong, ZHANG Xun, LIU Hanyun, et al. Global reliability analysis of running safety of a train traversing a bridge under crosswinds[J]. Journal of Wind Engineering and Industrial Aerodynamics, 2022, 224: 104979.
[9] 苏成,钟春意,周立成.车桥耦合系统随机振动的时域显式解法[J].应用数学和力学,2017,38(1):99-108.
SU Cheng, ZHONG Chunyi, ZHOU Licheng. Time domain explicit solution of random vibration of vehicle bridge coupling system [J]. Applied Mathematics and Mechanics, 2017,38(1):99-108.
[10] XIANG Huoyue, TANG Ping, ZHANG Yuan, et al. Random dynamic analysis of vertical train–bridge systems under small probability by surrogate model and subset simulation with splitting[J]. Railway Engineering Science, 2020, 28(3): 305-315.
[11] 唐兆,董少迪,罗仁,等.人工智能算法在铁道车辆动力学仿真中的应用进展[J].交通运输工程学报,2021,21(1):250-266.
TANG Zhao, DONG Shaodi, LUO Ren, et al. Application advances of artificial intelligence algorithms in dynamics simulation of railway vehicle [J]. Journal of Traffic and Transportation Engineering, 2021,21(1):250-266.
[12] XIANG Zhengliang, CHEN Jiahui, BAO Yuequan, et al. An active learning method combining deep neural network and weighted sampling for structural reliability analysis[J]. Mechanical Systems and Signal Processing, 2020, 140: 106684.
[13] LIU Haitao, ONG Yewsoon, CAI Jianfei. A survey of adaptive sampling for global metamodeling in support of simulation-based complex engineering design[J]. Structural and Multidisciplinary Optimization, 2018, 57(1): 393-416.
[14] ZHOU Tong, PENG Yongbo, LI Jie. An efficient reliability method combining adaptive global metamodel and probability density evolution method[J]. Mechanical Systems and Signal Processing, 2019, 131: 592-616.
[15] ZHENG Peijuan, WANG Chien Ming, ZONG Zhouhong, et al. A new active learning method based on the learning function U of the AK-MCS reliability analysis method[J]. Engineering structures, 2017, 148: 185-194.
[16] XIAO N C, ZUO M J, ZHOU C N. A new adaptive sequential sampling method to construct surrogate models for efficient reliability analysis[J]. Reliability Engineering & System Safety, 2018, 169: 330-338.
[17] ZHOU Tong, PENG Yongbo. Active learning and active subspace enhancement for PDEM-based high-dimensional reliability analysis[J]. Structural Safety, 2021, 88: 102026.
[18] MO Shaoxing, LU Dan, SHI Xiaoqing, et al. A Taylor expansion-based adaptive design strategy for global surrogate modeling with applications in groundwater modeling[J]. Water Resources Research, 2017, 53(12): 10802-10823.
[19] HAN Xu, XIANG Huoyue, LI Yongle, et al. Predictions of vertical train-bridge response using artificial neural network-based surrogate model[J]. Advances in Structural Engineering, 2019, 22(12): 2712-2723.
[20] LI Huile, WANG Tianyu, WU Guang. A Bayesian deep learning approach for random vibration analysis of bridges subjected to vehicle dynamic interaction[J]. Mechanical Systems and Signal Processing, 2022, 170: 108799.
[21] ZHU Zhihui, GONG Wei, WANG Lidong, et al. Efficient assessment of 3D train-track-bridge interaction combining multi-time-step method and moving track technique[J]. Engineering Structures, 2019, 183: 290-302.
[22] CHEN Jianbing, YANG Junyi, LI Jie. A GF-discrepancy for point selection in stochastic seismic response analysis of structures with uncertain parameters[J]. Structural Safety, 2016, 59: 20-31.
[23] 朱志辉,王力东,龚威,等.多种垂向轮轨关系的对比及改进的车-线-桥系统迭代模型的建立[J].中南大学学报(自然科学版),2017,48(6):1585-1593.
ZHU Zhihui, WANG Lidong, GONG Wei, et al. Comparative analysis of several types of vertical wheel/rail relationship and construction of an improved iteration model for train−track−bridge system [J]. Journal of Central South University (Science and Technology),2017,48(6):1585-1593.
[24] 韩旭,向活跃,李永乐.考虑参数不确定性的列车-桥梁垂向耦合振动的PC-ARMAX代理模型研究[J].工程力学,2021,38(11):180-188.
HAN Xu, XIANG Huoyue, LI Yongle. Coupled vibration of vertical random vehicle-bridge system considering parameter uncertainty based on PC-ARMAX model [J]. Engineering Mechanics, 2021,38(11):180-188.
