Page 71 - 《振动工程学报》2025年第9期
P. 71
第 9 期 李 扬,等:数据驱动的非高斯随机过程模拟 2001
表 2 超越概率误差对比 [7] WINTERSTEIN S R. Nonlinear vibration models for extremes
and fatigue[J]. Journal of Engineering Mechanics, 1988,
Tab. 2 Comparison of exceedance probability errors
114(10):1772-1790.
方法 超越概率误差 [8] 何桢,吴静,孔祥芬. 基于 Johnson 转换的非正态过程能力
本文方法 0.0088 研究 [J]. 组合机床与自动化加工技术,2007(8):104-107.
传统多项式转换模型 0.0227 HE Zhen, WU Jing, KONG Xiangfen. A study of non-
normal process capability using Johnson transformation
统一多项式转换模型 0.0155
system[J]. Modular Machine Tool & Automatic Manufactur-
ing Technique,2007(8):104-107.
4 结 论 [9] GRIGORIU M. Simulation of stationary non-Gaussian transla-
tion processes[J]. Journal of Engineering Mechanics, 1998,
124(2):121-126.
本文提出了一种数据驱动的非高斯随机过程模 [10] XIAO Q. Evaluating correlation coefficient for Nataf transfor-
拟方法,解决了确定样本转换关系和潜在高斯功率 mation[J]. Probabilistic Engineering Mechanics,2014,37:1-6.
[11] LI Z L,LIU R Y,FAN W L,et al. Simulation of stationary
谱这两个关键问题,能够基于样本数据实现非高斯
Gaussian/non-Gaussian stochastic processes based on stochas-
随机过程模拟。采用数值算例和工程实例,对本文 tic harmonic functions[J]. Probabilistic Engineering Mechan-
方法进行了验证,并得出以下结论: ics,2021,66:103141.
[12] 毛健,赵红东,姚婧婧. 人工神经网络的发展及应用 [J]. 电
(1)样本转换模型的优势。数据驱动构建的样 子设计工程,2011,19(24):62-65.
本转换模型具备精确、高效和广泛适用的优势,能 MAO Jian, ZHAO Hongdong, YAO Jinjging. Application
够准确地将高斯样本转换为非高斯样本,相对于多 and prospect of artificial neural network[J]. Electronic Design
Engineering,2011,19(24):62-65.
项式转换模型具有显著优势。 [13] 刘章军,刘增辉. 随机脉动风场的谱表示降维模拟 [J]. 振动
(2)潜在高斯功率谱的有效性。直接通过反向 工程学报,2018,31(1):49-56.
LIU Zhangjun, LIU Zenghui. Dimension reduction based
传播网络获得的潜在高斯功率谱证明了其有效性,
spectral representation of stochastic fluctuation wind field
避免了繁琐耗时的求解过程。 simulation[J]. Journal of Vibration Engineering, 2018, 31
(3)通过对比样本的统计矩、累积概率和超越概 (1):49-56.
[14] SHINOZUKA M, JAN C M. Digital simulation of random
率,可以发现:本文方法生成的样本具有参考数据的 processes and its applications[J]. Journal of Sound and Vibra-
概率特征,其非高斯特性与数据保持一致,同时样本 tion,1972,25(1):111-128.
[15] RIPLEY B D. Neural networks and related methods for classi-
的频域信息与目标功率谱吻合良好。这验证了本文
fication[J]. Journal of the Royal Statistical Society:Series B
方法的有效性和适用性,并展示了其在精度、效率 (Methodological),1994,56(3):409-437.
和适用性等方面相对于多项式转换模型的优越性。 [16] XU J, DANG C. A new bivariate dimension reduction
method for efficient structural reliability analysis[J]. Mechani-
cal Systems and Signal Processing,2019,115:281-300.
参考文献: [17] 罗成汉. 基于 MATLAB 神经网络工具箱的 BP 网络实现
[J]. 计算机仿真,2004,21(5):109-111.
LUO Chenghan. Realization of BP network based on neual
[1] 辛亚兵,刘志文,邵旭东,等. 山区地形实测风非平稳特
network tool kit in MATLAB[J]. Computer Simulation,
性和非高斯特性分析 [J]. 振动与冲击,2018,37(21):
2004,21(5):109-111.
247-252.
[18] ZHAO Z Z,XIN H P,REN Y Q,et al. Application and
XIN Yabing, LIU Zhiwen, SHAO Xudong, et al. Non-
comparison of BP neural network algorithm in MATLAB[C]//
stationary and non-Gaussian features of mountain terrain
measured wind[J]. Journal of Vibration and Shock, 2018, Proceedings of International Conference on Measuring Techno-
logy and Mechatronics Automation. IEEE,2010,1:590-593.
37(21):247-252. [19] KAIMAL J C,WYNGAARD J C,IZUMI Y,et al. Spec-
[2] GARCIA-ROMEU-MARTINEZ M A ,ROUILLARD V. On
tral characteristics of surface-layer turbulence[J]. Quarterly
the statistical distribution of road vehicle vibrations[J]. Packag-
Journal of the Royal Meteorological Society, 1972,
ing Technology and Science,2011,24(8):451-467.
[3] SHIELDS M D, DEODATIS G, BOCCHINI P. A simple 98(417):563-589.
[20] WINTERSTEIN S R,KASHEF T. Moment-based load and
and efficient methodology to approximate a general non-Gaus-
response models with wind engineering applications[J]. Jour-
sian stationary stochastic process by a translation process[J].
nal of Solar Energy Engineering,2000,122(3):122-128.
Probabilistic Engineering Mechanics,2011,26(4):511-519. [21] ZHANG X Y,ZHAO Y G,LU Z H. Unified Hermite poly-
[4] PHOON K K, HUANG S P, QUEK S T. Simulation of
nomial model and its application in estimating non-Gaussian
second-order processes using Karhunen-Loeve expansion[J].
processes[J]. Journal of Engineering Mechanics, 2019,
Computers & Structures,2002,80(12):1049-1060.
[5] SENGUPTA D, KAY S. Efficient estimation of parameters 145(3):04019001.
for non-Gaussian autoregressive processes[J]. IEEE Transac-
tions on Acoustics,Speech,and Signal Processing,1989, 第一作者:李 扬(1993—),男,博士研究生。
37(6):785-794.
E-mail:lyang@hnu.edu.cn
[6] LAHCENE B. On Pearson families of distributions and its
通信作者:徐 军(1986—),男,博士,教授。
applications[J]. African Journal of Mathematics and Computer
Science Research,2013,6(5):108-117. E-mail:xujun86@hnu.edu.cn