☆ 4.6 Article

A copula-based Gaussian mixture closure method for stochastic response of nonlinear dynamic systems

PROBABILISTIC ENGINEERING MECHANICS (2020)

Journal

PROBABILISTIC ENGINEERING MECHANICS

Volume 59, Issue -, Pages -

Publisher

ELSEVIER SCI LTD

DOI: 10.1016/j.probengmech.2019.103015

Keywords

Moment equation; Non-Gaussian closure; Copula; Gaussian mixture model; Random vibration

Categories

Engineering, Mechanical Mechanics Statistics & Probability

Funding

National Natural Science Foundation of China [11672091, 11972009, 51609054]

Ask authors/readers for more resources

Protocol

Community support

Reagent

Community support

Abstract

Gaussian closure method is commonly used in the analysis of nonlinear stochastic systems. However, Gaussian closure may lead to unacceptable errors when system response is very much different from being Gaussian, and accuracy of the method decreases as the nonlinearity of the system increases. The need for better accuracy in strongly non-linear problems has caused the development of non-Gaussian closure schemes. In this paper, we develop a new copula-based Gaussian mixture closure method for randomly excited nonlinear systems. Our method relies on the assumption of marginal PDF of response in terms of finite Gaussian mixture model, and the derivation of joint PDF with aid of dependence modeling of Gaussian copula. By substituting the non-Gaussian PDF representation into moment equations of nonlinear system, we further develop an optimization-based closure scheme for the solution of the unknown parameters in joint PDF. In this way, PDF and thus, moments of response of highly nonlinear system can be described in a more flexible and robust way. Effectiveness of the new closure method is demonstrated by a nonlinear and a Duffing oscillator that are subjected to Gaussian white noise. The results are compared with the Gaussian closure and exact solution. It has been shown that Gaussian closure is a special case of the new closure method, and accuracy of Gaussian closure is the lower bound of that of the new closure method.

Authors

I am an author on this paper

Click your name to claim this paper and add it to your profile.

Reviews

Primary Rating

4.6

Not enough ratings

Secondary Ratings

Novelty

-

Significance

-

Scientific rigor

-

Rate this paper

Recommended

Article Acoustics

Gaussian closure technique for a simple tank model with non-zero mean random load and elasto-plasticity

Holger Waubke, Christian H. Kasess

Summary: Gaussian closure is used to calculate the behavior of chain-like structures with hysteretic elements under white noise excitation, resulting in analytic expressions for the temporal evolution of the statistical moments. The method is generalized to the case of a filtered white noise with a slowly varying intensity, which is relevant in earthquake scenarios. The model is further complex when there is a lack of elastic restoring force at the hysteretic node that couples the tank model with the surface.

JOURNAL OF LOW FREQUENCY NOISE VIBRATION AND ACTIVE CONTROL (2023)

Add to Collection

Article Engineering, Mechanical

Stationary response analysis for a linear system under non-Gaussian random excitation by the equivalent non-Gaussian excitation method and the Hermite moment model

Takahiro Tsuchida, Koji Kimura

Summary: This paper presents an approximate analytical method to obtain the stationary response probability density and mean upcrossing rate for a linear system under non-Gaussian random excitation. The method first utilizes the equivalent non-Gaussian excitation method to derive a closed set of moment equations for the system response, and then calculates the exact solutions of response moments up to the fourth order from these equations. The proposed method demonstrates its effectiveness by comparing the analytical results with Monte Carlo simulation results for linear systems under random excitations with different probability densities.

PROBABILISTIC ENGINEERING MECHANICS (2023)

Add to Collection

Article Engineering, Multidisciplinary

Numerical solution of the Fokker-Planck equation using physics-based mixture models

Armin Tabandeh, Neetesh Sharma, Leandro Iannacone, Paolo Gardoni

Summary: This paper presents a novel numerical method based on physics-based mixture models for solving the transient and steady-state solutions of the Fokker-Planck equation. The unknown parameters of the mixture model are estimated using Bayesian inference with constraints on model parameters. An importance sampling algorithm is developed to reduce computational demand. The performance of the proposed method is demonstrated with benchmark problems.

COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING (2022)

Add to Collection

Article Engineering, Civil

Multivariate structural seismic fragility analysis and comparative study based on moment estimation surrogate model and Gaussian copula function

Yexiang Yan, Hongwei Huang, Limin Sun

Summary: This paper proposes a parameterized method for component and system-level fragility analysis through multivariate seismic fragility analysis. The study demonstrates that compared to existing methods, the proposed approach can efficiently and accurately generate multivariate fragility functions, and provides optimal choices for sample size, surrogate model, and IM selection.

ENGINEERING STRUCTURES (2022)

Add to Collection

Article Acoustics

Power exchanged between subsystems with non-diffuse fields in statistical energy analysis

V. Tyrode, N. Totaro, L. Maxit, A. Le Bot

Summary: This article discusses the necessity and validity of the assumption of diffuse field in statistical energy analysis, as well as the coupling power proportionality which states that the vibrational power exchanged between coupled subsystems is proportional to the difference of their modal energies. It proposes a reformulation of the coupling power proportionality in terms of local energy density instead of modal energy. The generalized form is shown to remain valid even if the vibrational field is not diffuse. Various causes of lack of diffuseness are studied, and numerical simulations and experimental results on flexural vibration of flat plates are provided to support the statements made.

JOURNAL OF THE ACOUSTICAL SOCIETY OF AMERICA (2023)

Add to Collection

Review Engineering, Aerospace

Stationary non-Gaussian random vibration control: A review

Ronghui Zheng, Guoping Chen, Huaihai Chen

Summary: This paper reviews various methods for controlling stationary non-Gaussian random vibration, discussing how to control time and frequency domain characteristics in non-Gaussian random vibration tests and generate a one frame stationary non-Gaussian random signal.

CHINESE JOURNAL OF AERONAUTICS (2021)

Add to Collection

Article Engineering, Multidisciplinary

The practical calculation method of the high-order moment spectra of quadratic Gaussian stochastic processes

Xiangqian Sheng, Wenliang Fan

Summary: This paper presents a practical calculation method for high-order moment spectra models of quadratic Gaussian stochastic processes. The expressions for the first four moment spectra of quadratic Gaussian stochastic processes are derived based on the relation between the high-order cumulant function and high-order moment function. The proposed method calculates the high-order moment spectra using the ratio of the corresponding component moment functions, and its efficiency and accuracy are validated through numerical examples.

APPLIED MATHEMATICAL MODELLING (2023)

Add to Collection

Article Engineering, Electrical & Electronic

Image compression based on Gaussian mixture model constrained using Markov random field

Jianjun Sun, Yan Zhao, Shigang Wang, Jian Wei

Summary: This paper introduces a Gaussian Mixture Model (GMM) constrained by Markov Random Field (MRF) framework for image compression, optimizing parameters selection with adjusted Expectation-Maximization (EM) algorithm and Mixture Model Optimization (MMO). The method encodes parameters using fixed-length bits and uses a codebook to improve efficiency, with the residual encoded using HEVC intra coding. Experimental results demonstrate superior performance compared to HEVC, JPEG 2000, and BPG.

SIGNAL PROCESSING (2021)

Add to Collection

Article Environmental Sciences

Definition of Spatial Copula Based Dependence Using a Family of Non-Gaussian Spatial Random Fields

Andras Bardossy, Sebastian Horning

Summary: The spatial structures of natural variables are often complex and exhibit non-Gaussian spatial dependence. Existing approaches to consider non-Gaussian behavior are limited. This study presents a flexible method for defining non-Gaussian spatial dependence, based on continuous deformation of fields with different Gaussian spatial dependence. The methodology is illustrated with theoretical examples and demonstrated in a real-life example of groundwater quality parameters.

WATER RESOURCES RESEARCH (2023)

Add to Collection

Article Engineering, Mechanical

The higher-order analysis method of statistics analysis for response of linear structure under stationary non-Gaussian excitation

Wenliang Fan, Xiangqian Sheng, Zhengliang Li, Yi Sun

Summary: The study presents a novel high-order statistical analysis method for effectively analyzing vibrations of structures under non-Gaussian excitation, which is more comprehensive compared to conventional methods. An analytical solution for higher-order statistics of response was derived using the mode superposition method, along with a theoretical expression for calculating higher-order moment spectrum of response.

MECHANICAL SYSTEMS AND SIGNAL PROCESSING (2022)

Add to Collection

Article Computer Science, Information Systems

Design and Analysis of an Effective Multi-Barriers Model Based on Non-Stationary Gaussian Random Fields

Zhi Li, Lei Liu, Jiaqiang Wang, Li Lin, Jichang Dong, Zhi Dong

Summary: In this paper, a Multi-Barriers Model is proposed to characterize an area of interest with different types of obstacles. The proposed model divides the area into sub-areas, one with sampling points and the other with obstacles. The correlation between points is determined by the obstruction degree of the obstacles, and stochastic partial differential equations are used to express continuous Gaussian fields. The model is verified using geostatistical and log-Gaussian Cox models, and its performance is compared with other models using real burglary data. The Multi-Barriers Model is found to better interpret spatial models with multiple obstacles and is closer to reality.

ELECTRONICS (2023)

Add to Collection

Article Engineering, Mechanical

Power spectrum and kurtosis separation method for multi-shaker non-Gaussian random vibration control

Ronghui Zheng, Guoping Chen, Huaihai Chen

Summary: An innovative power spectrum and kurtosis separation control strategy is proposed for non-Gaussian random vibration test, aiming to address the practical situation where spectrum and kurtosis control channels are not completely coincident. The feasibility of the proposed control method is verified through simulation examples and biaxial separation and non-separation non-Gaussian random vibration tests.

MECHANICAL SYSTEMS AND SIGNAL PROCESSING (2022)

Add to Collection

Article Mathematics, Applied

DIRECT FLUX GRADIENT APPROXIMATION TO MOMENT CLOSURE OF KINETIC EQUATIONS

Ruo Li, Weiming Li, Lingchao Zheng

Summary: A new closure approach is proposed in this paper, which approximates the flux gradient instead of the flux function, providing significant advantages over the canonical approach. This method allows for deriving globally hyperbolic moment models while preserving conservative variables in low order moments.

SIAM JOURNAL ON APPLIED MATHEMATICS (2021)

Add to Collection

Article Computer Science, Artificial Intelligence

An improved gaussian mixture hidden conditional random fields model for audio-based emotions classification

Muhammad Hameed Siddiqi

Summary: This study focuses on improving the emotional speech classifier by introducing a novel methodology to address the limitations of existing classifiers, achieving significant improvement in emotional recognition. The proposed method has been validated and evaluated on two datasets, showing significantly improved classification performance. In terms of computation, the technique is also more cost-effective compared to state of the art works.

EGYPTIAN INFORMATICS JOURNAL (2021)

Add to Collection

Article Computer Science, Interdisciplinary Applications

Machine learning moment closure models for the radiative transfer equation I: Directly learning a gradient based closure

Juntao Huang, Yingda Cheng, Andrew J. Christlieb, Luke F. Roberts

Summary: In this paper, a data-driven approach using machine learning is applied to solve the moment closure problem for the radiative transfer equation in slab geometry. Instead of learning the unclosed high order moment, the gradient of the high order moment is directly learned using neural networks. This new approach is consistent with the exact closure for the free streaming limit and provides a natural output normalization. Various benchmark tests confirm the accuracy and generalizability of the machine learning closure model.

JOURNAL OF COMPUTATIONAL PHYSICS (2022)

Add to Collection

No Data Available

Article Engineering, Mechanical

Surrogate-assisted investigation on influence of epistemic uncertainties on running safety of high-speed trains on bridges

R. Allahvirdizadeh, A. Andersson, R. Karoumi

Summary: The operational safety of high-speed trains on ballasted bridges relies on preventing ballast destabilization. This study explores the impact of epistemic uncertainties on the system using ISRA. Neglecting these uncertainties can lead to overestimation of permissible train speeds and reduced system safety.

PROBABILISTIC ENGINEERING MECHANICS (2024)

Add to Collection

Article Engineering, Mechanical

Two-phase optimized experimental design for fatigue limit testing

Lujie Shi, Leila Khalij, Christophe Gautrelet, Chen Shi, Denis Benasciutti

Summary: This study proposes an innovative Two-phase method based on the Langlie method and the D-optimality criterion to overcome the intrinsic shortcomings of the staircase method used in estimating the fatigue limit distribution. Through simulation-based study, it is demonstrated that the proposed method improves the estimation performance for the mean and standard deviation of the fatigue limit distribution.

PROBABILISTIC ENGINEERING MECHANICS (2024)

Add to Collection

Article Engineering, Mechanical

A comparative study of various metamodeling approaches in tunnel reliability analysis

Axay Thapa, Atin Roy, Subrata Chakraborty

Summary: This article compares different metamodeling approaches for reliability analysis of tunnels to evaluate their performance. The study found that Kriging and support vector regression models perform well in estimating the reliability of underground tunnels.

PROBABILISTIC ENGINEERING MECHANICS (2024)

Add to Collection

Article Engineering, Mechanical

A novel reliability updating based method for efficient estimation of failure-probability global sensitivity

Jiaqi Wang, Zhenzhou Lu, Lu Wang

Summary: This paper proposes an efficient method to estimate the FP-GS using reliability updating, avoiding the time-consuming double-loop structure analysis. By utilizing the likelihood function and adaptive Kriging model, the unconditional FP and all conditional FPs can be estimated simultaneously.

PROBABILISTIC ENGINEERING MECHANICS (2024)

Add to Collection

Article Engineering, Mechanical

Statistical analysis of wind load probabilistic models considering wind direction and calculation of reference wind pressure values in Liaoning Province, China

Jiaxu Li, Ming Liu, Xu Yan, Qianting Yang

Summary: Wind pressure is essential for architectural design, and this study found that using different probabilistic distribution models can improve the accuracy of reference wind pressure calculation. In the research conducted in Liaoning Province, the extreme value type III model and moment method achieved the best fit. Additionally, probability density functions for wind speed and wind direction were established for further analysis of wind pressure.

PROBABILISTIC ENGINEERING MECHANICS (2024)

Add to Collection

Article Engineering, Mechanical

Time-dependent reliability analysis of planar mechanisms considering truncated random variables and joint clearances

Yufan Cheng, Xinchen Zhuang, Tianxiang Yu

Summary: This paper proposes a time-dependent kinematic reliability analysis method that takes into account the truncated random variables and joint clearances, effectively addressing the issues of dimension variables and correlation between joint clearance variables. The proposed method transforms time-dependent reliability into time-independent reliability, greatly reducing computational complexity and obtaining upper and lower bounds of failure probability.

PROBABILISTIC ENGINEERING MECHANICS (2024)

Add to Collection

© Peeref 2019-2024. All rights reserved.