Article
Engineering, Industrial
Yanzhong Wang, Bin Xie, Shiyuan E
Summary: A new reliability method, RVM-MIS, is proposed in this research by combining relevance vector machine and Markov-chain-based importance sampling. The method improves computational efficiency in evaluating the failure probability of engineering systems with complex implicit performance functions. Relevance vector machine provides predicted values and variances, and active learning functions can be applied to improve accuracy. Markov-chain-based importance sampling is used to generate important samples, which are then predicted by RVM to obtain failure probability, reducing computation time significantly.
RELIABILITY ENGINEERING & SYSTEM SAFETY
(2022)
Article
Engineering, Multidisciplinary
Bin Xie, Chong Peng, Yanzhong Wang
Summary: This paper proposes a new reliability analysis method by combining relevance vector machine and subset simulation importance sampling, which improves the efficiency and accuracy of evaluating the failure probability of engineering structures involving implicit performance functions.
APPLIED MATHEMATICAL MODELLING
(2023)
Article
Computer Science, Interdisciplinary Applications
Chunyan Ling, Zhenzhou Lu
Summary: The proposed method introduces a novel two-stage meta-model importance sampling based on support vector machine (SVM) to efficiently estimate structural failure probability. It provides an algorithm to efficiently deal with multiple failure regions and rare events, with the SVM model accurately recognizing the states of samples. Several examples are performed to show the feasibility of the proposed method.
STRUCTURAL AND MULTIDISCIPLINARY OPTIMIZATION
(2021)
Article
Computer Science, Artificial Intelligence
Ning Wei, Zhenzhou Lu, Yingshi Hu
Summary: By introducing a Beta-hypersphere in traditional Radial-Based Importance Sampling (RBIS) method, efficiency of reliability analysis can be improved by avoiding the evaluation of Limit State Function (LSF) in safety samples located in the Beta-hypersphere. However, evaluation of LSF in safety samples outside the Beta-hypersphere is still necessary. To further enhance the efficiency, an Eccentric RBIS (ERBIS) method with an eccentric hypersphere is proposed, which can envelop more safety samples and avoid excess evaluation of LSF. The ERBIS method is demonstrated to significantly reduce LSF evaluations compared to the RBIS method.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Jie Tian, Mingdong Hou, Hongli Bian, Junqing Li
Summary: Due to the curse of dimensionality, applying surrogate-assisted evaluation algorithms (SAEAs) to high-dimensional expensive problems remains challenging. This paper proposes a variable surrogate model-based particle swarm optimization (VSMPSO) and extends it to solve 200-dimensional problems. By constructing a single surrogate model through simple random sampling, different promising areas are explored in different iterations. The variable model management strategy is used to better utilize the current global model and accelerate the convergence rate of the optimizer. Comparisons with state-of-the-art algorithms demonstrate that VSMPSO achieves high-quality solutions and computational efficiency for high-dimensional problems.
COMPLEX & INTELLIGENT SYSTEMS
(2023)
Article
Computer Science, Interdisciplinary Applications
Fen Li, Zhenzhou Lu, Kaixuan Feng, Xia Jiang
Summary: This article proposes an importance sampling-based algorithm for small failure probability problems. By using standard normal transformation and establishing an augmented performance function, the efficiency of estimating failure probability function and failure chance index is improved. Additionally, the efficiency is further enhanced by combining adaptive Kriging model and reduction of candidate sample pool.
STRUCTURAL AND MULTIDISCIPLINARY OPTIMIZATION
(2022)
Article
Agronomy
Gunnar Lischeid, Heidi Webber, Michael Sommer, Claas Nendel, Frank Ewert
Summary: This study utilized machine learning methods to investigate the impact of climatic and soil hydrological factors on the yield of four crops, highlighting the uniqueness of key predictors. Random Forest and Support Vector Machine models achieved between 50% and 70% capture of spatial and temporal variance, with different sets of predictors performing similarly. In light of climate change, excess precipitation and heat effects are seen as important factors in crop breeding and modeling.
AGRICULTURAL AND FOREST METEOROLOGY
(2022)
Article
Engineering, Multidisciplinary
Mohsen Rashki
Summary: This study introduces a new probability concept, random probability density function (PDF), as an efficient alternative to random sampling for probability/reliability analysis of multivariate problems. By drawing a few random PDFs instead of millions of random samples, accurate and efficient statistical moment estimation and reliability analysis of multivariable problems can be achieved.
APPLIED MATHEMATICAL MODELLING
(2021)
Article
Computer Science, Interdisciplinary Applications
Bin Xie, Yanzhong Wang, Yunyi Zhu, Fengxia Lu
Summary: This paper proposes a novel reliability analysis method, RVM-SS, which combines relevance vector machine (RVM) and subset simulation (SS). It improves the efficiency and accuracy of reliability analysis by using RVM to approximate limit states and performing SS based on the constructed RVM. The updated RVM has high prediction accuracy, resulting in accurate failure probability estimation.
STRUCTURAL AND MULTIDISCIPLINARY OPTIMIZATION
(2023)
Article
Construction & Building Technology
Yongxin Wu, Hao Bao, Juncheng Wang, Yufeng Gao
Summary: Accurate prediction of tunnel convergence is crucial for ensuring the stability of tunnel structures. This study conducted reliability analysis based on Monte Carlo simulations, showing that the Young's modulus of soils varies spatially and discussing the influence of various factors on tunnel convergence. The results highlighted the significant impact of distribution type on the outcomes, providing a detailed analysis of tunnel convergence mechanisms in spatially random soils.
TUNNELLING AND UNDERGROUND SPACE TECHNOLOGY
(2021)
Article
Engineering, Geological
Ze Zhou Wang, Changlin Xiao, Siang Huat Goh, Min-Xuan Deng
Summary: The paper proposes a novel and computationally efficient metamodeling technique using convolutional neural networks for random field finite-element analyses, showing promising potential for reliability analysis in spatially variable soils. The CNN outputs demonstrated good agreement with FEM predictions, indicating the effectiveness of using CNNs as metamodels to replace expensive simulations.
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING
(2021)
Article
Green & Sustainable Science & Technology
Jian Wang, Xiang Gao, Zhili Sun
Summary: This paper proposes an importance sampling framework for time-variant reliability analysis, which utilizes both time-invariant random variables and stochastic processes, applying time-invariant IS and crude Monte Carlo simulation simultaneously.
Article
Computer Science, Hardware & Architecture
Nan Ye, Zhenzhou Lu, Xiaobo Zhang, Kaixuan Feng
Summary: In this article, a metamodel-based directional importance sampling method (Meta-DIS-AK) is proposed to improve the efficiency of reliability analysis. The main novelty of Meta-DIS-AK lies in constructing the quasi-optimal DIS density (DIS-D) using the Kriging model and accurately extracting DIS-D samples through a simple rejection sampling algorithm. Compared with existing methods, Meta-DIS-AK overcomes the difficulties of constructing DIS-D and extracting important direction vector samples, while maintaining the advantages of DIS in dealing with high-dimensional and small failure probability problems.
IEEE TRANSACTIONS ON RELIABILITY
(2023)
Article
Engineering, Marine
Jinwan Park, Jung-Sik Jeong
Summary: This study introduces an enhanced machine learning method to estimate ship collision risk, with the relevance vector machine (RVM) showing more accurate and efficient results compared to the conventional support vector machine (SVM). By supporting more reliable decision-making for navigators through precise risk estimation, early evasive actions can be facilitated.
JOURNAL OF MARINE SCIENCE AND ENGINEERING
(2021)
Article
Engineering, Multidisciplinary
Chunyan Ling, Zhenzhou Lu
Summary: The proposed method utilizes a compound kriging-based importance sampling strategy to efficiently estimate the failure probability of systems with multiple failure modes. The algorithm involves two stages of constructing and refining kriging models to estimate the system failure probability. The system failure probability is estimated by the product of component augmented failure probabilities and a correction factor.
ENGINEERING OPTIMIZATION
(2022)