Adaptive Physics-Informed Neural Network Based Directional Sampling Method for Efficient Reliability Analysis

被引:0
|
作者
Yan, Yuhua [1 ]
Lu, Zhenzhou [1 ]
机构
[1] Northwestern Polytech Univ, Sch Aeronaut, State Key Lab Clean & Efficient Turbomachinery Pow, Natl Key Lab Aircraft Configurat Design, Xian 710072, Shaanxi, Peoples R China
基金
中国国家自然科学基金;
关键词
Physics-informed neural network; Directional sampling; Partial differential equation; Failure probability; Adaptive strategy;
D O I
10.2514/1.J064926
中图分类号
V [航空、航天];
学科分类号
08 ; 0825 ;
摘要
To efficiently estimate the failure probability for structures with performance function governed by the partial differential equation, this paper proposed an adaptive physics-informed neural network based directional sampling (A-PINN-DS) method. The proposed A-PINN-DS possesses both the low dependence on labeled data from PINN and the high failure domain exploring efficiency from DS. In the proposed method, DS is firstly employed to generate the random input target collocation sample pool (TCSP) with high failure domain exploring efficiency. Then in TCSP, the neural network is trained by the partial differential physics information governing the performance function, and the neural network trained to be convergent in TCSP is used to predict the performance function response of TCSP, on which the failure probability can be estimated by DS. Because the proposed method only uses partial differential physics information instead of finite element simulation to train neural network in the TCSP for predicting performance function, it remarkably reduces the computational cost for estimating failure probability. To improve the efficiency of training PINN in TCSP, this paper designs a strategy of adaptively updating loss weights and a convergence distinguishing strategy for training PINN. Finally, the feasibility and superiority of the proposed method are verified by several examples.
引用
收藏
页数:13
相关论文
共 50 条
  • [1] Physics-informed neural network classification framework for reliability analysis
    Shi, Yan
    Beer, Michael
    Expert Systems with Applications, 2024, 258
  • [2] Sensitivity-based Adaptive Sampling for Physics-Informed Neural Networks
    Chang, Shuji
    Agarwal, Piyush
    McCready, Chris
    Ricardez-Sandoval, Luis
    Budman, Hector
    IFAC PAPERSONLINE, 2024, 58 (14): : 325 - 330
  • [3] Adaptive-Sampling Physics-Informed Neural Network for Viscoacoustic Wavefield Simulation
    Zhao, Haixia
    Hao, Yufan
    Xu, Wenhao
    Li, Chen
    IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, 2024, 21
  • [4] A Gaussian mixture distribution-based adaptive sampling method for physics-informed neural networks
    Jiao, Yuling
    Li, Di
    Lu, Xiliang
    Yang, Jerry Zhijian
    Yuan, Cheng
    ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE, 2024, 135
  • [5] Physics-Informed Neural Networks with Generalized Residual-Based Adaptive Sampling
    Song, Xiaotian
    Deng, Shuchao
    Fan, Jiahao
    Sun, Yanan
    ADVANCED INTELLIGENT COMPUTING TECHNOLOGY AND APPLICATIONS, PT II, ICIC 2024, 2024, 14863 : 320 - 332
  • [6] A physics-informed neural network enhanced importance sampling (PINN-IS) for data-free reliability analysis
    Roy, Atin
    Chatterjee, Tanmoy
    Adhikari, Sondipon
    PROBABILISTIC ENGINEERING MECHANICS, 2024, 78
  • [7] Simulation-free reliability analysis with importance sampling-based adaptive training physics-informed neural networks: Method and application to chloride penetration
    Song, Chaolin
    Xiao, Rucheng
    Zhang, Chi
    Zhao, Xinwei
    Sun, Bo
    RELIABILITY ENGINEERING & SYSTEM SAFETY, 2024, 246
  • [8] Physics-Informed Neural Network Based Digital Image Correlation Method
    Li, B.
    Zhou, S.
    Ma, Q.
    Ma, S.
    EXPERIMENTAL MECHANICS, 2025, 65 (02) : 221 - 240
  • [9] Efficient training of physics-informed neural networks via importance sampling
    Nabian, Mohammad Amin
    Gladstone, Rini Jasmine
    Meidani, Hadi
    COMPUTER-AIDED CIVIL AND INFRASTRUCTURE ENGINEERING, 2021, 36 (08) : 962 - 977
  • [10] Cascading Failure Analysis Based on a Physics-Informed Graph Neural Network
    Zhu, Yuhong
    Zhou, Yongzhi
    Wei, Wei
    Wang, Ningbo
    IEEE TRANSACTIONS ON POWER SYSTEMS, 2023, 38 (04) : 3632 - 3641