基于混合机器学习模型的两级加载下金属材料的剩余疲劳寿命预测方法

doi:10.3901/JME.2025.12.073

摘要/Abstract

摘要： 剩余疲劳寿命预测对于保障工程结构的安全可靠至关重要。特别是，当结构物遭受变幅值载荷时，由于疲劳损伤累积的复杂性性以及数据中固有的噪声和不确定性，使得剩余疲劳寿命预测具有挑战性。为了解决这个问题，建立混合机器学习模型GA-BP-GPR方法，该方法使用GA-BP方法(遗传算法-反向传播神经网络)映射输入和剩余寿命之间的复杂关系结合GPR (高斯过程回归)获取相应的不确定性，用于估计两级加载下金属材料的剩余疲劳寿命。在包含12种金属材料、总共328个样本的数据集上对GA-BP-GPR方法进行全面评估。GA-BP-GPR方法对测试集所有预测结果都落在3倍分散带内；与5种机器学习方法和2种传统疲劳预测方法对比结果表明所提方法对两步载荷下的剩余寿命预测具有较高的准确性和可靠性。此外，所提方法对没有参与训练的Al-2024-T42的预测结果都在2倍分散带内，表明所提方法对新材料剩余疲劳寿命预测的泛化能力。

关键词: 剩余疲劳寿命, 疲劳寿命预测, 遗传算法, 高斯过程回归, 机器学习

Abstract: Remaining fatigue life prediction is crucial for the safety and reliability of engineering structures. When the structure is subjected to multi-step loading, it becomes challenging to predict the remaining fatigue life due to the complexity of fatigue damage accumulation and inherent noise and uncertainty in the data. To address this problem, a hybrid machine learning method, the GA-BP-GPR method, is established, which uses the GA-BP method (genetic algorithm-backpropagation neural network) to map the complex relationship between input features and remaining fatigue life and combines with GPR (Gaussian process regression) to obtain the corresponding uncertainty, for estimating the remaining fatigue life of metal materials under two-step loading. The GA-BP-GPR method is comprehensively evaluated on a dataset of 328 samples from 12 metal materials. The predicted results of the test set using the proposed method fall within the ±3 scatter band. In comparison with the five machine learning methods and two traditional methods, the results indicate that the proposed method can achieve higher accuracy and reliability in predicting remaining fatigue life under two-step loading. In addition, the predicted results of the proposed method for Al-2024-T42 fall within the ±2 scatter band, indicating the proposed method has generalization ability for remaining fatigue life prediction under two-step loading for new materials.

Key words: remaining fatigue life, fatigue life prediction, genetic algorithm, Gaussian process regression, machine learning

中图分类号:

O346

徐自力, 高京京, 覃曼青, 何孟夫. 基于混合机器学习模型的两级加载下金属材料的剩余疲劳寿命预测方法[J]. 机械工程学报, 2025, 61(12): 73-82.

XU Zili, GAO Jingjing, QIN Manqing, HE Mengfu. Hybrid Machine Learning Method for Remaining Fatigue Life Prediction of the Metallic Materials under Two-step Loading[J]. Journal of Mechanical Engineering, 2025, 61(12): 73-82.

参考文献

[1] JORDAN M I，MITCHELL T M. Machine learning：Trends，perspectives，and prospects[J]. Science，2015，349(6245)：255-260.
[2] FARID M. Data-driven method for real-time prediction and uncertainty quantification of fatigue failure under stochastic loading using artificial neural networks and Gaussian process regression[J]. International Journal of Fatigue，2021，155：106415.
[3] AFSHARI S S，ENAYATOLLAHI F，XU X，et al. Machine learning-based methods in structural reliability analysis：A review[J]. Reliability Engineering & System Safety，2022，219：108223.
[4] CHEN J，LIU Y. Fatigue modeling using neural networks：A comprehensive review[J]. Fatigue & Fracture of Engineering Materials & Structures，2022，45(4)：945-979.
[5] BOCK F E，AYDIN R C，CYRON C J，et al. A review of the application of machine learning and data mining approaches in continuum materials mechanics[J]. Frontiers in Materials，2019，6(110)：1-23.
[6] JOHNSON N S，VULIMIRI P S，TO A C，et al. Invited review：Machine learning for materials developments in metals additive manufacturing[J]. Additive Manufacturing，2020，36：101641.
[7] NASIRI S，KHOSRAVANI M R，WEINBERG K. Fracture mechanics and mechanical fault detection by artificial intelligence methods：A review[J]. Engineering Failure Analysis，2017，81：270-293.
[8] WANG C，TAN X P，TOR S B，et al. Machine learning in additive manufacturing：State-of-the-art and perspectives[J]. Additive Manufacturing，2020，36：101538.
[9] GAO J，WANG C，XU Z，et al. Gaussian process regression based remaining fatigue life prediction for metallic materials under two-step loading[J]. International Journal of Fatigue，2022，158：106730.
[10] GALANOPOULOS G，MILANOSKI D，ELEFTHEROGLOU N，et al. Acoustic emission-based remaining useful life prognosis of aeronautical structures subjected to compressive fatigue loading[J]. Engineering Structures，2023，290：116391.
[11] LEI G，ZHAO X，WU H，et al. Estimation of remaining fatigue life under two-step loading based on kernel-extreme learning machine[J]. International Journal of Fatigue，2021，148：106190.
[12] LI J，CAO X，CHEN R，et al. Prediction of remaining fatigue life of metal specimens using data-driven method based on acoustic emission signal[J]. Applied Acoustics，2023，211：109571.
[13] GAN L，WU H，ZHONG Z. Integration of symbolic regression and domain knowledge for interpretable modeling of remaining fatigue life under multistep loading[J]. International Journal of Fatigue，2022，161：106889.
[14] FENG S Z，HAN X，LI Z，et al. Ensemble learning for remaining fatigue life prediction of structures with stochastic parameters：A data-driven approach[J]. Applied Mathematical Modelling，2022，101：420-431.
[15] GU H，WANG R，ZHU S，et al. Machine learning assisted probabilistic creep-fatigue damage assessment[J]. International Journal of Fatigue，2022，156：106677.
[16] HENG J，ZHENG K，FENG X，et al. Machine learning-assisted probabilistic fatigue evaluation of Rib-to-Deck joints in orthotropic steel decks[J]. Engineering Structures，2022，265：114496.
[17] CHEN J，LIU Y. Probabilistic physics-guided machine learning for fatigue data analysis[J]. Expert Systems with Applications，2021，168：114316.
[18] HU D，SU X，LIU X，et al. Bayesian-based probabilistic fatigue crack growth evaluation combined with machine-learning-assisted GPR[J]. Engineering Fracture Mechanics，2020，229：106933.
[19] GAO J，WANG J，XU Z，et al. Multiaxial fatigue prediction and uncertainty quantification based on back propagation neural network and Gaussian process regression[J]. International Journal of Fatigue，2023，168：107361.
[20] CYBENKO G V. Approximation by superpositions of a sigmoidal function[J]. Mathematics of Control，Signals，and Systems，1989，2(4)：303-314.
[21] KATOCH S，CHAUHAN S S，V K. A review on genetic algorithm：Past，present，and future[J]. Multimedia Tools and Applications，2021，80：8091-8126.
[22] SHENG H，LIU X，BAI L，et al. Small sample state of health estimation based on weighted Gaussian process regression[J]. Journal of Energy Storage，2021，41：102816.
[23] HAN G S，LEE J. Prediction of pricing and hedging errors for equity linked warrants with Gaussian process models[J]. Expert Systems with Applications，2008，35(1-2)：515-523.
[24] PAVLOU D G. A phenomenological fatigue damage accumulation rule based on hardness increasing，for the 2024-T42 aluminum[J]. Engineering Structures，2002，24(11)：1363-1368.
[25] MESMACQUE G，GARCIA S，AMROUCHE A，et al. Sequential law in multiaxial fatigue，a new damage indicator[J]. International Journal of Fatigue，2005，27(4)：461-467.
[26] MANSON S S，FRECHE J C，ENSIGN C R. Application of a double linear damage rule to cumulative fatigue[R]. National Aeronautics and Space Administration，1967.
[27] KAMAYA M，KAWAKUBO M. Loading sequence effect on fatigue life of type 316 stainless steel[J]. International Journal of Fatigue，2015，81：10-20.
[28] 谢里阳，吕文阁，师照峰. 两级载荷作用下疲劳损伤状态的试验研究[J]. 机械强度，1994，16(3)：52-54. XIE Liyang，LÜ Wenge，SHI Zhaofeng. Experimental study on fatigue damage under two level loading. Journal of Mechinical Strength，1994，16(3)：52-54.
[29] ZHU S，LIAO D，LIU Q，et al. Nonlinear fatigue damage accumulation：Isodamage curve-based model and life prediction aspects[J]. International Journal of Fatigue，2019，128：105185.
[30] SUBRAMANYAN S. A cumulative damage rule based on the knee point of the S-N curve[J]. Journal of Engineering Materials & Technology，1976，98(4)：316.
[31] XIA F，ZHU S，LIAO D，et al. Isodamage curve-based fatigue damage accumulation model considering the exhaustion of static toughness[J]. Engineering Failure Analysis，2020，115：104575.
[32] DATTOMA V，GIANCANE S，NOBILE R，et al. Fatigue life prediction under variable loading based on a new non-linear continuum damage mechanics model[J]. International Journal of Fatigue，2006，28(2)：89-95.
[33] 方义庆，胡明敏，罗艳利. 基于全域损伤测试建立的连续疲劳损伤模型[J]. 机械强度，2006，28(4)：582-586. FANG Yiqing，HU Mingmin，LUO Yanli. New continuous fatigue damage model based on whole damage field measurement[J]. Journal of Mechanical Strength，2006，28(4)：582-586.
[34] AID A，AMROUCHE A，BOUIADJRA B B，et al. Fatigue life prediction under variable loading based on a new damage model[J]. Materials & Design，2011，32(1)：183-191.
[35] SHANG D，YAO W. A nonlinear damage cumulative model for uniaxial fatigue[J]. International Journal of Fatigue，1999，21(2)：187-194.
[36] PEREIRA H F S G，DE JESUS A M P，RIBEIRO A S，et al. Fatigue damage behavior of a structural component made of P355NL1 steel under block loading[J]. Journal of Pressure Vessel Technology，2009，131(2)：021407.
[37] FANG Y，HU M，LUO Y. New continuous fatigue damage model based on whole damage field measurement[J]. Journal of Mechanical Strength，2006，28(4)：582-586.
[38] HECHT NIELSEN R. Kolmogorov’s mapping neural network existence theorem[C]//Proceedings of the International Conference on Neural Networks，New York, NY, USA：IEEE Press, 1987：11-14.
[39] RIPLEY B D. Pattern recognition and neural networks[M]. New York：Cambridge University Press，2007.
[40] WANG C. A theory of generalization in learning machines with neural network applications[M]. Philadelphia：University of Pennsylvania，1994.
[41] KAASTRA I，BOYD M. Designing a neural network for forecasting financial and economic time series[J]. Neurocomputing，1996，10(3)：215-236.
[42] SHIBATA K，IKEDA Y. Effect of number of hidden neurons on learning in large-scale layered neural networks[C]//ICROS-SICE International Joint Conference，August 18-21，2009，Fukuoka International Congress Center，Japan，ICCAS-SIC，New York：IEEE，2009：5008-5013.
[43] YE D，WANG Z. A new approach to low-cycle fatigue damage based on exhaustion of static toughness and dissipation of cyclic plastic strain energy during fatigue[J]. International Journal of Fatigue，2001，23(8)：679-687.
[44] MINER M A. Cumulative damage in fatigue[J]. Journal of Applied Mechanics，1945，12(3)：A159-A164
[45] WANG S，XIA P，CHEN K. Prediction and optimization model of sustainable concrete properties using machine learning，deep learning and swarm intelligence：A review[J]. Journal of Building Engineering，2023：108065.
[46] SUN Q，ZHANG H C，SUN Z. Ridge regression and artificial neural network to predict the thermodynamic properties of alkali metal Rankine cycles for space nuclear power[J]. Energy Conversion and Management，2022，273：116385.