High-cycle fatigue strength prediction for 7050 aluminum alloy using data augmentation and hybrid neural network in small sample scenarios

ZHAO Dewang; XU Yufan; YANG Wenping; ZHAO Yanguang; Junling

doi:10.11990/jheu.20260612013

Chinese

您当前的位置：

首页 >

文章列表页 >

High-cycle fatigue strength prediction for 7050 aluminum alloy using data augmentation and hybrid neural network in small sample scenarios

更新时间：2026-09-18

- High-cycle fatigue strength prediction for 7050 aluminum alloy using data augmentation and hybrid neural network in small sample scenarios
- Journal of Harbin Engineering University Pages: 1-14(2026)
- 作者机构：
  
  1.安徽理工大学机电工程学院，安徽淮南232000
  2.哈尔滨工程大学航空航天学院，黑龙江哈尔滨 150001
  3.大连理工大学力学与航空航天学院，辽宁大连 116000
  4.中国飞机强度研究所强度与结构完整性全国重点实验室，陕西西安 710065
- 作者简介：
- 基金信息：
- DOI：10.11990/jheu.20260612013
  CLC： TB114.3;TG146.22;TP183
- Received：12 June 2026，
  
  Revised：2026-08-15，
  
  Accepted：08 September 2026，
  
  Online First：18 September 2026，
- 稿件说明：
移动端阅览
赵德望,徐雨帆,杨文平等.小样本下基于数据衍生与混合神经网络的7050铝合金高周疲劳强度预测[J].哈尔滨工程大学学报,

ZHAO Dewang，XU Yufan，YANG Wenping，et al.High-cycle fatigue strength prediction for 7050 aluminum alloy using data augmentation and hybrid neural network in small sample scenarios［J］.Journal of Harbin Engineering University，DOI：10.11990/jheu.20260612013.
赵德望,徐雨帆,杨文平等.小样本下基于数据衍生与混合神经网络的7050铝合金高周疲劳强度预测[J].哈尔滨工程大学学报, DOI：10.11990/jheu.20260612013.

ZHAO Dewang，XU Yufan，YANG Wenping，et al.High-cycle fatigue strength prediction for 7050 aluminum alloy using data augmentation and hybrid neural network in small sample scenarios［J］.Journal of Harbin Engineering University，DOI：10.11990/jheu.20260612013. DOI：

摘要

针对7050铝合金高周疲劳试验周期长、成本高，且传统模型在小样本条件下预测精度低、高周外推鲁棒性差的问题，本文开展了5种不同应力比与应力集中系数工况的全寿命疲劳试验，获取了材料基础疲劳性能数据。提出一种融合浅层神经网络数据衍生与ANN+CNN-GRU-Attention混合网络的小样本疲劳强度预测方法，引入含物理先验的趋势感知损失函数抑制外推失稳，设计滑动时间窗增量迭代训练策略实现向10

次循环的逐步外推。结果表明：模型有效克服了长短期记忆模型在外推阶段的预测失稳问题，预测精度较单次全量训练提升13.56%；增量迭代训练使5种工况疲劳强度预测平均偏差降至5.29%，在本文工况下表现出良好的工程适用性。本文方法可缩短材料高周疲劳性能试验周期，为航空铝合金疲劳性能快速评估提供了一种可行的数据驱动技术途径。

Abstract

To address the problems of long cycle time and high cost in high-cycle fatigue testing of 7050 aluminum alloy， as well as the low prediction accuracy and poor high-cycle extrapolation robustness of traditional models under small-sample conditions， full-life fatigue tests under five different stress ratios and stress concentration factor conditions were carried out to obtain the basic fatigue performance data of the material. A small-sample fatigue strength prediction method combining shallow neural network-based data augmentation and an ANN+CNN-GRU-Attention hybrid network was proposed. A trend loss function with physical priors was introduced to suppress extrapolation instability， and a sliding-window incremental iterative training strategy was designed to achieve gradual extrapolation to 10⁷ cycles. The results show that the proposed model effectively overcomes the prediction instability problem of the LSTM model during the extrapolation stage， and the prediction accuracy is improved by 13.56 percentage points compared with single full-batch training. The incremental iterative training reduces the average deviation of fatigue strength prediction under the five working conditions to 5.29%， demonstrating good engineering applicability under the conditions investigated in this study. The proposed method can reduce the high-cycle fatigue testing duration and provides a feasible data-driven technical approach for the rapid evaluation of fatigue performance of aviation aluminum alloys.

关键词

Keywords

references

王金龙，彭文杰，冀秀坤，等 . 不同类型表面粗糙度对钛合金疲劳失效影响的试验研究［J］. 哈尔滨工程大学学报， 2025 ， 46 （ 5 ）： 1032 - 1038， 1046 .

WANG Jinlong ， PENG Wenjie ， JI Xiukun ， et al . Experimental study on the effect of different types of surface roughness on fatigue failure of titanium alloys ［J］. Journal of Harbin Engineering University ， 2025 ， 46 （ 5 ）： 1032 - 1038， 1046 .

ZAWADA-MICHAŁOWSKA M ， PIEŚKO P ， MRÓWKA-NOWOTNIK G ， et al . Effect of the technological parameters of milling on residual stress in the surface layer of thin-walled plates ［J］. Materials ， 2024 ， 17 （ 5 ）： 1193 .

VIJAYANANDH R ， KUMAR K N ， KUMAR G R ， et al . Comparative approaches for fatigue life estimation of aluminium alloy for aerospace applications ［J］. International journal of vehicle structures and systems ， 2018 ， 10 （ 4 ）： 282 .

SWANSON S R . Handbook of Fatigue Testing ［M］. West Conshohocken ： ASTM International ， 1974 .

MANSON S S ， HALFORD G R . Practical implementation of the double linear damage rule and damage curve approach for treating cumulative fatigue damage ［J］. International journal of fracture ， 1981 ， 17 （ 2 ）： 169 - 192 .

SMITH K . A stress-strain function for the fatigue of metals ［J］. Journal of Materials ， 1970 ， 5 ： 767 – 778 .

ZOU Li ， SUN Yibo ， YANG Xinhua . An entropy-based neighborhood rough set and PSO-SVRM model for fatigue life prediction of titanium alloy welded joints ［J］. Entropy ， 2019 ， 21 （ 2 ）： 117 .

HENG Fei ， GAO Jianxiong ， XU Rongxia ， et al . Multiaxial fatigue life prediction for various metallic materials based on the hybrid CNN-LSTM neural network ［J］. Fatigue & fracture of engineering materials & structures ， 2023 ， 46 （ 5 ）： 1979 - 1996 .

WANG Yanju ， ZHU Zhenyu ， SHA Aixue ， et al . Low cycle fatigue life prediction of titanium alloy using genetic algorithm-optimized BP artificial neural network ［J］. International journal of fatigue ， 2023 ， 172 ： 107609 .

ZHAN Zhixin ， HU Weiping ， MENG Qingchun . Data-driven fatigue life prediction in additive manufactured titanium alloy： a damage mechanics based machine learning framework ［J］. Engineering fracture mechanics ， 2021 ， 252 ： 107850 .

JING Hutao ， YUAN Shenfang ， CHEN Jian ， et al . A multi-scale deep residual network-based guided wave imaging evaluation method for fatigue crack quantification ［J］. Measurement science and technology ， 2025 ， 36 （ 1 ）： 016115 .

LIAO Huiya ， PAN Jun ， SU Xihui ， et al . A path-dependent adaptive physics-informed neural network for multiaxial fatigue life prediction ［J］. International journal of fatigue ， 2025 ， 193 ： 108799 .

张红哲，王德鹏，鲍永杰，等 . 基于长短期记忆网络的航空用TC4钛合金缺口件断裂行为分析［J/OL］. 哈尔滨工程大学学报， 2026 （ 2026-04-09 ）. https：//link.cnki.net/urlid/23.1390.u.20260408.1623.002 https://link.cnki.net/urlid/23.1390.u.20260408.1623.002 .

ZHANG Hongzhe ， WANG Depeng ， BAO Yongjie ， et al . Fracture behavior analysis of TC4 titanium alloy notched parts for aviation based on long-term and short-term memory networks ［J/OL］. China industrial economics ， 2026 （ 2026-04-09 ）. https：//link.cnki.net/urlid/23.1390.u.20260408.1623.002 https://link.cnki.net/urlid/23.1390.u.20260408.1623.002 .

国家质量监督检验检疫总局，中国国家标准化管理委员会 . 金属材料疲劳试验轴向力控制方法： GB/T 3075—2008 ［S］. 北京：中国标准出版社， 2009 .

General Administration of Quality Supervision ， Inspection and Quarantine of the People’s Republic of China ， Standardization Administration of the People’s Republic of China . Metallic materials - fatigue testing-axial-force-controlled method ： GB/T 3075—2008 ［S］. Beijing ： Standards Press of China ， 2009 .

YI Min ， TANG Wei ， ZHU Yiqi ， et al . A holistic review on fatigue properties of additively manufactured metals ［J］. Journal of materials processing technology ， 2024 ， 329 ： 118425 .

SHEN Xingkeng ， DAI Ying ， CHEN Xinmin ， et al . Fatigue performance testing and life prediction of welded fuel pipes ［J］. International journal of fatigue ， 2024 ， 187 ： 108453 .

郭宏超，关晓迪，王万祯，等 . 高强钢焊接接头疲劳性能及损伤演化规律研究［J/OL］. 建筑钢结构进展， 2026 ， 1 - 10 ［ 2026-03-17 ］. https：//link.cnki.net/urlid/31.1893.TU.20250916.1726.04 https://link.cnki.net/urlid/31.1893.TU.20250916.1726.04 .

GUO Hongchao ， GUAN Xiaodi ， WANG Wanzhen ， et al . Research on fatigue properties and damage evolution laws of high-strength steel welded joints ［J/OL］. Advances in Building Steel Structures ， 2026 ， 1 - 10 ［ 2026-03-17 ］. https：//link.cnki.net/urlid/31.1893.TU.20250916.1726.04 https://link.cnki.net/urlid/31.1893.TU.20250916.1726.04 .

姚喆赫，何淼峰，王发博，等 . 基于数据增强的激光增材制造GH4169疲劳寿命预测方法研究［J/OL］. 中国激光， 2026 ， 1 - 17 ［ 2026-03-30 ］. https：//link.cnki.net/urlid/31.1339.TN.20251201.0941.074 https://link.cnki.net/urlid/31.1339.TN.20251201.0941.074 .

YAO Zhehe ， HE Miaofeng ， WANG Fabo ， et al . Research on fatigue life prediction method for GH4169 in laser additive manufacturing based on data augmentation ［J/OL］. Chinese Journal of Lasers ， 2026 ， 1 - 17 ［ 2026-03-30 ］. https：//link.cnki.net/urlid/31.1339.TN.20251201.0941.074 https://link.cnki.net/urlid/31.1339.TN.20251201.0941.074 .

LI Yunze ， DONG Chao ， JIANG Chao ， et al . Fatigue limit prediction of 7050 aluminium alloy based on experimental and shallow + deep hybrid neural network ［J］. Engineering failure analysis ， 2025 ， 167 ： 108994 .

LEVENBERG K . A method for the solution of certain non-linear problems in least squares ［J］. Quarterly of applied mathematics ， 1944 ， 2 （ 2 ）： 164 - 168 .

李龙斌 . 基于深度学习的裂纹转子剩余寿命预测［D］. 成都：西南交通大学， 2023 .

LI Longbin . Prediction of residual life of cracked rotor based on deep learning ［D］. Chengdu ： Southwest Jiaotong University ， 2023 .

张波，胡昌华，张浩，等 . 机器学习应用于随机退化设备剩余寿命预测的综述［J］. 哈尔滨工程大学学报， 2024 ， 45 （ 9 ）： 1783 - 1790 .

ZHANG Bo ， HU Changhua ， ZHANG Hao ， et al . A review of remaining useful life prediction for stochastic degrading devices based on machine learning ［J］. Journal of Harbin engineering university ， 2024 ， 45 （ 9 ）： 1783 - 1790 .

赵德望，李运泽，尹汉琪，等 . 一种基于自注意力机制的深度学习小样本增量迭代训练的金属疲劳寿命预测方法及系统： CN120449668A ［P］. 2025-08-08 .

裴洪，张澳，胡昌华，等 . 考虑阶段划分不确定性的轴承剩余寿命预测方法［J/OL］. 哈尔滨工程大学学报， 1 - 8 ［ 2026-06-15 ］. https：//link.cnki.net/urlid/23.1390.U.20260518.1424.002 https://link.cnki.net/urlid/23.1390.U.20260518.1424.002 .

PEI Hong ， ZHANG Ao ， HU Changhua ， et al . Remaining useful life prediction for bearings considering stage division uncertainty ［J/OL］. Journal of Harbin Engineering University ， 1 - 8 ［ 2026-06-15 ］. https：//link.cnki.net/urlid/23.1390.U.20260518.1424.002 https://link.cnki.net/urlid/23.1390.U.20260518.1424.002 .

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Research on verification and optimization of intelligent fault diagnosis methods under nuclear power plant automatic startup conditions

Improving the accuracy of data-driven diagnostic models guided by confusion matrix

Research on offline reinforcement learning method for core power control with data quality constraints

Computational models for static response of thin plates based on physics-informed neural networks

Aerodynamic parameter identification based on an improved extreme learning machine and ensemble learning

Related Author

TAN Sichao

TIAN Ruifeng

MIAO Zhuang

WEI Hongyang

WANG Zixiao

XU Zhao

ZHU Dahuan

QIN Sulin

Related Institution

China Nuclear Power Engineering Co.， Ltd.

College of Nuclear Science and Technology， Harbin Engineering University

Nuclear Power Institute of China

College of Physics， Sichuan University

College of Mathematical Sciences， Harbin Engineering University

AI问答

Postal code：100079
Tel：（010）53879206 Email：tmw@bjxintong.com.cn
Technical support is provided by Beijing Founder electronics co., LTD 京ICP备09082226号-64 京公网安备11010602201714号
It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.

⁰