Scholarlink

<p>Wide distribution, small size, and diverse shapes of micropores lead to difficulty in predicting fatigue life of micro-arc oxidation (MAO) coated Al alloys. In this study, the MAO coating of 5.1&#xa0;μm is deposited on 2024-T3 Al alloy. Image processing technology based on machine learning (ML) was used to analyze the micropores. The average micropore size is statistically 0.82&#xa0;μm. Combined with Murakami and crack propagation theory, fatigue life prediction physical model was established. In addition, generative adversarial networks are used to augment fatigue data. Random forest (RF), extreme gradient boosting and k-nearest neighbor algorithm are used to train fatigue data. Furthermore, a genetic algorithm (GA) is proposed to optimize the hyperparameters of the RF model. The GA-RF model possesses high prediction accuracy with coefficient of determination <i>R</i><sup>2</sup> of 0.95 and mean absolute percentage error of 8.11%. Subsequently, utilizing limited-memory broyden-fletcher-goldfarb-shanno algorithm, the physical parameters are inversely identified through the integration of GA-RF model and the physical model for fatigue life. Thus, the fatigue life prediction equation of the 2024-T3 Al alloy with thin MAO coating is established. In addition, the fatigue life prediction methodology and its equation are validated for MAO coated Ti-6Al-4&#xa0;V alloy and the bare 2024-T3 Al alloy with Ra = 0.2&#xa0;μm, respectively. The physics-data hybrid driving method provides an effective new approach for predicting fatigue life of thin MAO coated samples.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Fatigue Life Prediction of 2024-T3 Al Alloy with Thin Micro-Arc Oxidation Coating

  • 期刊论文

  • 发表日期：

    2025年12月11日

查看全文
  • Ce Zhang,
  • Chenguang Guo,
  • Weibing Dai,
  • Haitao Yue,
  • Qiang Li,
  • Jianzhuo Zhang,
  • Changyou Li

摘要

Wide distribution, small size, and diverse shapes of micropores lead to difficulty in predicting fatigue life of micro-arc oxidation (MAO) coated Al alloys. In this study, the MAO coating of 5.1 μm is deposited on 2024-T3 Al alloy. Image processing technology based on machine learning (ML) was used to analyze the micropores. The average micropore size is statistically 0.82 μm. Combined with Murakami and crack propagation theory, fatigue life prediction physical model was established. In addition, generative adversarial networks are used to augment fatigue data. Random forest (RF), extreme gradient boosting and k-nearest neighbor algorithm are used to train fatigue data. Furthermore, a genetic algorithm (GA) is proposed to optimize the hyperparameters of the RF model. The GA-RF model possesses high prediction accuracy with coefficient of determination R2 of 0.95 and mean absolute percentage error of 8.11%. Subsequently, utilizing limited-memory broyden-fletcher-goldfarb-shanno algorithm, the physical parameters are inversely identified through the integration of GA-RF model and the physical model for fatigue life. Thus, the fatigue life prediction equation of the 2024-T3 Al alloy with thin MAO coating is established. In addition, the fatigue life prediction methodology and its equation are validated for MAO coated Ti-6Al-4 V alloy and the bare 2024-T3 Al alloy with Ra = 0.2 μm, respectively. The physics-data hybrid driving method provides an effective new approach for predicting fatigue life of thin MAO coated samples.