中国机械工程

• 可持续制造 • 上一篇    下一篇

基于磁记忆和表面纹理特征融合的再制造毛坯疲劳损伤评估

刘涛1;鲍宏2;朱达荣1;汪方斌1;雷经发1   

  1. 1.安徽建筑大学机械与电气工程学院,合肥,230601
    2.合肥工业大学工业培训中心,合肥,230601
  • 出版日期:2018-07-10 发布日期:2018-07-10
  • 基金资助:
    国家自然科学基金资助项目(51505119);
    安徽省自然科学基金资助项目(1508085QE91,1708085ME130);
    安徽省教育厅高校自然科学研究重大项目(KJ2017ZD42)
    National Natural Science Foundation of China (No. 51505119)
    Anhui Provincial Natural Science Foundation of China (No. 1508085QE91,1708085ME130)

Fatigue Damage Evaluation of Remanufacturing Cores Using Feature Fusion of Magnetic Memory and Surface Texture

LIU Tao1;BAO Hong2;ZHU Darong1;WANG Fangbin1;LEI Jingfa1   

  1. 1.School of Mechanical and Electrical Engineering,Anhui Jianzhu University,Hefei,230601
    2.Industrial Training Center,Hefei University of Technology,Hefei, 230601
  • Online:2018-07-10 Published:2018-07-10
  • Supported by:
    National Natural Science Foundation of China (No. 51505119)
    Anhui Provincial Natural Science Foundation of China (No. 1508085QE91,1708085ME130)

摘要: 为提高再制造损伤评估的精度和效率,提出了一种基于磁记忆和表面纹理特征融合的构件损伤评估方法。通过磁记忆特征和表面纹理特征建模,提取磁信号及其梯度的样本熵参数以及表面纹理的能量、熵、反差和相关参数,分析了各特征参数的损伤时序变化规律。分别从数据层和指标层建立特征融合模型。在数据层将磁特征和表面纹理特征参数作为非线性映射的输入,得到其与损伤之间的映射关系。指标层采用D-S证据理论方法,进行磁和表面纹理特征量之间的信息融合,将各损伤状态作为识别框架,通过Bayes近似,得到融合后的各证据信度函数值,并依据函数值进行损伤状态评估。最后选取疲劳试样进行了该方法的验证。

关键词: 再制造, 疲劳损伤评估, 磁记忆, 表面纹理, 特征融合

Abstract: In order to improve the accuracy and effectiveness of the damage evaluation for remanufacturing, a method to assess the fatigue damage of remanufacturing cores was proposed using feature fusion of magnetic memory and surface texture. The sample entropy parameters from magnetic signals and their gradients, and energy, entropy, contrast, and correlation parameters from surface texture were extracted by feature modeling of magnetic memory and surface texture. Meanwhile, the sequential variation regularities of each parameters were analyzed. The feature fusion models were established from the data layer and the index layer respectively. The parameters of magnetic memory and surface texture features were defined as non-linear mapping inputs in the data layer, and their mapping relationships with damage were obtained. Later, in the index layer, the information fusion between magnetic and surface texture eigenvalues was carried out by Dempster-Shafer (D-S) evidence theory. The Bayes approximation was used to obtain the belief function values of each evidence, then, according to the function value, the damage states were evaluated. Finally, fatigue test samples were selected to illustrate the capability of the proposed method.

Key words: remanufacturing, fatigue damage evaluation, magnetic memory, surface texture, feature fusion

中图分类号: