中国机械工程

• 机械基础工程 • 上一篇    下一篇

基于反演识别法的有限元模型校正

陈永亮;刘德帅;刘玉坤;彭涛   

  1. 天津大学,天津,300072
  • 出版日期:2016-11-10 发布日期:2016-11-10
  • 基金资助:
    国家自然科学基金资助项目(51275347);天津市重大科技专项(12ZCDZGX45000);天津市自然科学基金重点资助项目(13JCZDJC35000)

Finite Element Model Updating Based on Inversion Method of Identification

Chen Yongliang;Liu Deshuai;Liu Yukun;Peng Tao   

  1. Tianjin University,Tianjin,300072
  • Online:2016-11-10 Published:2016-11-10
  • Supported by:
     

摘要: 提出了一种在动态分析中对机构的结合面进行参数识别的方法。建立机械结构的有限元模型并将其关键结合面以弹簧-阻尼单元代替,将模态计算的结果与实验结果相结合建立目标函数;通过BP神经网络拟合、遗传算法参数寻优,得出最优结合面参数。以立式圆台磨床为例,运用该方法,对其结合面参数进行了识别。结果表明,通过该方法进行有限元模型中结合面参数的识别是可行的。

关键词: 立式磨床, 模态试验, 有限元计算, 神经网络, 遗传算法

Abstract: In dynamic analysis, a method to identify parameters of contact surfaces was put forward. The finite element model of mechanical structure was set up, and its key contact surface was replaced by spring-damper elements. Then the modality calculation was carried out. The objective function was built by combining the calculation results with the experimental results. Finally, the optimal parameters were obtained by the fitting of BP neural network and parameter optimization of genetic algorithm. The method was applied to the vertical grinder to obtain the parameters of contact surface. The results show that the method of identifying the parameters of contact surfaces is feasible.

Key words: vertical grinding machine, modal test, finite element calculation, neural network, genetic algorithm

中图分类号: