• CN:11-2187/TH
  • ISSN:0577-6686

机械工程学报 ›› 2017, Vol. 53 ›› Issue (7): 47-54.doi: 10.3901/JME.2017.07.047

• 机构学及机器人 • 上一篇    下一篇

柔性机构动态可靠性分析的先进极值响应面方法*

张春宜1, 宋鲁凯1,2, 费成巍3, 郝广平1, 路成1   

  1. 1. 哈尔滨理工大学机械动力工程学院 哈尔滨 150080;
    2. 北京航空航天大学能源与动力工程学院 北京 100191;
    3. 香港理工大学机械工程系 香港 999077
  • 出版日期:2017-04-05 发布日期:2017-04-05
  • 作者简介:

    张春宜(通信作者),男,1960年生,博士,教授,博士研究生导师。主要研究方向为机械设计,优化设计,可靠性工程。

    E-mail:zhangchunyi@hrbust.edu.cn

  • 基金资助:
    * 国家自然科学基金(51275138)和黑龙江省教育厅科学基金(12531109)资助项目; 20160501收到初稿,20170124收到修改稿;

Advanced Extremum Response Surface Method for Dynamic Reliability Analysis on Flexible Mechanism

ZHANG Chunyi1, SONG Lukai1,2, FEI Chengwei3, HAO Guangping1, LU Cheng1   

  1. 1. School of Mechanical and Power Engineering, Harbin University of Science and Technology, Harbin 150080;
    2. School of Energy and Power Engineering, Beihang University, Beijing 100191;
    3. Department of Mechanical Engineering, Hong Kong Polytechnic University, Hong Kong 999077
  • Online:2017-04-05 Published:2017-04-05

摘要:

为了提高柔性机构可靠性分析的精度和效率,提出动态可靠性分析的先进极值响应面法。该方法将智能算法与可靠性分析的极值响应面法相结合,利用蒙特卡洛法抽取样本,经过网络训练建立先进极值响应面法的数学模型。以柔性机械臂为例,以柔性机械臂材料密度、弹性模量、构件截面尺寸为输入随机变量,构件变形为输出响应;利用蒙特卡罗法、极值响应面法和先进极值响应面法分别进行动态可靠性仿真计算,得出了各自输出响应量的分布特征及可靠度。通过方法比较表明:先进极值响应面法在保证计算精度的前提下,大大提高了计算速度,验证了先进极值响应面法在柔性机构动态可靠性分析中的可行性和适用性;也为柔性机构可靠性优化开辟了有效途径。

关键词: 人工神经网络, 柔性机械臂, 先进极值响应面法, 智能算法, 可靠性分析

Abstract:

To improve the precision and efficiency of reliability analysis of flexible mechanism, advanced extremum response surface method of reliability analysis is proposed. Based on the intelligent algorithm and extremum response surface method, the mathematical model of the proposed method is established through the generation of samples by Monte Carlo method and the network training. A flexible robot manipulator is chosen as the target of study in the reliability simulation where its material density, elastic modulus, section size of components are taken as input random variables and deformation of components is taken as output response. Distribution characteristics of output response and reliability are assessed by Monte Carlo method, extremum surface method and advanced extremum surface method respectively. The comparison with various other methods shows that this advanced extremum response surface method has greatly improved the computational speed while keeping acceptable computational accuracy, advanced extremum response surface method is proved to be a high precision and high efficiency in dynamic reliability analysis of flexible mechanism, and this method has opened up an effective way for reliability optimization of flexible mechanism.

Key words: advanced extremum response surface method, artificial neural network, flexible robot manipulator, intelligent algorithm, reliability analysis