东北大学学报(自然科学版) ›› 2003, Vol. 24 ›› Issue (10): 966-969.DOI: -

• 论著 • 上一篇    下一篇

基于贝叶斯统计的汽车半轴可靠性设计

王铁;张国忠;周淑文   

  1. 东北大学机械工程与自动化学院;东北大学机械工程与自动化学院;东北大学机械工程与自动化学院 辽宁沈阳 110004
  • 收稿日期:2013-06-24 修回日期:2013-06-24 出版日期:2003-10-15 发布日期:2013-06-24
  • 通讯作者: Wang, T.
  • 作者简介:-
  • 基金资助:
    国家自然科学基金资助项目(59835050)·

Reliability design of vehicle axles on Bayes

Wang, Tie (1); Zhang, Guo-Zhong (1); Zhou, Shu-Wen (1)   

  1. (1) Sch. of Mech. Eng. and Automat., Northeastern Univ., Shenyang 110004, China
  • Received:2013-06-24 Revised:2013-06-24 Online:2003-10-15 Published:2013-06-24
  • Contact: Wang, T.
  • About author:-
  • Supported by:
    -

摘要: 分析了半轴的受力及贝叶斯统计的可靠性·为了实现给定可靠度求出半轴直径或给定半轴直径求出可靠度,对贝叶斯统计后验期望方法进行了简化整理,即利用传统可靠性的知识将其他分布转化为正态分布,再利用指数族共轭分布的特点计算出正态先后验分布参数,最后运用符号换元积分和Newton Raphson迭代法得出数值解(收敛速度极快),并通过例子予以验证,在计算机上用VB编程实现·克服了传统方法的保守性,使设计更合理·特别适于经典统计方法不适应的个性化和特殊化设计,对设计方案的频繁变更及贫乏的实验和现场数据都能得出比传统方法更精确的解·

关键词: 贝叶斯统计, 统计设计, 车辆工程, 可靠性, 半轴

Abstract: Reliabilities of force condition analysis of vehicle axles and Bayes statistics are both discussed with the intention of determining the axle diameter on given reliability basis and vice versa, the posterior expectation of Bayes statistics is simplified in order in sequence: (1) converting the non-normal distribution into equivalent normal one by use of conventional approaches to reliability; (2) calculating the prior and posterior parameters of normal distribution by use of the characteristics of exponent group in conjugate distribution; (3) getting numerical solutions with very high convergence by use of symbolic integration by substitution and Newton-Ralphson iteration method. The solution is verified through an instance. The algorithm is implemented on computer with VB programming, thus avoiding out-of-date methods and rationalizing conventional design procedure. In particular the method proposed here applies to the personalized/specialized designs to which the classical statistics method is unavailable, and it may provide more accurate solutions than traditional methods for those designs which will be changed frequently or poor in either experimental or in-site data.

中图分类号: