基于贝叶斯理论的白车身测量数据相关性分析

The Relativity Analysis of BIW Checking Data Based on Bayesian

构造白车身(BIW)偏差流是提高汽车制造质量的重要手段,当前有着多种方法去构造白车身的偏差流,但效果不是特别满意。基于相关性分析等统计手段去构造偏差流的方法正以其相对优越性受到了人们的关注,但由于车身测量数据的特殊性,传统的相关性分析在此失效。本文首先分析了车身测量数据的频率不稳定性和小样本特性,并针对这些特点提出了通过求数据日均值,并通过贝叶斯方法的改善小样本的方法,实例证明该方法可有效改进相关性分析问题。

It is a key way for improving the automobile quality to build the variation stream of body in white(BIW).There are several ways to build it,but these ways are not always effective.Now,people pay more attention to a new way based on relativity analysis for its advantages.However,because of the particularity of auto checking data,it is a difficult problem how to identify the relativity between two different data sequences.This paper discusses the instability of frequency and small sample characteristic of auto checking data,and brings forward a new relativity analysis way based on Bayesian theory,which is testified that the research problems can be solved effectively.