失效分布类型的概率图模糊判别

本文以概率图和模糊综合评判方法为基础,提出了一种在常见可靠性分布下,对原始失效数据进行分布类型判别的新方法——概率图模糊判别法(PPFR)。该法通过对概率图上的数据点进行最小二乘拟合,并结合对拟合残差的分析,从中选定一组评价拟合程度的指标作为评判因素集,以五种常用的失效分布作为评判对象集。在此基础上,应用模糊综合评判法进行分布类型的综合识别,从而获得原始失效数据的最佳分布类型。

Fuzzy Recognition of Probability Plots for Failure Distribution

The recognition of the model for failure distribution is of great importance in the study of reliability. A new method called PPFR (Probability Plot Fuzzy Recognition) is proposed, which can be used directly to distinguish the model for failure distribution from the original failure data without any prerequisite assumptions. The basic idea as well as the principle of PPFR are described. Based on a general probability plot, the coordinate transformation corresponding to a distribution model is carried out first with the original failure data and the estimated value of cumulative distribution function and then linear least square fit is made for the transformed coordinates. After that, the multifactor fuzzy evaluation method is used to evaluate the fitting conditions for various failure distribution models, and a best method is chosen. This method makes full use of the simple and clear features of probability plot and has brought about an effective compensation for accuracy. It does not matter what type of the sample and how large it is and whether or not any unknown parameters are presented. An example is given together with the flowchart.