Improved Statistical method for Batch determination, including a Life Data application example

摘要

The purpose of this paper is to demonstrate a new 90% Confidence method for the existing MRR-MLE Weibull slope comparison for batch identifications. This method is developed to improve the currently available statistical batch detection methods in Weibull analysis. In Life Data Analysis the failure distribution is not always homogeneous. Subgroups or mixture of failure modes can be in the data. A batch problem is often the reason for the data not being homogeneous. When the failed parts are not available for detailed batch detection examination, a statistical batch analysis can be applied. In this paper the standard 3 step batch analysis [2] is complimented with a 90% Confidence method for the MRR - MLE slope comparison to prove that the slopes are significantly different. The application of the 3 step batch analysis, including the new 90% Confidence method, is demonstrated by an example, showing the steps followed and the results.