Trace model based abnormal degradation data inspection and analysis method

Abstract

A trace model based abnormal degradation data inspection and analysis method includes steps of 1), firstly inspecting homogeneity of error variances when the error variances are unequal and relation thereof is unknown, and then inspecting the same with subsequent methods when two groups of the error variances have no significant differences; 2), acquiring point estimation of the error variances of model coefficients of two groups of degradation data by means of maximum likelihood estimation in case that abnormal data can return, and subjecting construction statistics to comply with a distribution F and judging whether the significant differences exist between the abnormal data and entirety or not by means of hypothesis inspection; 3), analyzing the model coefficients of data in case that the abnormal data cannot return, subjecting the construction statistics to comply with another distribution F, and utilizing the hypothesis inspection to judge whether the significant differences exist between the abnormal data and the entirety or not; 4) judging whether the abnormal degradation data are brought by differences between model intercept a of the two groups of degradation data or by differences between slope b of the two groups of degradation data, or by both.

Description

technical field [0001] The invention proposes a trajectory model-based inspection and analysis method for abnormal degraded data, which belongs to the technical field of reliability for abnormality inspection of degraded data. Background technique [0002] With the improvement of product reliability and service life, a large number of non-failure data appear, and the traditional reliability evaluation method based on failure data has been difficult to meet the requirements of engineering evaluation accuracy. In this regard, degradation data has been introduced into the field of reliability assessment and has been widely used, becoming an important technical approach to solve the reliability assessment problems of high reliability and long-life products. [0003] However, factors such as test errors, differences in manufacturing status or testing status will cause changes in product performance degradation characteristics, resulting in abnormal points, abnormal segments or ab...

AI Technical Summary

Problems solved by technology

[0003] However, factors such as test errors, differences in manufacturing status or testing status will cause changes in product performance degradation characteristics, resulting in abnormal points, abnormal segments or abnormal samples in the performance degradation data of the same group of products

These abnormal data cannot reflect the real degradation law of the product and cannot be directly used for product reliability evaluation

However, due to the dispersion introduced by factors such as materials, manufacturing processes, and test errors, product performance degradation data must be random

Images