CN-122020185-A - Acceleration coefficient interval estimation system

CN122020185ACN 122020185 ACN122020185 ACN 122020185ACN-122020185-A

Abstract

The invention discloses an acceleration coefficient interval estimation system, which is characterized in that sufficient complete Failure sample data is obtained through Test-to-Failure Test, a double-layered Bootstrap sampling rule of a stress layer and a degradation characteristic layer is designed to generate a plurality of groups of Failure sample combinations, meanwhile, whole-course degradation data of the Failure samples under double stress are integrated, a cross-stress integrated joint likelihood function is constructed, acceleration coefficients of each group of combinations are directly solved, and finally, the acceleration coefficient interval is determined through statistical analysis. By adopting the technical scheme of the invention, the problems that the existing acceleration coefficient estimation can only realize point estimation, effective interval estimation without adapting Test-to-Failure data characteristics, interval results cannot cover Failure scene parameter fluctuation, and error accumulation exists in traditional step-by-step solution are solved, and the accuracy of the acceleration coefficient estimation and interval reference value are improved.

Inventors

WANG HAOWEI
Sang Shuohai
LI YANG
HAO WEIXIA

Assignees

杭州市北京航空航天大学国际创新研究院(北京航空航天大学国际创新学院)

Dates

Publication Date: 20260512
Application Date: 20260413

Claims (7)

1. An acceleration factor interval estimation system, comprising: the first processing module is used for acquiring failure sample data; the second processing module is used for carrying out wiener process modeling according to the failure sample data, integrating the degradation increment and the failure time of all the failure samples under low stress and high stress, and constructing a cross-stress integrated joint likelihood function; And the third processing module is used for sampling the failure sample data through the double-layered Bootstrap to obtain a plurality of groups of failure sample combinations, and obtaining an acceleration coefficient interval based on the integrated joint likelihood function.
2. The acceleration factor interval estimation system of claim 1, wherein the first layer is a stress layer and the second layer is a degradation feature layer.
3. The acceleration factor interval estimation system of claim 2, wherein the second hierarchical partitioning rule is to partition the samples into short, medium and long failure time sub-layers according to failure time average values of all the effective samples in each stress layer, and if the number of samples in a certain sub-layer is less than 1, the samples are merged with adjacent sub-layers to ensure that each sub-layer has effective samples.
4. The acceleration factor interval estimation system of claim 3, wherein the double layered Bootstrap sampling is to randomly extract 1 sample from each sub-layer of the low stress layer, the total extraction number m is equal to or greater than 2, and the high stress layer extracts m samples according to the same rule to form 1 group of "low stress and high stress" failure sample combinations, and the sampling mode is replaced sampling.
5. The acceleration factor interval estimation system of claim 4, wherein the third processing module samples the failure sample data through a bi-layered Bootstrap, the bi-layered including a first layered divided by acceleration stress levels, a second layered divided by time-to-failure averages of each stress layer, and extracts samples from sub-layers of each layered to form a plurality of groups of failure sample combinations, directly solving the acceleration factors of each group of combinations through the integrated joint likelihood function, and obtaining the acceleration factor interval through statistical analysis.
6. The acceleration factor interval estimation system of claim 5, wherein the first processing module obtains Failure sample data through Test-to-Failure testing.
7. The acceleration factor interval estimation system of claim 6, wherein the parameters under stress are 、 The parameters under high stress are related by the acceleration coefficient A: Wherein A is the acceleration coefficient between two different stress levels, As a result of the drift coefficient, Is the diffusion coefficient; the integrated joint likelihood function is constructed as follows: wherein n represents the number of samples at stress level 1, m represents the number of samples at stress level 2, Representing the amount of degradation of the ith sample at stress level 1, Representing the amount of degradation of the ith sample at stress level 2, A measurement interval representing the performance of the ith sample at stress level 1, Representing the measurement interval of the i-th sample property at stress level 2.

Description

Acceleration coefficient interval estimation system Technical Field The invention belongs to the technical field of accelerated degradation experiments and reliability tests, and particularly relates to an acceleration coefficient interval estimation system. Background The acceleration coefficient is used as a core parameter for connecting high acceleration stress and data under normal stress, and the reliability of an estimation result directly determines the accuracy of equipment reliability assessment. The Test-to-Failure Test can obtain complete degradation track from initial performance to Failure and Failure sample data by continuously applying stress to the Failure of the sample, and provides a high-quality data source for acceleration coefficient estimation. However, the existing acceleration factor estimation techniques have the following key pain points: the interval estimation method is missing, that is, the traditional technology can only realize the estimation of the acceleration coefficient points through a random process parameter relation, has no interval estimation scheme aiming at Test-to-Failure samples, and cannot quantify the fluctuation range of the acceleration coefficient, so that the subsequent reliability estimation lacks risk boundary references; 1. The sampling method is not adaptive, if the existing interval estimation adopts simple random sampling, the specificity of Test-to-Failure 'Failure samples' is not considered, the sampling combination possibly contains non-Failure samples, parameter fluctuation under a real Failure scene can not be reflected, and the interval result has low reference value; 2. The interval coverage is insufficient, the traditional interval estimation is mostly based on theoretical distribution assumption (such as normal distribution), and statistical characteristics of Test-to-Failure actual Failure data are not combined, so that the fluctuation deviation of acceleration coefficients of a theoretical interval and an actual Failure scene is large, and the coverage rate is low; 3. the verification mechanism is lack, and the validity verification standard aiming at the interval estimation result is lacking, so that whether the interval actually covers the acceleration coefficient true value can not be judged, and the reliability of interval estimation is further reduced. Disclosure of Invention In order to solve the problems in the prior art, the invention provides an acceleration coefficient interval estimation system. In order to achieve the above object, the present invention provides the following solutions: an acceleration factor interval estimation system, comprising: the first processing module is used for acquiring failure sample data; the second processing module is used for carrying out wiener process modeling according to the failure sample data, integrating the degradation increment and the failure time of all the failure samples under low stress and high stress, and constructing a cross-stress integrated joint likelihood function; And the third processing module is used for sampling the failure sample data through the double-layered Bootstrap to obtain a plurality of groups of failure sample combinations, and obtaining an acceleration coefficient interval based on the integrated joint likelihood function. Preferably, the first layer is a stress layer and the second layer is a degradation feature layer. Preferably, the second layered division rule is to divide the samples into a short failure time sub-layer, a middle failure time sub-layer and a long failure time sub-layer according to the failure time average value of all the effective samples in each stress layer, and if the number of samples of a certain sub-layer is less than 1, the samples are combined with the adjacent sub-layers to ensure that the effective samples of each sub-layer are all available. Preferably, the double-layered Bootstrap sampling is carried out by randomly extracting 1 sample from each sub-layer of a low-stress layer, wherein the total extraction number m is more than or equal to 2, and extracting m samples from a high-stress layer according to the same rule to form 1 group of 'low-stress and high-stress' failure sample combinations, and the sampling mode is substitution sampling. Preferably, the third processing module samples the failure sample data through a double-layered Bootstrap, the double-layered Bootstrap comprises a first layered part divided according to acceleration stress level and a second layered part divided according to failure time average value of each stress layer, samples are extracted from sub-layers of each layered part to form a plurality of groups of failure sample combinations, acceleration coefficients of each group of combinations are directly solved through the integrated joint likelihood function, and an acceleration coefficient interval is obtained through statistical analysis. Preferably, the first processing module acquires Failure