Theory And Method Of Step-down-stress Accelerated Life Testing

The quantitative evaluation for high-reliability and long-life has been a technical problem in reliability engineering for a long period of time. With the rapid development of science and technology, the level of reliability is enhanced continuously, and this problem is more stringent increasingly. The accelerated life testing (ALT) technology, developed in 1960s, provides a possible solution for this problem. The theory, method, and application of ALT have been given more and more research concerns in recent years.Aimed at the basic requirement for high-reliability and long-life evaluation, this dissertation carries on a systemic research on the theory and method of a new kind of step-down-stress ALT (SDS-ALT), to improve the efficiency of the present ALT, and to provide a set of high-reliability and long-life evaluation testing method eventually. The main contents are as follows:1. A new ALT method, viz. SDS-ALT, is brought forward for the first time for high-reliability and long-life evaluation described by Weibull distribution, and theory models and numerical simulation are also presented to illustrate its efficiency principle.(1) In the research of theory model, cumulative degradation model (CDM) of failure mechanism and cumulative exposure model (CEM) of probability statistic for SDS-ALT are established, and the efficiency principle for SDS-ALT is qualitatively analyzed based on them.(2) The efficiency principle for SDS-ALT is also studied by Monte-Carlo simulation. The results reveal that SDS-ALT is more efficient than the widely used step-up-stress ALT (SUS-ALT) when applicable to this kind of life evaluation, and that it may reduce testing cost consequently.2. The fundamental problems in statistical analysis for SDS-ALT, such as the theory of acceleration factor, the recognition of failure mechanism drift, and the method of failure detection, are investigated.(1) From the basic idea of failure time conversion under equal degradation accumulation delivered by CDM, the statistical inferences for acceleration factor are discussed under common life distributions. The statistical expressions for acceleration factors are gained, and it is proven that the restriction for distribution parameters is the indispensable condition of failure mechanism consistency.(2) In the research of failure time detection for life testing under Weibull distribution, an iterative estimation method is established to remove the theory error of the present method, and to improve the detection precision. This problem is an effective testing data acquisition actually.(3) The recognition of failure mechanism drift is an effective testing data validation problem in ALT. The statistical recognition method is studied by the indispensable condition of failure mechanism consistency, and the probability plot test is applied in SDS-ALT to solve the3. The statistical analysis method for SDS-ALT under Weibull distribution occasion is further studied.(1) The conversion analysis method based on accelerated model, which is the main statistical analysis method in SUS-ALT, is applied to SDS-ALT at first. It reveals that the method is applicable to SDS-ALT under simple accelerated model (Inverse Power Rule Model and Arrhenius Model), and that the complexity of the method will leap under complex accelerated models, and that the method is difficult to software realization and engineering application.(2) To simplify the above method, 3-step analysis method for SDS-ALT under Weibull distribution occasion is put forward, which analyzes SDS-ALT by converting the data into equivalent constant stress ALT data. The analysis procedure of 3-step analysis method is independent of the complexity of accelerated model, so it is applicable to SDS-ALT under complex accelerated models. 3-step analysis method is of procedural characteristic, easy to software realization.(3) The data conversion of 3-step analysis method is discussed in detail, and a conversion analysis method based on acceleration factor is...