Research On Assembly Level Accelerated Testing For Storage Life Assessment Of Equipments

The main characteristic of some weapon equipments, such as missiles, torpedoes, is so call“long-term storage, one shot”. In other words, those equipments spend the majority of time in storage environment during their whole life cycle. Therefore, storage life assessment is the key point in the process of life decision and life extension, which plays an important role on improving usable rate and exerting benefit of equipments. Storage life assessment based on field storage testing, which is the traditional method, has some disadvantages, such as time consuming, high expenditure and weak prediction. Accelerated testing provides a feasible way for assessing storage life faster relatively, and becomes more and more popular in the field of storage reliability of equipments.To meet the requirement of storage life assessment mentioned above, this dissertation proposes that accelerated testing should be applied to assembly level products in storage life assessment of equipments, by compromising of assessment effect and maneuverability. The dissertation carried on a systemic research on a set of problems of assembly level accelerated testing (ALAT), including mechanism and model, data analysis, consistency analysis and fusion evaluation, optimal design, which provides technical support for storage life assessment of equipments. The main contributions are summarized as follows:1. As foundation of the dissertation, mechanism of storage life assessment by ALAT is analyzed firstly. And then ALAT models, such as life model and degradation model, are established to describe the relation between accelerated stresses and storage life. These models can be used in constant stress ALAT, step stress ALAT and constant stress ALAT of old product, which is referred to the samples put into ALAT has been stored for a period of time in normal circumstance.2.Methods for data analysis of constant stress ALAT, step stress ALAT and constant stress ALAT of old product are presented respectively.(1) Data of constant stress ALAT in the common form (hard failure mode and degradation failure mode may exist in the same time, and data might be incomplete) is described and basic steps of data analysis are established. Maximum likelihood estimation (MLE) based on the expectation maximization (EM) algorithm is proposed for statistic analysis of hard failure mode. Results turns out that, the EM algorithm are more accurate and stable than the direct maximum algorithm which is used generally. Pseudo lifetime MLE and least square estimation (LSE) for analyzing degradation failure mode are researched respectively. And comparison between the two methods is carried out, which turns out that pseudo lifetime MLE is better than the other.(2) On the basis of researches on constant stress ALAT, data analysis of step stress ALAT and constant stress ALAT of old product are proposed by considering the cumulated effect of different level stresses, and examples are demonstrated to confirm the validity, which provide technical support for data analysis for main types of ALAT.3. Several criterions are presented to measure the consistency of analysis results from ALAT and field storage. Then Bayesian method and MLE method for fusion evaluation are proposed respectively to improve the consistency.(1) Bayesian methods for fusion evaluation based on Gamma prior distribution and Dirichlet prior distribution are proposed respectively, which can be used to combine ALAT data and several aspects of information from similar product or expert experience. Analysis results turn out that, the methods can improve the consistency effectively in the case of small sample size in ALAT.(2) MLE method for fusion evaluation is proposed to combine data from ALAT of old product and field data. Validity of the method is confirmed by simulation analysis and examples, which prove that the method can improve the consistency effectively by combining field data.4. Main flow of design and optimizing for ALAT is constructed, within which two key problems, quantitative analysis of dominant failure modes and optimal design of test plan, are studied respectively.(1) MLE and Bayesian methods for quantitative analysis of dominant failure modes are proposed. Simulation analysis and examples turn out that, comparing to frequency analysis method which is generally used, MLE method is a little bit better and suitable in the case of large sample size. Because of the ability of combining other information, Bayesian method can distinguish dominant failure modes effectively in the case of small sample size.(2) Simulation based optimal design for ALAT is presented, which can optimizing test elements to get better assessment effect. Results prove that, the method overcomes limitation of complicated procedure in analytical solutions of optimal design, which is very suitable for application.5. The methods of this dissertation for ALAT are applied to storage life assessment of some pilot-operated safety valve, which provides an engineering case for the study of ALAT. Results turn out that, assessment from component level accelerated testing is quite different from ALAT. Comparing to component level accelerated testing, ALAT provides more accurate assessment result, because the testing condition and technical configuration in ALAT are closer to the real state.In summary, drawn by application requirements in storage life assessment of equipments, theories and techniques of ALAT are studied systematically in this dissertation by means of theoretical analysis, simulation analysis and application. Several important theoretical and technical questions about model, data analysis, fusion evaluation, and optimal design, are systematically studied. The research results are meaningful for the application of ALAT in storage life assessment of equipments.