Research On Methods And Applications Of Reliability Growth Test For Solid Rocket Motor

During the development of solid rocket motor (SRM), the reliability growth tests are adopted to eliminate the weaknesses in SRM design or manufacture. As a result, the SRM reliability is improved gradually. The reliability growth test data of SRM is a dynamical statistic population, and it is difficult to get scientific and reasonable analysis results. Therefore, the SRM reliability analysis is an important theory problem of the SRM reliability growth test. In this dissertation, the SRM reliability growth test theory and methods are researched. Aimed at the theoretical and practical problems of the SRM reliability growth test, the reliability growth test management, programming and analysis methods are proposed, which make use of the diverse information synthetically.In this dissertation, the growth test programming methods are proposed as the efficient tools of scientific management and decision-making of the reliability growth test. Under the restrictions of the development cost and reliability, the two test programming models are put forward. The reliability forecast frame of SRM is proposed based on the fault modes. The development actions including test, manufacture and design modification are expressed mathematically, and the practical process of the reliability growth test is depicted in detail. The prediction of the reliability growth test quantity is achieved, and the test stop rule is proposed. With the application of the above methods and models, the test management and decision-making are scientized, and the disturbance of subjective factors is decreased.The stayed and instant modification strategies during the development of SRM are analyzed in detail in this dissertation, and two reliability growth models are developed to suit the different test modification strategies. The application of discrete AMSAA model and Nonhomogeneous Geometric (NHG) model are extended. The computational problem in the model application is solved by the adoption of sequential number-theoretic method for optimization. The difference of two models is deeply researched by theoretical analysis and numerical simulation, and the impact of model misspecification on the reliability growth analysis is discussed.Based on the strong inherited attribute in the development of SRM, the Bayesian reliability growth model is proposed. There are much history and expert information. To make use of the diverse information synthetically, the information fusion methods are put forward. Based on the Mazzuchi-Soyer model, the Bayesian reliability growth model is built. This model can not only estimate the reliability growth, but also predict the reliability of posterior development period, which overcomes the defect that the general Bayesian model can only estimate reliability growth. The model robustness and scientificity are analyzed by simulation. The model is used to analyze the SRM storage test, which is a new application domain of the Bayesian reliability growth model.In view of the SRM overstress test and overtime test, two reliability analysis models are proposed in this dissertation. The reliability analysis method of multi-overstress is introduced based on that of single overstress, and the Bayesian reliability growth model to analyze the SRM overtime test is proposed. The historical and expert information are synthetically applied, and the joint posterior distribution of the reliability is deduced for Type I censoring life test. The Gibbs method is adopted to solve the complex posterior distribution inference problem.In view of the 3-parameter Weibull distribution estimation, the compound estimation method is proposed. In this method, the Maximum Likelihood Estimation (MLE) and the Minimum Distance Estimation (MDE) are integrated to settle the location parameter estimator. On this basis, the mixed Weibull distribution is adopted to fit the multi-stresses competing failure dada, and the corresponding model parameter estimation method is proposed. The EM method is applied to estimate the parameters and mixture weights, and the Bootstrap method is adopted to compute the standard error. These methods are effective to fit the multi-stresses competing failure dada.To solve much data processing problems of the SRM reliability analysis, the SRM reliability growth evaluation system is developed, which has the function of reliability growth test management, the reliability assessment, reliability forecast of future test, overstress test reliability analysis, and so on. It is a powerful tool of SRM reliability growth test management and SRM reliability assessment.The research in this dissertation provides important guidance to the SRM reliability growth test programming, estimation and analysis, and has high practical value to improve the SRM reliability growth test scientificity.