Multi-source Information Fusion Method Based On Bayes Small Sample Theory Research

According to the dominant characteristics of test analysis and assessment in small sample circumstance, the Bayesian method in small sample statistical inference and fusion theory is adopted as an important and suitable approach. The main research work includes Bayesian multi-information fusion using credibility and data fusion of different environments in this dissertation.Firstly, the problem of multiple a priori information fusion has been studied under the condition of the same test environment, such as the consistency test of a priori information, the credibility analysis of a priori information, the calculation of credibility with information likelihood rate, the expression of fusion posterior distribution based on the credibility of multiple a priori information. Then, the method of multi-information fusion with Bernoulli distribution or exponential distribution has been studied by simulation.Secondly, the problem of test data fusion has been studied under the condition of different test environment. Through analyzing the disadvantages of former data fusion method with environmental factor, a truncated Bayesian fusion algorithm of environment factors has been presented. Then, the dynamic modeling method has been studied and used in fusion data under the condition of different environments from the failure mechanism. In addition, a Bayesian fusion method based on the proportion factors of mixed exponential distribution has been also presented in this dissertation.Finally, Bayesian fusion algorithm with Weibull distribution has been studied in detail by using He-Ne Laser life test data., which includes three aspects, that is, the selection of Weibull distribution model , the posterior distribution of Weibull parameters with multiple a priori information and the dynamic modeling method of Weibull distribution parameters under the condition of different test environments.