Study On Analysis Methods For Reliability Data Of Civil Airplane Based On Information Fusion

During the process of design and assessment of civil aircrafts, special attention was paid toreliability information. However, on the one hand, we are in the lack of reliability data in the designand assessment of civil aircrafts. On the other hand, during the design, manufacturing, test and usageof civil aircraft, a large number of reliability information was produced, and even similar productsreliability information was also available. Therefore, how to use this reliability informationcomprehensively and rationally is very important.Surrounding the utilization of reliability information, based on information fusion, followingtasks were finished in this paper.Focused on analysis of multi-sourced reliability data, the weighted integration method, entropymethod, Bayes method was studied. Methods to determine the prior distribution and to integrate priordistributions were put forward, based on Bootstrap approach and marginal distribution separately. Therationality of using a linear weighting method to merge multi-sourced prior distributions wasdemonstrated.Improved AHP was used to fusion reliability data.A new approach was provided toimprove the consistence of inconsistency matrix.Aimed at analysis of multi-phased reliability data, based on Bayes law in succession, a model wasestablished for multi-phase reliability data analysis.How to decide the phase transfer factors wasstudied. When the reliability grows with phase conducting, a new approach based on AMSAA modelwas available. Finally, an example was given to put this method into application.To take out analysis of multi-environmental reliability data, environmental conversion factorswere used to convert these data under different environments to the same real one. Basic assumptionsand model were established to definate this factor. Point estimation and interval estimation methods ofenvironmental convert factors for normal distribution were studied. Simulation was carried out tocompare and analynize these methods.Finally, a system aimed to analynize reliability data based on information fusion was established.