Maintainability verification for airplanes with small samples based on similarity degree

Classical methods of maintainability verification usually requires a large number of samples. In reality, it is always difficult to obtain enough data to perform the maintainability verification for a variety of airplanes during the operational test and evaluation stage due to the high cost involved in the tests. Bayesian method has been recognized as an effective approach to deal with small sample problem. In the process of maintainability verification with small samples by using Bayesian method, how to obtain a set of reasonable weights for information from multiple sources is a key problem. In this paper, the similarity degree between the new airplane and its predecessors is introduced as the fusion weights of prior information. The prior distribution of the new airplane is obtained by fusing the historical maintenance data of the existing airplanes. The posterior distribution function of maintenance index of the new airplane, such as MTTR, can be updated by Bayesian method via a small sample of maintenance data of the new airplane. This paper proposed a suitable approach to make use of the historical maintenance information of existing airplanes as a compensation for the lack of sufficient samples of new airplanes. The method not only reduces the sample size and costs, but also guarantees the accuracy of maintainability verification. A case study is presented to demonstrate the proposed maintainability verification method.