Missile competing fault prediction based on degradation data and fault data

Aiming at the problem of missile competing fault prediction which has multivariate degradation data, the characteristics of missile degradation are analyzed and missile competing fault prediction model which has multivariate degradation data is established considering the correlation between sudden fault and degradation fault. The distribution patterns of performance degradation data and sudden fault data are determined through statistical inference, and on this basis the parameters for competing fault prediction model are solved. Aiming at the distribution parameters of performance degradation data having the feature of small sample and nonlinearity, least square support vector machine (LS-SVM) prediction algorithm is used to predict the distribution parameters of performance degradation data in order to get the future distribution function. Aiming at the correlation between sudden fault and degradation data, the correlation parameters of sudden fault and degradation data are obtained using location-scale model to analyze the relations between sudden fault and degradation data. Furthermore, the future missile competing fault probability can be obtained according to the missile competing fault prediction model. The efficiency and validity of missile competing fault prediction model are verified by case analysis and contrasting with other prediction methods. © 2016, Beijing University of Aeronautics and Astronautics (BUAA). All right reserved.