Remaining useful life prediction of weighted k-out-of-n systems based on dynamic random weights of importance

Abstract

With the increasing complexity of various industrial systems, the reliability assessment and remaining useful life prediction of weighted k-out-of-n systems considering stochastic correlation have become a concern in the field of reliability. The components’ varying locations and operating environments result in fluctuating reliability and different impacts on normal system operation. In this study, the Birnbaum importance of components is modeled for a weighted k-out-of-n continuous degradation system consisting of different types of components with stochastic correlation. Furthermore, a reliability model and a remaining useful life prediction model based on kernel diffeomorphism estimation with dynamic random weights of Birnbaum importance are developed for weighted k-out-of-n systems. First, the component state reliability model is established by a nonparametric method of adaptive kernel diffeomorphism estimation. Second, the multivariate copula function is used to characterize the stochastic correlation of different types of component degradation. In addition, the Birnbaum importance model with stochastic correlation of components was developed using Akaike Information Criterion for copula function preference. A remaining useful life prediction model for the weighted k-out-of-n continuous degradation system is established considering the dynamic random weights of Birnbaum importance measures. Finally, the validity and accuracy of the proposed weighted k-out-of-n system remaining useful life model are verified using a cluster of plant protection unmanned aerial vehicles.