Reliability modeling of modular k-out-of-n systems with functional dependency: A case study of radar transmitter systems.

• A new modular k -out-of- n system model with functional dependency is developed. • Bayesian network is used to model the reliability of modular k -out-of- n systems. • An algorithm is designed to automatically generate the CPTs of the BN model. • DBN model is developed to update the system reliability using observations. • A real-world radar transmitter system in the space launch site is studied. The k -out-of- n systems are among the most important redundancy structures in engineering practices, and their reliability assessment has been extensively studied in the past decades. However, components in a k -out-of- n structure are often subject to functional dependency (FDEP), in which component states are affected by other components' states in the system. In this article, we study a new system structure, namely modular k -out-of- n system with FDEP. In such a system, the failure of some specific components will disable some components in the k -out-of- n structure. Bayesian network (BN) models are used to construct the structure function of modular k -out-of- n systems. The parameters encoded in the graphical structure of the modular k -out-of- n system are automatically generated by a customized algorithm. Furthermore, a dynamic BN (DBN) is developed to update the reliability of modular k -out-of- n system dynamically when observation data are collected from either component or system level. The Birnbaum importance measure of the different types of components in the modular k -out-of- n system is also evaluated by the DBN model via inserting evidence of the components' states overtime. A real-world case of a radar transmitter system in the space launch site is studied to demonstrate the effectiveness of the proposed method. [ABSTRACT FROM AUTHOR]