Model-based Systems Engineering (MBSE) is a noval approach to support complex system development by formalizing system artifacts and development using models. Though MBSE models provide a completely structural formalisms about system development for system developers, such large of domain specific knowledge represented by models cannot be captured as what the developers expect. This leads to a big challenge when MBSE can be widely used for complex system development. In this paper, a knowledge management approach is proposed to support an intelligent question answering scenario when implementing MBSE in system lifecycle. We make use of the GOPPRR approach to support MBSE formalisms which are transformed to knowledge graph models. Then such models provide cues for intelligent question answers through reasoning. In the case study, we make use of an auto-braking system scenario to develop MBSE models and to implement the intelligent question answering. Finally, we find the availability of our approach is evaluated which the domain engineers enable to capture their domain knowledge more efficiently.