A model and knowledge-driven reliability optimisation approach for complex systems

Complex systems' low reliability is often caused by information uncertainties like component coupling, fault propagation and fault modes. Accurately identifying and optimising these uncertainties is vital for systems' optimal operation. However, their evolution patterns are hard to describe precisely, and the current data-driven mode is often resource-intensive. Consequently, inefficient information extraction and knowledge reuse in complex systems have led to suboptimal reliability optimisation decisions. To address this challenge, a model and knowledge-driven reliability optimisation approach is proposed in this paper. Firstly, a model-driven reliability analysis approach combining SysML modelling language of model-based system engineering (MBSE) with fault mode and effects analysis (FMEA) is proposed to support accurate fault extraction information and reliability assessment. Then, an ontology-based knowledge representation model and a reliability optimisation-oriented knowledge graph are developed to efficiently manage and reuse the reliability knowledge derived from historical operation and maintenance (OM) data of complex systems. After that, the naive Bayes classifier is used to perform probabilistic fault cause reasoning, providing quantitative guidance for reliability optimisation decisions. Finally, an application scenario study of an avionics system demonstrates that the proposed approach improves the display control unit reliability of the avionics system by 22%.