Fault Diagnosis Strategy for Complex Systems Based on Multi-Source Heterogeneous Information Under Epistemic Uncertainty

Abstract

Technological innovation in modern systems has significantly improved their performance. However, fault characteristics such as epistemic uncertainty and dynamic failure modes often occur when these systems break down, which greatly raises some new challenges in fault diagnosis. A new fault diagnosis strategy for complex systems is presented based on multi-source heterogeneous information considering epistemic uncertainty in this paper. Specifically, in view of the epistemic uncertainty, the failure distribution parameters of basic events are described with interval numbers and test cost of these events are evaluated using domain experts and intuitionistic fuzzy linguistic set; Aiming at the problem of dynamic failure modes, a dynamic fault tree (DFT) is used to establish the dynamic failure model and is converted into a dynamic evidential network to calculate some reliability parameters; Furthermore, a diagnostic decision table is constructed based on multi-attribute heterogeneous information such as test cost and some reliability results; Finally, a novel fault diagnosis strategy is designed based on distance-based VIKOR algorithm, which can provide some decision support for fault diagnosis and locate the fault as quickly as possible.

Keywords

• Intuitionistic fuzzy linguistic set
• reliability assessment
• diagnosis strategy
• D-S evidence theory
• VIKOR algorithm