Multivariate ensembles-based hierarchical linkage strategy for system reliability evaluation of aeroengine cooling blades

Abstract

To improve the computing accuracy and efficiency of system reliability evaluation for aeroengine cooling blades, by fusing the benefits of multivariate ensembles model (ME) into the hierarchical linkage technique (HL), a multivariate ensembles-based hierarchical linkage strategy (ME-HL) is proposed. In the ME-HL modeling, the complex evaluation system is first decomposed into multiple subsystems (i.e., frail site and failure mode) by the developed HL strategy, after that the multiple output responses of subsystems are synchronously mapped by proposing the ME model, and the multi-level system reliability framework is finally built with the Copula-based correlation quantification. The reliability evaluation of a typical aeroengine turbine cooling blade is regarded as a case, to verify the effectiveness of the proposed strategy. From the reliability evaluation results, we observe that the acquired reliability degree considering the failure correlation (i.e., 0.9547) is higher than that of without considering the failure correlation (i.e., 0.9356). From the methods comparison, we discover that the computing efficiency of the ME-HL method is 86.89%, 80.12%, 80.63%, and 18.85% greater than that of the ANN, BT, ME, and BT-HL methods, respectively; and the computing efficiency of the ME-HL method is 12.47%, 11.26%, 6.11%, and 0.93% higher than that of the ANN, BT, ME, and BT-HL methods, respectively. The evaluation and comparison results demonstrate that the proposed ME-HL holds significant advantages in computing accuracy and efficiency in system reliability evaluation problems. The current study can shed a light on the complex multi-level structural system reliability evaluation.