Improved resonance reliability and global sensitivity analysis of multi-span pipes conveying fluid based on active learning Kriging model

In this paper, the improved resonance reliability and global sensitivity analysis of multi-span pipes conveying fluid is proposed via active learning Kriging model. The natural frequency of multi-span functionally graded materials pipes conveying fluid is calculated using the dynamic stiffness method. A new improved resonance performance function is proposed which is suitable for both broadband and multi-frequency resonance failure. This paper extends resonance reliability to the global sensitivity analysis, and proposes a moment-independent global sensitivity index based on the resonance failure probability. An importance ranking of random variables is obtained, which provides vital guidance and advice for the optimal design of pipe anti-resonance. A new resonance reliability analysis method via the active learning Kriging model is established, which greatly improves the application of pipelines resonance reliability analysis in engineering practice. Based on active learning Kriging model, the effects of fluid velocity, pressure and volume fraction of functionally graded material on resonance reliability are analyzed. The results demonstrate that the proposed method has high performance for the anti-resonance analysis of multi-span functionally graded materials pipes conveying fluid.