Improved Kriging-based hierarchical collaborative approach for multi-failure dependent reliability assessment

Reliability assessment considering multi-failure dependency brings in highly complex computing tasks, which is impracticable for some complex structures like turbine cooling blades. To efficiently investigate the failure dependency between multiple frail sites and multiple failure modes, an improved Kriging (IK) model is first established by optimizing model parameters with dynamic hybrid ant colony optimal algorithm, and the IK-based hierarchical collaborative (IK-HC) method is further developed by absorbing the advantages of IK model into HC strategy. Based on the multidimensional Copula function with dependency capture power and the IK-HC method with high-fidelity computing power, the multi-failure dependent reliability framework is first built. The computing efficiency and accuracy of the proposed IK-HC method are pre-verified by several numerical cases, and the method is applied to the multi-site (i.e., blade root, front edge) multi-mode (i.e., low-cycle fatigue, creep fatigue) failure dependent reliability assessment of a high-pressure turbine cooling blades. Method comparisons reveal that the proposed method can perform the multi-failure dependent reliability assessment with high accuracy and efficiency.