Probabilistic framework for reliability analysis of gas turbine blades under combined loading conditions

This paper establishes a probabilistic framework for reliability analysis of gas turbine blades under combined high and low cycle fatigue (CCF) loadings. Initially, the dynamic reliability model of turbine blades with respect to load application times is developed by using the stress-strength interference (SSI) theory under combined loading conditions. Considering the expensive computing cost of the Monte Carlo simulation (MCS) integrated into finite element (FE) models (MCS-FE), an improved least squares support vector machines (ILS-SVM) approach is presented by employing the modified seagull optimization algorithm (MSOA) to seek for the optimal model parameters of LS-SVM. Subsequently, the distribution characteristics of maximum equivalent stress of gas turbine blades under the uncertainties involved in CCF assessment induced by working loads and material properties can be obtained with the built ILS-SVM model. Accordingly, the reliability estimation considering strength degradation is reached. The probabilistic framework is demonstrated via a numerical example of a turbine blade, and the results confirmed that ILS-SVM is an effective probabilistic analysis methodology holding high computing accuracy and convergence speed.