Vectorial surrogate modeling approach for multi-failure correlated probabilistic evaluation of turbine rotor

For complex structures like aeroengine turbine rotor, its reliability performance is jointly determined by multiple correlated failure modes. Probabilistic evaluation is an effective way to reveal the output response traits and quantify the structural reliability performance. However, for the requirement of evaluating the multivariate output responses and considering the correlation relationships, the multi-failure correlated probabilistic evaluation often shows the complex characteristics of high-nonlinearity and strong-coupling, leading to the conventional evaluation methods are hard to meet the requirements of accuracy and efficiency. To address this problem, a vectorial surrogate model (VSM) method is proposed by fusing the linkage sampling technique and model updating strategy. First, the linkage sampling technique is developed to build the vectorial sample set and the initial VSM by collaboratively extracting multidimensional input variables and multivariate output responses; moreover, the model updating strategy (MU) is presented to find the optimal undetermined parameters and construct the final VSM by addressing the issues of premature convergence and over-fitting problems. Regarding a typical high-pressure turbine rotor with multiple correlated failure modes (i.e., deformation failure, stress failure, strain failure) as engineering application case, the response distributions, reliability degree, sensitivity degree, correlation relationships for each/all failure modes of turbine rotor are obtained by the proposed method. Through the comparison of methods (direct Monte Carlo simulation, polynomial response surface, random forest, support vector regression, artificial neural network, VSM-I (without MU strategy), VSM-II (with MU strategy)), it is verified that the proposed VSM method can efficiently and accurately accomplish the multi-failure correlated probabilistic evaluation.