Failure correlation evaluation for complex structural systems with cascaded synchronous regression

Abstract

Failure correlation evaluation of complex structural systems is a complex analysis problem involving multiple disciplinary analyses and multiple response objectives, whose high-nonlinearity performance function makes the traditional methods are difficult to acquire acceptable computing accuracy and efficiency. In this case, a cascaded synchronous regression (CSR) method is proposed by combining the multi-level decomposition cooperation strategy and the random regression tree modeling algorithm. In CSR modeling, the multi-level and multi-stage complex structural system is first decomposed into several correlated single-level and single-stage distributed subsystems; moreover, based on the random regression tree algorithm, the corresponding nonlinear mapping model for each distributed subsystem are established; finally, by performing the CSR-based system reliability simulation and Copula functions, the failure correlation relationships can be quantified. The effectiveness of proposed method is validated by the multi-site and multi-mode failure correlation evaluation of aeroengine turbine disc. The current efforts open up an effective way to realize the high-accuracy failure correlation evaluation of complex structural systems, and shed a light on the structural reliability design theory.