LAUSR

Abstract Compressor blade seriously affects the reliability and stability of aircraft engine. Probabilistic flutter evaluation of compressor blade can effectively quantify failure risk and improve aeroelastic stability. To improve the computational efficiency of probabilistic flutter evaluation, a surrogate modeling approach (called as PSO-LSSVR) is presented by absorbing the strength of particle swarm optimization (PSO) algorithm and least-squares support vector regression (LSSVR). We first mathematically modeled the PSO-LSSVR and then introduce the corresponding probabilistic flutter framework. With respect to high-nonlinearity and strong coupling of limit-state function, the probabilistic flutter evaluation of NASA Rotor 37 is performed to evaluate the proposed approach. Compared with Monte Carlo method, quadratic polynomial (QP), regular support vector regression (SVR and LSSVR), the presented PSO-LSSVR possesses the highest computing efficiency while keeping acceptable computing precision.