Probabilistic analysis of blade flutter based on particle swarm optimization-deep extremum neural network

Abstract The improvement of aero-engine performance has posed a serious threat to aeroelastic stability, thereby compromising the reliability of aero-engines. An effective approach to quantify the risk of compressor blade instability and enhance aeroelastic stability is through flutter probability evaluation. This study proposes a prediction method called the Particle Swarm Optimization-Deep Extremum Neural Network model (PSO-DENN) to improve the modeling accuracy and computational efficiency of compressor blade flutter probability analysis in aero-engines. Through deterministic analysis, the flutter response distribution of the blade is obtained. To account for the randomness of boundary conditions and time-varying loads, the flutter reliability of compressor blades is evaluated, providing insights into distribution characteristics, and reliability associated with aeroelastic instability. Comparative analysis of different methods demonstrates that the proposed PSO-DENN method improves calculation efficiency while ensuring accuracy.