关键词: 转子系统, 振动, 可靠性, 灵敏度, 神经网络

Abstract: Aiming at the random rotor system with complex constructure in practical engineering, taking the influence of bearing, gyroscopic moment and mass unbalance on vibration characteristics of rotor system as considerable factor, the explicit relationship expression between random response and original random variables was obtained by integrating Finite Element Method (FEM) and Artificial Neural Network (ANN) technique. According to the assumption that the maximum amplitude peak of unbalance response should be no more than the allowable amplitude peak, the reliability mode was defined, and the formula of reliability-based sensitivity was given under the condition that the first four moments of basic random variables was known. The effect of working parameter′s randomness on rotor system vibration's reliability was studied, and the sensitivity of reliability to the mean value and variance of original random variables was obtained. The results indicated that the working speed and performance parameters of intermediate-pressure compressor left bearing were main influencing factors of vibration. In practical working process, the changes of these parametes should be control and monitor strictly.

Key words: rotor system, vibration, reliability, sensitivity, artificial neural network, finite element method