Study on Typical Fault Detection Method for Insufficient Turbine Rotor Fault Samples
Currently,deep learning has achieved greater popularity and development in fault diagnosis of mechanical systems,and these intelligent models require a large amount of training data to ensure their generalization capability.However,the lack of or difficulty in obtaining actual turbine rotor fault data poses a new challenge for intelligent fault diagnosis.In this paper,a method for generating fault data and performing intelligent fault diagnosis based on numerical simulation of turbine rotors is proposed.By building a finite element model of the rotor,the fault information which reflects the operating condition of the rotor is generated to provide data samples for the intelligent model.Then,a high-precision rotor finite element model based on actual rotor signals is established,which can effectively solve the problem of insufficient fault samples and increase the accuracy of intelligent diagnosis.Through the combination of finite element technique and deep convolutional neural network,the proposed method can realize the end-to-end intelligent fault diagnosis of turbine rotors under the condition of insufficient fault samples and the difficulty of measuring some faults,meanwhile it has the advantages of high accuracy and strong robustness.
国家科技期刊平台
NSTL
万方数据
