Fault Diagnosis for Marine Diesel Engine Cylinder Liner-Piston Rings Based on Improved EEMD-MB1 DCNN
Aiming at the problems of non-linear and non-stationary vibration signals of marine medium-high speed diesel en-gine cylinder liner-piston rings,the similar time and frequency domain characteristics of vibration signals,difficulty in fault iden-tification for the same type of faults with different damage degrees,the vibration signals was used to identify the faults,and a new end-to-end cylinder liner-piston rings fault diagnosis method was set forth based on ensemble empirical mode decomposition(EE-MD)and multi-block 1-D convolutional neural network(MB1DCNN).Through designing IMF information quality screening cri-teria,the IMFs through EEMD decomposed were reordered,to obtaine the reconstructed signals containing more salient fault fea-ture components,which were input into the MB1DCNN network model.The performance of the designed IMF information quality screening criteria and the model were evaluated by vibration signal's analyzing and comparing with existing methods.Experimen-tal results showed that this method can accurately and effectively identify the fault type of cylinder liner-piston rings.It has high accuracy in fault diagnosis of the same type of faults with different wear degrees for the wearing parts,fault feature extraction and classification can be carried out effectively.
万方数据
维普
