Locomotive Traction System Fault Diagnosis Based On Fractional Fourier Transform Under Variable Speed Conditions

The fault diagnosis of locomotive traction system is the key to ensure safe operation and avoid major accidents.When running on the electric railway,passing over phasesplitter and the variation of line parameters may cause the locomotive to speed-varying condition.In this case,the signals of the traction system collected by sensors are timevarying and non-stationary.At the same time,the fault frequency will change continuously with the rotation speed,which makes the extraction of fault features quite difficult.In this thesis,Fractional Fourier Transform(FRFT)and Variational Mode Decomposition(VMD)are introduced into the analysis of locomotive vibration signals and motor current signals in speed-varying condition,and the following results are obtained:The fault diagnostic models of bearing and gear in speed-varying condition are discussed,and the signal characteristics of vibration signal and motor current signal in time-domain and frequency-domain in speed-varying condition are analyzed.The TimeFrequency Analysis method is introduced,and the time-frequency relationship of the fault characteristic frequency is given,which lays the foundation for the fault diagnosis in speed-varying condition.A fault diagnostic approach for locomotive traction system in speed-varying condition based on VMD-FRFT is proposed.First,the collected signal is decomposed by VMD,and then the most suitable intrinsic mode function is selected and analyzed using the FRFT.The optimal order and aggregation position of the target components are determined in the fractional domain.Finally,the parameters of frequency-modulated component are estimated and the fault characteristic frequency is extracted.Experiments show that the proposed approach can take advantage of the two Time-Frequency Analysis methods and extract the fault feature accurately.Aiming at the difficulty of the optimal order selecting of FRFT,an order selection method based on the hybrid entropy algorithm is proposed.By combining the sample entropy and the fuzzy entropy,the hybrid entropy is constructed to measure the ambiguity and randomness changes of signal components.The optimal order of energy convergence in fractional domain is determined by means of the principle of local minimum entropy.Experiments show the proposed method can select the optimal order of FRFT quickly and accurately.Based on the above research,a fault diagnostic system is developed and applied to the operation tests of locomotive in speed-varying condition.The vibration signal and motor current signal is analyzed and the faults of the traction system were diagnosed successfully.