Energy Operator Demodulation Approach And Its Application In Fault Diagnosis Of Transmission

Transmission is a vital component of automobile power train，its fault diagnosis andcondition monitoring can reduce accident rates and improve the vehicle performance. Orderanalysis method is an important approach to process non-stationary signals of rotatingmachinery. The method based on instantaneous frequency estimation without a specific ofcomplex hardware devices, can overcomes the problems of too expensive and inconvenientinstallation in traditional order analysis method. For a mono-component amplitudemodulation and frequency modulation (AM-FM) signal, its instantaneous amplitude andinstantaneous frequency can be obtained by Energy Operator Demodulation method.Compared with the conventional demodulation method-Hilbert transform, the EnergyOperator Demodulation has higher time-frequency resolution, better adaptability to transientcharacteristics and lower computational complexity. Therefore it is applied to obtain theinstantaneous frequency. However, Energy Operator Demodulation is subjected tomono-component signals and sensitive to noise, so studying a new Energy OperatorDemodulation Algorithm which is suitable to multi-component signal and strong resistance tonoise will be of great significance.The main consists are as follows:(1) The impact of the sliding filter length L on demodulation accuracy for SmoothEnergy Separation Algorithm is discussed, and points out that in the practical application ofSmoothing the energy separation algorithm, the best choice for the sliding window length isequal to the value of sampling frequency divided by the two times of the carrier frequency,and sliding window length may be increased when the signal containing large noise.(2) Because of Smooth Energy Operator Demodulation method subjected tomono-component signal, the multi-component signal separation technique based ongeneralized demodulation is studied. Based on the deep study of the limitations of traditionalgeneralized demodulation method, the complex analytical Gabor filter is used to overcome thenegative frequency interference, and a new estimation method of initial phase functionrequired by the generalized demodulation is proposed. Combining the improved generalized demodulation with Smooth Energy Separation Algorithm, an instantaneous frequencyestimation method suitable for noisy multi-component signal is proposed, which furtherexpands the application range of Energy Operator Demodulation. The results of simulationand engineering application examples show that the new method has higher accuracy thantraditional methods.(3) A fault diagnosis method for transmission running in non-stationary conditions isproposed, the instantaneous frequency estimation method based on generalized demodulationand Energy Operators is combined with order analysis to diagnose the fault features of gearsand bearings in transmission. The results of simulation analysis and experimental verificationshow that the proposed method can identify the teeth number of fault gear and the gear shaft,and can extract the fault feature order coefficient of bearing effectively.