The Application Of Generalized Demodulation Time-Frequency Analysis To Multi-Component Signal

In actual applications, most of the signals are time-varying, non-stationary and multi-component. For example, during the dynamic governing process of the induction motor fed withvariable frequency power, current harmonic is very much, and its frequency is time-varying,fault feature component also has this property. Most effective way of analyzing them is that thesignal is decomposed into some single-component signals whose instantaneous amplitudes andfrequencies are of physical significance, then analyze each single component signal, so that wecan precisely get the variation of the frequency components with time.Based on the analysis and summary of the common time-frequency analysis methods,aimed the deficiency of them, the theory and the application of the generalized demodulationtime-frequency analysis is further studied, the main research work and achievements are asfollows:First, we introduce some background knowledge of the generalized demodulationtime-frequency analysis, including the principle and some algorithms.Secondly, an new method to decompose the time-varying, non-stationary and multi-component signal into some single components.Thirdly, the proposed method is compared with the EMD time-frequency method. Thesoftware is provided for the algorithm.Fourthly, an formula is derived to estimate the instantaneous frequency when the signal ismeasured in the presence of additive white noise. The noise effect of the algorithm is analyzedon the correlation structure of the real and imaginary parts of the resulting complex vectors. Inthe presence of additive white noise, the simulation results demonstrate that there was nosignificant bias, except for some small fluctuations. Therefore, in order to eliminate theinterference, the original signal should be filtered before analyzing the signal.Finally, in order to verify the validity of the method, analyzed stator electric current signalsof the induction motor fed with variable frequency power in regular and diagnosis statesrespectively, on the one hand, converted the main power component to the constant frequencyand then filter it, on the other hand, separated the fault characteristics. Simulation shows that themethod can extract each signal component and highlight the fault features.