Modal Parameter Identification Of Signal With Big Noise And Close Modes

This paper studies on improving identification accuracy of wavelet transform based modalparameter identification method, when the test signal is short data sequence with low signal to noiseratio and close modes. After several key issues of this method are researched, it is able to preciselyestimate modal frequency and damping ratio to solve engineering problems better.The main contents of this paper are as follows:1. An improved Morlet wavelet is adopted in wavelet transform based modal parameteridentification method. And then, several important issues, such as wavelet parameters, time-frequencyresolution, edge-effect, fitting interval, are discussed for optimizing parameter selection of waveletfunction and automatically selecting fitting interval. The simulation results show this method iseffective on improving the identification accuracy of modal frequency and damping ratio.2. Identifing each mode in operational conditions is difficult, when the test signal is short datasequence with close modes and weak signal submerged in noise. Consequently, a combine methodbased on transmissibility function and correlation function is presented. Transmissibility functionapproximate to FRF is calculated before achieving wavelet transform based modal parameteridentification. This combine method is able to obtain high precision modal frequency and to improvethe identification accuracy of damping ratio. Especially, transmissibility is good at identifying weakmodes and close modes.3. A new method based on chirplet transform and Carmona ridge extraction is presented toenhance signal to noise ratio and to extract signal features for linear sweeping signal. After signalreconstruction according to location of ridge line and chirplet coefficients, a good FRF is achieved toimprove modal identification accuracy.4. Modal identification software based on above algorithms is developed using GUI interfacetools in Matlab. It can be easily applied on both input-output system and output-only system. Theeffectiveness of the software has been confirmed through numberical simulation and experimental teston a frame model.