Research On The High Precision Time-frequency Analysis Method And Technology

The traditional analysis method of signal is Fourier transform, but to the typical non-stationary signal such as seismic signal, this method can't reach the requirement, it is need to use time and frequency domain to analysis. As the widely use of time-frequency analysis method in petroleum exploration, it's researched by many people.This research mainly through continuous wavelet transform and inverse spectral decomposition research to form high-precision time-frequency analysis method, then use these methods in theoretical calculation and practical data processing.In CWT, there are three important parameters: wavelet function, the space of scale and the step. Adopting different wavelet function to analyse the signal will get different resolution results; The scale is related to frequency, the much better choice of the scale range the better frequency results will be obtained; The choice of step is related to the partial feature of the signal. The result of wavelet transform is time-scale range, we can't obtain the information of frequency directly in the transform result; Also analysis the signal need to meet the reconstruction to let it has a practical significance. So I had studied the inverse problem about wavelet transform to certain the data processing has practical significance.The inverse spectral decomposition is based on inverse wavelet transform and constraint condition. The choice of wavelet function and constraint condition has important influence to the decomposition result. Compared with other methods such as short time Fourier transform and continuous wavelet transform this method has better time and frequency resolution. In inverse spectral decomposition, the paper mainly studied the influence of constraint condition. In theoretical models mainly used three constraints condition minimum L2 norm, minimum L1 norm and sparse spike or minimum support constraint. The accuracy of these constraints is raised, and the sparse spike has the best accuracy.This paper use accuracy time-frequency to the processing and processing of real seismic data base on the successful research of high precision time-frequency methods, and have a good application effect in frequency data interpretation and high resolution processing.