Time-Frequency Attributes Extraction Methods And Application

Time-frequency analysis is a new signal processing method in recent years, itprovides the signal transform between time domain and frequency domain, and canclearly describe the energy distribution which gathering near the instantaneousfrequency in particular time. Seismic attributes used in seismic data interpretationbecome more and more with the development of seismic attribute analysis technology,but the traditional attributes, such as instantaneous amplitude, instantaneous frequency,and instantaneous phase, are still the basis of science study and calculation, and otherattributes are associated with them. So this paper takes time-frequency attributesanalysis as the breakthrough point to research, and on the basis of studying theconcepts and properties of time-frequency analysis methods, compares thecharacteristics of various methods, and chose the reasonable method to analyze theseismic signals according to specific geological tasks and characteristics of seismicdata. Finally extracting precise seismic time-frequency properties to interpret theseismic data and predict reservoir of oil and gas.This paper firstly analysis the basic theory of several time-frequency analysismethods, including short-time Fourier transform, wavelet transform, S transform,Wigner-Vill distribution and adaptive optimum kernel time-frequency analysis.According to these theory researches, write the programs of numericalimplementation, and compare the effects of various time-frequency analysis methods.The resolution of short-time Fourier transform is fixed, but STFT is calculated simply.Wavelet transform has the characteristics of multiresolution analysis, but it easilyproduce information leakage when analysis the signals because of the multiplicity ofwavelet function. S transform, which absorbed the advantages of short-time Fouriertransform and wavelet transform, is a relatively good linear time-frequencyrepresentation; Wigner-Vill distribution has good focus degrees both in time and frequency, it also produce serious cross-terms, which confused the correctinterpretation of signal. Adaptive time-frequency analysis has the sametime-frequency localization accuracy as Wigner-Vill distribution, and not interferedwith cross terms of interference.Secondly, extract the instantaneous frequency of analog signals using the methodof spectral peaking on the basis of time-frequency analysis. Adaptive optimum kerneltime-frequency analysis has good resolution both in time and frequency, and is lessinfluenced by the cross terms, so it extracts the most accurate frequency values,whose duration time is corresponding to the time range of a single frequencycomponent best. Then establish the geological model of mudstone wedge form, andinverse the stratum thickness separately with spectrum decomposition techniquebased on S transformation and adaptive optimum kernel time-frequency method,which obtained the satisfactory effect.Finally, extract the time-frequency attributes of real seismic data, and analyze thedata according to the geological significance of these attributes, and predict thefavorable oil and gas reservoir in seismic interpretation with spectral decompositiontechnology. Results tested with well data show that the application of time-frequencyattributes in the practical seismic data interpretation achieved good effects.