Research On HHT Method And Its Application In Seismic Reservoir Prediction

Seismic signal is a typical non-stationary signal which contains lots of geological information.Predicting reservoirs effectively with seismic data is becoming a research focus in geophysics.Time-frequency analysis method is a signal analysis method base on time-frequency joint domain,it can transform seismic signal from time domain of one-dimensional to time-frequency domain of two-dimensional,and reflect the time-varying characteristics of seismic signal frequency and energy.It can excavate the information hidden in seismic signal effectively,and applied to reservoir prediction widely.The time-frequency analysis methods based on Fourier transform have great limitations in processing seismic signals.Hilbert Huang Transform(HHT)is an adaptive analysis method for time-varying and non-stationary signals,which can reflect the time-varying characteristics of seismic data better,and greatly improve the time-frequency resolution.However,conventional HHT HSA end effect and mode mixing,it restricted development of HHT in geophysics.In this paper,the end effect and mode mixing are studied in depth.On this basis,some seismic attribute analysis method based on an improved fast ensemble empirical mode decomposition(IFEEMD)are proposed,and applied to reservoir prediction in Tahe block.In this paper,the Short-time Fourier Transform(STFT),Continuous Wavelet Transform(CWT),S Transform(ST)and HHT are compared by the simulation of analog signals.Calculation method of STFT is simple,but the time-frequency resolution is single;CWT can analyze signals with multi-scale,its spectrum energy is easy to leak;ST HSA multi-resolution and adaptability,but the time and frequency resolution of can't be optimized at the same time.HHT can decompose the signal into single component or narrowband signals adaptively according to the characteristics of the signal.It can take into account of both time and frequency resolution.The accuracy of time-frequency analysis is higher,but there are end effect and mode mixing.Aiming at the end effect and mode mixing of HHT,this paper studies the Piecewise Cubic Hermite Interpolating Polynomial(PCHIP)?end extending method based on signal local eigenvalue?ensemble empirical mode decomposition(EEMD)and Fast ensemble empirical mode decomposition(EEMD).Proposing an improved FEEMD(IFEEMD)algorithm,which effectively improves the end effect and mode mixing effectively,and improves the accuracy and efficiency of time-frequency analysis.Based on the IFEEMD,the methods of seismic instantaneous spectrum decomposition are studied.Proposing the seismic instantaneous spectrum analysis methods of IFEEMD-HSA and IFEEMD-TK energy operator.And the normalized energy absorption index is developed to visually reflect the difference of energy between low frequency and high frequency volumes of seismic signal.In order to improve the accuracy of reservoir prediction,this paper studies a feature extraction method based on spectrum in speech signal field.Mel cepstrum coefficient method is used to extract the corresponding parameters of seismic data based on the marginal spectrum of seismic signal obtained by IFEEMD-HT,and it obtain a new seismic attribute called seismic voiceprint attribute.Reservoir prediction based on single attribute often HSA multiple solutions.Fusing energy absorption index and seismic voiceprint attributes based on attribute fusion method.It improves accuracy of reservoir prediction.The results of this study improve the interpretation accuracy of seismic data.It provides new technical support for seismic reservoir prediction.It has achieved good results in the reservoir prediction of Tahe block.It effectively improves the prediction accuracy,and reduces the exploration risk.