Research Of Geological Precdiction On Beach-Bar Sandstone Reservoir

Now it is a hotspot to research thin-layer or thin interbed. Firstly, we have analyzed thecharacteristics in time domain and frequency domain of the reflection coefficient of thin-layerin theory, and tried to analysis new methods to describe the subsurface geologic structuremuch more truly and meticulous of beach-bar sandstone reservoir in frequencies domain.The basic idea of time-frequency analysis is to design the joint function of time andfrequency, which is used to describe the different times and frequency of the energydistribution that can make the local characteristics of the time-frequency two-dimensionalplane clearly displayed. This thesis starts from the Fourier transform to analyze theadvantages and limitations; then with an example of a short-time Fourier transform, wavelettransform and Wigner-Ville distribution, they are more classic methods of time-frequencyanalysis, but there are some problems in practical application. The recent emerged adaptivetime-frequency analysis, which separates the signal without the constraints of Heisenberguncertainty principle, have a good time and frequency resolution, and high noise immunity.Hilbert-Huang transform has ends flying wing phenomenon. In order to solve thisproblem, several approaches to restraining the end effect were proposed, and all have a certaindegree of restraint effect in analying seismic data. Matching pursuit is another method ofadaptive time-frequency analysis. In this thesis, we focus on the research of improving thecomputational efficiency of MP by using particle swarm algorithm and multichannelmatching pursuit. Through model analysis, the improved adaptive time-frequency analysishave good time-frquency performance and can accurately show the local characteristics of thesignal.Based on the adaptive time-frequency analysis, we have analyed several seismicprediction methods, such as dividing sedimentary cycles, multiwavelets theory and spectraldecomposition. All of the methods have been used in beach-bar sandstone, with comparison of multiple attribution clustering analyses. The results indicated that the technologies based onadaptive time-frequency analysis can be used to improve the prediction accuracy of reservoir.