De-Noising Method Of Seismic Exploration Data Based On Embedded Delay Coordinates

Seismic exploration is the most effective method for geophysical exploration of oil in the process of the exploration and development of oil and gas fields. The aim of seismic exploration is to make use of ground-based observations of the seismic records to study the changes of underground medium stratum lithology, from which we can deduce the spacial location, geometric shape and the changes of its internal physical parameters of the basic unit of geological Objects. To meet the needs of the rapid development of oil exploration, the following three techniques in seismic exploration technology must be further developed - three-dimensional exploration technology, three high (high signal to noise ratio, high resolution and fidelity) processing technology and reservoir prediction technology. Among them the anterior two technologies have been regarded as the two pillars of the seismic exploration technology; and the last technology is considered as perspective technology. In term of the seismic data processing, three-high-tech is the long-term goal of geophysicists, and enhancing the SNR of the seismic data is the basis of seismic data processing.In the practical production, the signal is often polluted by noise, which obstructs the reconstruction of the important information in the signal. It is true in all of the signal processing areas including radar, sonar, communications, seismology, as well as biomedical. Generally, SNR is the performance measurement of the Signal enhancement algorithms, and improving the SNR of the seismic data is the basis of seismic data processing. Many signal processing algorithms work well in the high SNR circumstances, but when the SNR dropped to a given threshold, the performances of most of them will deteriorate. Under such situations, if the algorithms can not provide a certain number of iterations, then pretreatment and filtering are needed to improve SNR.There are two signal enhancement approached for non-stationary signal corrupted by noise: one is the adaptive enhancement algorithms including LMS, LS and KF algorithm; the other is deterministic algorithms. Both of them have their own advantages and the scope of application: adaptive techniques are generally superior in performance to fixed methods when the signal statistics are nonstationary and unknown. Adaptive filters perform poorly in certain conditions, however. An example of this is filtering of a nonstationary signal whose spectral content changes quickly with time. The filter designed using LMS approach, for instance, may not adapt quickly enough to track the rapidly changing signal. In many biomedical signal processing applications such as electroencephalogram (EEG) data, the structure of the underlying signal is often unknown and too complicated to model accurately.D. Napoletani proposed a new signal enhancement algorithm - embedded delay coordinates algorithm in 2006, which is based on nonlinear dynamics theory to eliminate the Gaussian white noise. In order to extract the useful signal polluted by the noise, we use the embedded delay coordinates of the measured signal as a data-processing tool and then use Fourier embedded estimator in a window frame. Through the threshold, the noise in signals is filtered and useful signals are recovered. After processing the measured signals, experimental results show that the algorithm base on the embedded delay coordinate has good performance for signals polluted by a variety of the white noise. This algorithm has already been applied to voice signals area.Seismic data is an important information source for Geological exploration, and the random interferences always appear in the earthquake record accompanied with seismic signals. When these disturbances are overloaded, it will affect the authenticity and reliability of the seismic imaging seriously, and make the interpretation of seismic data more difficulty, thereby the geological exploration cost is increased. In this paper, the embedded delay coordinates algorithm is used for filtering seismic exploration data immerged in random noise, and it is found that this method can reduce the random noise in the seismic exploration data effectively, which means it has good prospects in the application of the seismic processing. In this article, the basic contents of the phase space reconstruction theory are introduced first, and then the principle information of the embedded delay coordinates algorithm, and the selection of corresponding parameters are covered as well. By processing the single-component and multi-component signals, the effectiveness of the algorithm is proven.Afterwards, the embedded delay coordinates algorithm is used to eliminate random noise in seismic exploration data in this thesis. First, practical characteristics of the seismic signal are introduced, and then seismic materials applicable to the algorithm in this paper are discussed. According to the properties of practical seismic signal, signal model which satisfies the reasonable constitution of recorded data is proposed. Secondly, simulation experiments on the synthesized signals are performed in seismic data with single-in-phase axis and multi-in-phase axis respectively.At the end of the article, practical seismic data are processed using embedded delay coordinates to test the feasibility and effectiveness of the model. Test results consistent with the theoretical analysis, and the simulation results meet the requirement. Simulation results how that the algorithm can reduce random noise effectively, restore the signal obliterated by the noise, and the restoration maintains the waveform characteristics of the original records basically. So the effectiveness and rationality of this algorithm when applied to seismic signal can be concluded.In the practical signal processing, there still exist some other problems, such as the reading and compensation of the signal. They all can be addressed by repeated experiments to obtain certain approaches. Practical signal processing results show that the artillery record of the resumption can express the location of the in-phase axis of original signals clearly and recover the wavelet Ricker obliterated by the noise roughly. Through the comparison of the wave shapes before and after filtering, 2D and 3D time-frequency distribution , it is seen that this method can reduce earthquake random noise in the seismic data and recover useful information effectively. Although a Gaussian white noise as a signal mix-type noise is used when discussing the signal model, the results of the practical seismic processing show that it is the feasible to use embedded delay coordinates algorithm to reduce random noise in seismic data.