Fault Detection Technology Using Seismic Data Gradient Coherence Combined With Different Mathematical Algorithms

The use of seismic data for fault detection has become an important part in the development of oil-gas exploration, because fault play an important role in controlling oil-gas migration and accumulation. The seismic data contain abundant information which can be used for fault detection. In the process of collecting seismic data in the field, there will be accompanied by a various kinds of uncertain factors, caused by man-made, instrument or poornatural environment, and thsese uncertainties result in a lot interference information irrelevant to the actual fault information in the final data collected. This will undoubtedly increase the difficulty of data interpreter to obtain effective fault information. In order to improve the ability to suppress noise interference in the fault detection process, many scholars have developed a lot of new fault detection methods.The method based on ant colony algorithm and the method based on Hough transform are more superiority in anti-noise performance and integrity of fault information than other fault detection methods.We have made efforts to improve the two methods in this dissertation, and proposed two new kinds of fault detection methods,which are improved fault detection method based on ant colony algorithm and improved fault detection method based on Hough transform.First of all, these two kinds of fault detection methods are the re strengthening treatment of the fault features included in the seismic attributes. Obviously, the quality of seismic attributes will directly affect the final results of the fault detection. Therefore, in order to obtain a higher quality seismic attribute data, this dissertation introduces the coherence that measure the intensity of local neighborhood anisotropy in texture image into seismic data processing. After verification, we can accord to the coherence of the square gradient vector direction of each point in the target point's neighborhood to identify the transverse discontinuity of the seismic wave event. Compared with the other attributes that characterize discontinuity of seismic wave events, the coherence is affected little by the stratigraphic dip, and can obtain higher quality data with fault characteristics without any dip calculation step, but it is only for the high SNR of seismic data at present.In order to improve the anti-noise, this disseratation puts weight on gradient coherence. By comparing the theoretical seismic model, it is found that the gradient weighted coherence improves the ability to suppress noise to a large extent, and it is significantly better than the coherence and variance attribute.Secondly, we combine seismic data gradient weighted coherence with colony algorithm and apply it to seismic data fault detection for the first time in this disseratation. Compared with the two fault detection results from ant colony algorithm based on the conventional variance and C3 coherent attribute,the proposed method in this disserataion has the following advantages: better anti-noise ability, better suppression of the pseudo fault information, better convergence ability of fault boundary, the longitudinal extension length of fault dentified by the proposed method is more complete on the basis of increasing the continuity of fault. But the method is still needed to improve the location accuracy of fault.Finally, in order to obtain more accurate fault location information from seismic data, this disseratation combines the seismic data gradient coherence with Hough transform to detect fault for the first time. Through theoretical model, the proposed method is superior to Hough transform previous proposed based on the coherence attribute in the location accuracy of fault. The original fault detection method based on Hough transform can only detect single fault, but can't detect multiple faults simultaneously. For this shortcoming, we use cluster analysis method to deal with the line segments of characterizing the fault feature detected by Hough transform for the first time, and then handle the other part of the fault line processing.Through the validation of theoretical model and real seismic data, it is proved that the improved method can detect multiple faults at the same time. In addition, we need to deal with seismic data gradient coherence by threshold segmentation in the early stage in this fault detection method, which is to provide the two valued data with the characteristic information of the fault, Because of artificially setting threshold, the original method is time-consuming and laborious, so we have made improvements in this dissertation. We use the Otsu method and iterative method to realize adaptive threshold segmentation of seismic data gradient coherence. After verification, the binary data from the two methods are able to provide effective fault information for Hough transform, which can greatly improve the computational efficiency.Compared the two new fault detection methods proposed in this dissertation,we can find that fault detection method based on the improved Hough transform has more significant advantages than fault detection based on ant colony algorithm in the location accuracy of fault, resolution and computational efficiency. But the clustering analysis, which is based on the measurement of the real distance between the points, can't separate the more complex fault combinations very well, such as the X or Y type fault and the faults very close to each other, but the improved fault detection method based on ant colony algorithm can detect the above types of faults better.