基于蚂蚁追踪的三维地震图像目标识别研究
发布时间：2018-11-01 20:04
【摘要】：在地质勘探中,为了获得油气等资源的分布情况,需要对地震数据进行解释,而断层解释是地震数据解释的重要内容。断层是由于地下岩石层因为受到一定程度的压力而产生破裂,并且沿着破裂面的方向产生相对错位移动的地质现象。断层导致油田的产生和分布,既是断块的油田的分界处,还是油气移动的通路。在勘探石油、天然气等资源的末期或者在开发这些资源时,解释和弄清断层在地下的大小和分布状况,不管是对评估产量、产量建设,还是对于油藏的挖潜和管理都具有重大意义。本文通过对地震数据进行研究,对地下断层大小和分布情况进行估计。常规的断层解释方法是用人工方法解释,这种方法不仅工作量大,周期长,并且非常依赖解释人员的专业知识和解释经验。因此,迫切需要一种断层自动解释方法。目前,国外商业软件中只有斯伦贝谢(Schlumberger)公司开发的Petrel软件实现了三维地震图像上的断层自动识别解释,但是国内在这一领域尚属空白,Petrel中断层自动识别解释的具体实现过程由于知识产权国内也无法获得。本文以Petrel的处理方法和蚂蚁追踪算法为基础,主要进行了如下工作:1、开发了基于蚂蚁追踪算法的三维地震图像断层自动识别方法。本文借鉴了Petrel的思想方法,首先用断层增强属性体进行预处理,突出断层并抑制非断层,然后用蚂蚁追踪算法在断层增强属性体上追踪断层,实现了断层的自动识别。2、针对Petrel追踪得到的结果中含有很多噪声干扰的问题,本文从蚂蚁追踪算法模型角度入手,引入人工种子点和梯度一致性,结合基于带精英策略的蚂蚁追踪算法和基于排序的蚂蚁追踪算法,对不同置信度的人工蚂蚁给予不同的信息素更新策略,使人工蚂蚁尽量在大断层上追踪,减少在非断层结构上的追踪。实际测试表明,该方法能有效抑制噪声干扰。针对追踪得到的断层面上留有毛刺、空洞、分叉、缺口等问题,本文还运用数学形态学方法平滑和填补断层面,并用三维地震图像骨骼化算法细化断层。实际测试表明,该方法能使断层面更加清晰完整。3、为了评估本文方法的效果,我们用实际地震工区数据进行测试。实际测试表明,本文的方法能有效识别地震图像上的断层。相比Petrel,本文方法识别出的断层连续性更好,大断层更加清晰明显,并且有效减少了噪声的干扰。
[Abstract]:In geological exploration, in order to obtain the distribution of oil and gas resources, seismic data need to be interpreted, and fault interpretation is an important part of seismic data interpretation. Fault is a geological phenomenon that the underground lithosphere is fractured because of a certain degree of pressure and is relatively dislocated along the direction of the fracture surface. The formation and distribution of oil field caused by fault is not only the boundary of fault block oil field, but also the path of oil and gas migration. At the end of exploration for oil, natural gas and other resources, or in the development of these resources, explain and understand the size and distribution of faults underground, whether it is for assessing production, production construction, Or for the reservoir to tap potential and management are of great significance. In this paper, the size and distribution of underground faults are estimated by studying seismic data. The conventional fault interpretation method is interpreted by manual method. This method not only has a large workload and long period, but also relies heavily on the professional knowledge and interpretation experience of the interpreters. Therefore, there is an urgent need for an automatic fault interpretation method. At present, only the Petrel software developed by Schlumberger (Schlumberger) Company has realized automatic fault recognition and interpretation on 3D seismic images in foreign commercial software, but this field is still blank in China. The realization process of automatic fault identification and interpretation in Petrel is not available because of intellectual property rights. Based on the processing method of Petrel and ant tracking algorithm, the main work of this paper is as follows: 1. The automatic recognition method of 3D seismic image fault based on ant tracking algorithm is developed. This paper draws lessons from Petrel's thought and method, first uses fault enhancement attribute body to pre-process, protrudes fault and suppresses non-fault, then uses ant tracing algorithm to trace fault on fault enhancement attribute body, realizes automatic fault recognition. In order to solve the problem of noise disturbance in the result of Petrel tracking, this paper introduces artificial seed points and gradient consistency from the point of view of ant tracking algorithm model. Combined with ant tracking algorithm with elite strategy and ant tracking algorithm based on sorting, different pheromone updating strategies are given to artificial ants with different confidence levels, so that artificial ants can be traced on large faults as far as possible. Reduce tracking on non-fault structures. The experimental results show that the proposed method can effectively suppress noise interference. Aiming at the problems of burr, cavity, bifurcation and gap on the fault plane obtained by tracing, this paper also uses mathematical morphology method to smooth and fill the fault plane, and thinning the fault with three-dimensional seismic image skeleton algorithm. The actual test shows that the method can make fault plane more clear and complete. 3. In order to evaluate the effect of this method, we use the actual seismic data to test. Practical tests show that the proposed method can effectively identify faults in seismic images. Compared with the Petrel, method, the fault continuity is better, the large fault is clearer and more obvious, and the noise interference is reduced effectively.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：P631.44;TP391.41
本文编号：2304955
[Abstract]:In geological exploration, in order to obtain the distribution of oil and gas resources, seismic data need to be interpreted, and fault interpretation is an important part of seismic data interpretation. Fault is a geological phenomenon that the underground lithosphere is fractured because of a certain degree of pressure and is relatively dislocated along the direction of the fracture surface. The formation and distribution of oil field caused by fault is not only the boundary of fault block oil field, but also the path of oil and gas migration. At the end of exploration for oil, natural gas and other resources, or in the development of these resources, explain and understand the size and distribution of faults underground, whether it is for assessing production, production construction, Or for the reservoir to tap potential and management are of great significance. In this paper, the size and distribution of underground faults are estimated by studying seismic data. The conventional fault interpretation method is interpreted by manual method. This method not only has a large workload and long period, but also relies heavily on the professional knowledge and interpretation experience of the interpreters. Therefore, there is an urgent need for an automatic fault interpretation method. At present, only the Petrel software developed by Schlumberger (Schlumberger) Company has realized automatic fault recognition and interpretation on 3D seismic images in foreign commercial software, but this field is still blank in China. The realization process of automatic fault identification and interpretation in Petrel is not available because of intellectual property rights. Based on the processing method of Petrel and ant tracking algorithm, the main work of this paper is as follows: 1. The automatic recognition method of 3D seismic image fault based on ant tracking algorithm is developed. This paper draws lessons from Petrel's thought and method, first uses fault enhancement attribute body to pre-process, protrudes fault and suppresses non-fault, then uses ant tracing algorithm to trace fault on fault enhancement attribute body, realizes automatic fault recognition. In order to solve the problem of noise disturbance in the result of Petrel tracking, this paper introduces artificial seed points and gradient consistency from the point of view of ant tracking algorithm model. Combined with ant tracking algorithm with elite strategy and ant tracking algorithm based on sorting, different pheromone updating strategies are given to artificial ants with different confidence levels, so that artificial ants can be traced on large faults as far as possible. Reduce tracking on non-fault structures. The experimental results show that the proposed method can effectively suppress noise interference. Aiming at the problems of burr, cavity, bifurcation and gap on the fault plane obtained by tracing, this paper also uses mathematical morphology method to smooth and fill the fault plane, and thinning the fault with three-dimensional seismic image skeleton algorithm. The actual test shows that the method can make fault plane more clear and complete. 3. In order to evaluate the effect of this method, we use the actual seismic data to test. Practical tests show that the proposed method can effectively identify faults in seismic images. Compared with the Petrel, method, the fault continuity is better, the large fault is clearer and more obvious, and the noise interference is reduced effectively.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：P631.44;TP391.41
【参考文献】
相关硕士学位论文 前1条
1 周彬;基于数学形态学的图像处理算法研究[D];华北电力大学（北京）;2008年
，本文编号：2304955
本文链接：https://www.wllwen.com/falvlunwen/zhishichanquanfa/2304955.html
