Volcano channel identification method based on unsupervised neural network

Abstract

The invention discloses a volcano channel identification method based on an unsupervised neural network, and aims to solve the problems that the quality of seismic data in a volcano channel is poor, the transverse change is very fast, and the characteristics are generally represented as weak-amplitude cluttered reflection characteristics; therefore, research on the spatial distribution characteristics of the volcano channel is always a difficulty in geophysics research. Along with continuous advancing of oil exploration and development work, especially for middle-deep oil fields with igneous rock development, fine recognition of volcano channels is more important. Description of the volcano channel directly affects deployment of a subsequent development well site. Therefore, a volcano channel identification technology based on an unsupervised neural network is developed, the method and an opposite propagation neural network are deeply fused to form an unsupervised intelligent learningmodel, the model is used for training to obtain new attributes depicting a volcano channel, and full-three-dimensional automatic interpretation of the volcano channel is realized.

Description

technical field [0001] The invention belongs to the technical field of petroleum exploration seismic data processing and interpretation, and in particular relates to a volcano channel identification method based on an unsupervised neural network. Background technique [0002] Due to the influence of volcanic tectonic movements, the quality of seismic data inside volcanic channels is poor, and the lateral changes are very fast; they usually appear as chaotic reflections with weak amplitudes; therefore, the study of the spatial distribution characteristics of volcanic channels has always been a difficult point in geophysical research. With the continuous advancement of petroleum exploration and development, especially in medium-deep oilfields with igneous rocks, the fine identification of volcanic channels directly affects the deployment of follow-up development wells. [0003] There are many methods to characterize the distribution characteristics of volcanic channels, mainly...

AI Technical Summary

Problems solved by technology

Conventional seismic attribute detection mainly uses edge detection techniques such as variance and curvature to describe the distribution range of volcanic channels. The calculation results of such methods are highly dependent on the quality of seismic data. Because the quality of seismic data around volcanic channels is poor, such attributes are easy to Brings many false interpretations

Seismic facies analysis technology is developed on the basis of seismic stratigraphy. There are mainly two methods: waveform classification and seismic structure attributes. It mainly analyzes the variation characteristics of sedimentary facies belts by distinguishing the difference in reflection characteristics between seismic traces. This type of method can effectively describe the planar distribution characteristics of igneous rocks, but it is difficult to describe the spatial distribution characteristics of volcanic channels

Seismic inversion is an important means to predict the vertical and lateral distribution characteristics of reservoirs and abnormal geological bodies, but the conventional inversion method that relies on geological models has great limitations, and accurate geological models are needed to have better results. Inversion results, but in fact it is difficult to provide, therefore, the inversion results do not achieve the desired effect

Images