Three-dimensional reconstruction method of shale digital core based on deep learning and support vector machine

Abstract

The present invention relates to a method for three-dimensional reconstruction of shale digital cores based on deep learning and Support Vector Machine (SVM, Support Vector Machine), comprising the following steps: S1, using a three-dimensional data template to scan volume data of real shale cores to obtain page The three-dimensional pattern library of the rock core; S2, using the deep belief network (DBN, Deep Belief Networks) to extract the features of the three-dimensional pattern library; S3, using the SVM to classify the extracted features to form a class set {Category i , i=1,2,3...}; S4, using the multi-point geostatistical method to reconstruct the digital core. Compared with the prior art, the present invention uses deep learning and support vector machine to reconstruct shale digital rock core, deep learning has a strong ability to extract the essential features of training images, and support vector machine can analyze the structural features of shale Classification, and then using the multi-point geostatistical method can effectively reconstruct the shale digital core.

Description

technical field [0001] The present invention relates to a method for three-dimensional reconstruction of digital rock cores, in particular to a method for three-dimensional reconstruction of shale digital rock cores based on deep learning and support vector machines. Background technique [0002] With the continuous depletion of conventional gas reservoirs, the development and production of unconventional oil and gas resources such as shale gas reservoirs has received increasing attention. The flow of shale gas in the reservoir is a complex multi-scale flow process, and the gas flow mechanism is obviously different from that of conventional gas reservoirs. As the storage and flow carrier of shale gas, the rock pore structure of shale reservoirs is complex, with pore sizes ranging from nanometers to micrometers, accompanied by naturally developed micro-fractures, and the pores and gas in fractures of different sizes Occurrence states are different from motion characteristics...

AI Technical Summary

Problems solved by technology

[0005] As mentioned above, due to the coexistence of pores of various scales in shale, including micro-scale and nano-scale pores and micro-fractures, the traditional shale digital core reconstruction method cannot describe the complex pore-micro-fracture structure in shale well. , making it difficult for the reconstructed shale core to meet the requirements of studying the microscopic seepage mechanism, which brings great challenges to reconstructing the digital core by numerical methods

Images