Method and device for fluid prediction based on deep learning using seismic data

Abstract

The invention discloses a method for fluid prediction based on deep learning using seismic data, which can finally output the probability distribution of oil and gas characteristics, improve the accuracy of predicting the spatial distribution of oil and gas, and provide important support for well location deployment. The method includes: inputting and preprocessing all seismic data within a specified area; using the first set of deep learning networks to perform nonlinear optimization and fitting on the linear features of seismic data; using the second set of deep learning networks to Classify the linear features of the seismic data to establish the first fluid feature model; use the residual network in the second deep learning network to iterate and correct the established first fluid feature model to obtain the second fluid feature model; The second fluid characteristic model and the activation function perform matrix set calculation on the seismic data in the block to be predicted, and obtain the probability distribution data of the fluid feature in the block to be predicted.

Description

technical field [0001] The invention relates to the technical field of seismic data processing and fluid prediction, in particular to a method and device for fluid prediction based on deep learning using seismic data. Background technique [0002] Oil and gas exploration and development targets are mostly affected by various geological factors such as structure and lithology. Using drilling, logging, geological and other data can accurately obtain oil and gas information in a certain interval near the wellbore, but it is difficult to describe the reservoir and oil and gas conditions between wells and other areas. Seismic data contain very rich information on reservoir physical properties and have good continuity in the lateral direction. Therefore, seismic exploration technology is an effective means for lateral prediction of reservoirs and oil and gas reservoirs. In seismic oil and gas exploration, it is always hoped that the location of oil and gas can be found directly b...

AI Technical Summary

Problems solved by technology

[0004] However, in seismic exploration, since the seismic wave produced by explosives is not a standard seismic signal, and due to the heterogeneity of the formation, the signal received by the geophone is affected by many factors, and it is necessary to remove all interference without losing the effective signal. It is very difficult. Due to many processing links in the processing process, it is difficult to obtain the seismic wave velocity accurately, resulting in ambiguity in the processing results.

This causes two problems in the process of fluid prediction using seismic attributes. One is that the signal source is a non-ideal regular signal, which leads to low prediction accuracy. greater limitations

Images