Oil well production prediction method based on deep learning algorithm

Abstract

The invention provides a method for predicting oil well production based on a deep learning algorithm. The method for predicting oil well production based on a deep learning algorithm includes: step 1, acquiring data and performing quality inspection; step 2, performing data processing and division; step 3, establishing Learning model; step 4, use the model built in step 3 to carry out training and verification; step 5, predict oil well production. The oil well production prediction method based on deep learning algorithm establishes the relationship between reservoir physical properties, working system, development stage and other factors and oil production and liquid production through training, and takes advantage of the data-driven algorithm to establish multi-factor oil well production prediction Model.

Description

technical field [0001] The invention relates to the technical field of oilfield development, in particular to an oil well production prediction method based on a deep learning algorithm. Background technique [0002] Oil well and oil field production prediction is one of the most important tasks in oil field production management, and the prediction results directly determine subsequent oil field development decisions. However, limited by geological conditions, technological level, development history, data quality and other conditions, it is very difficult to predict the change of oil well production with time. At present, the commonly used methods in mines include: reservoir engineering method and numerical simulation method. The reservoir engineering method represented by Arps decline curve is a direct fitting to the phenomenon of oil well production decline. It is simple to operate and is not limited by the type of oil and gas reservoir. The defects of this method are v...

AI Technical Summary

Problems solved by technology

The reservoir engineering method represented by the Arps decline curve is a direct fitting of the production decline phenomenon of oil wells. It is easy to operate and is not limited by the type of reservoir. The defect of this method is very obvious. The prediction must assume that the historical production conditions will remain unchanged in the future change, cannot be used to analyze data in unsteady flow state

Although the subsequent improved methods make up for the differences in reservoir types and flow stages to varying degrees, they are always limited to the basic process of typical mathematical models-field data fitting, and the assumptions of typical theoretical models are the limitations of this method.

Reservoir numerical simulation is based on the understanding of the real flow process of underground porous media. It is a typical physical-driven data analysis method, which can consider more factors in more detail, and the prediction results are more objective than reservoir engineering. However, the modeling digital simulation process is very time-consuming , especially when the geological conditions are complex or the seepage mechanism is not clear, the prediction results still need to be improved

Images