Method for shale gas production prediction, device, and storage medium

Patent metadata

Field	Value
Publication number	US-12203352-B2
Application number	US-202418426597-A
Country	US
Kind code	B2
Filing date	Jan 30, 2024
Priority date	Feb 27, 2023
Publication date	Jan 21, 2025
Grant date	Jan 21, 2025

Abstract

Provided is a method for shale gas production prediction, including: acquiring real shale gas production data, and determining a target production decline curve model according to the real shale gas production data; setting a time step for production prediction, and obtaining a linear production decline curve by performing production prediction by using the target production decline curve model based on the time step; determining target shale gas production residuals according to the linear production decline curve and the real shale gas production data, inputting the target shale gas production residuals to a long short-term memory, and obtaining a residual prediction result by performing production prediction according to the long short-term memory and the time step; and determining a target production prediction result of shale gas well to be subjected to production prediction based on the linear production decline curve and the residual prediction result.

First claim

What is claimed is:

1. A method for shale gas production prediction, executed by a processor, comprising:
acquiring real shale gas production data of a shale gas well to be subjected to production prediction, and determining a target production decline curve model for production prediction of the shale gas well to be subjected to production prediction according to the real shale gas production data, wherein the shale gas well to be subjected to production prediction is a shale gas well which needs to be subjected to predicting production, and the real shale gas production data is shale gas production corresponding to each of past several years obtained by exploiting the shale gas well to be subjected to production prediction;
setting a time step for production prediction of the shale gas well to be subjected to production prediction, and obtaining a linear production decline curve corresponding to the shale gas well to be subjected to production prediction by performing production prediction on the shale gas well to be subjected to production prediction by using the target production decline curve model based on the time step, wherein performing production prediction on the shale gas well to be subjected to production prediction by using the target production decline curve model indicates that obtaining shale gas production corresponding to each of future several years by using the target production decline curve model;
determining target shale gas production residuals according to the linear production decline curve and the real shale gas production data, inputting the target shale gas production residuals into a long short-term memory, and obtaining a residual prediction result by performing production prediction according to the long short-term memory and the time step;
determining a target production prediction result of the shale gas well to be subjected to production prediction based on the linear production decline curve and the residual prediction result, and verifying the target production prediction result by using a preset test set corresponding to the shale gas well to be subjected to production prediction; and
using the target production prediction result to adjust at least one of subsequent development or deployment of the shale gas well or a new shale gas well;
wherein said determining the target production decline curve model for production prediction of the shale gas well to be subjected to production prediction according to the real shale gas production data comprises:
determining an initial production decline curve model for production prediction of the shale gas well to be subjected to production prediction according to the real shale gas production data, wherein a model parameter of the initial production decline curve model is a preset initial parameter; and
fitting the real shale gas production data by using the initial production decline curve model based on the preset initial parameter of the initial production decline curve model, so as to determine a target model parameter according to a fitting result, and to determine the target production decline curve model corresponding to the target model parameter.

2. The method for shale gas production prediction according to claim 1 , wherein prior to determining the target shale gas production residuals according to the linear production decline curve and the real shale gas production data, the method further comprises:
determining a real shale gas production curve based on the initial production decline curve model and the real shale gas production data of the shale gas well to be subjected to production prediction, so as to determine the target shale gas production residuals according to the linear production decline curve and the real shale gas production curve.

3. The method for shale gas production prediction according to claim 1 , wherein said inputting the target shale gas production residuals into the long short-term memory comprises:
standardizing the target shale gas production residuals and inputting the standardized target shale gas production residuals into the long short-term memory; and
correspondingly, said obtaining the residual prediction result by performing production prediction according to the long short-term memory and the time step comprises:
obtaining an initial prediction result by performing production prediction according to the long short-term memory and the time step, and obtaining the residual prediction result by anti-standardizing the initial prediction result.

4. The method for shale gas production prediction according to claim 1 , wherein upon inputting the target shale gas production residuals into the long short-term memory, the method further comprises:
setting initial values of the number of hidden units and the number of training rounds of a corresponding model of the long short-term memory, so as to obtain the residual prediction result by performing production prediction according to the corresponding model of the long-short memory and the time step.

5. The method for shale gas production prediction according to claim 4 , wherein said obtaining the residual prediction result by performing production prediction according to the long short-term memory and the time step comprises:
fitting the target shale gas production residuals according to the long short-term memory, and adjusting a memory parameter corresponding to the long short-term memory according to a residual fitting result; and
determining the adjusted long short-term memory based on the adjusted memory parameter, and obtaining the residual prediction result by performing production prediction according to the adjusted long short-term memory and the time step.

6. The method for shale gas production prediction according to claim 1 , wherein said determining the target production prediction result of the shale gas well to be subjected to production prediction based on the linear production decline curve and the residual prediction result comprises:
obtaining the target production prediction result of the shale gas well to be subjected to production prediction by summing a linear prediction result corresponding to the linear production decline curve and a nonlinear prediction result corresponding to the residual prediction result.

7. An electronic device, wherein the electronic device comprises a processor and a memory configured to store one or more computer programs, wherein the one or more computer programs, when loaded and executed by the processor, cause the processor to perform the method for shale gas production prediction as defined in claim 1 .

8. A non-transitory computer-readable storage medium storing one or more computer programs therein, wherein the one or more computer programs, when loaded by a processor, cause the processor to perform the method for shale gas production prediction as defined in claim 1 .

Assignees

• Univ China Petroleum East China

Inventors

• Sun Hai
• Luo Fei
• Fan Dongyan
• Zhang Lei
• Yao Jun
• Fu Shuaishi
• Zhang Kai
• Yang Yongfei

Classifications

• G01V20/00

  Geomodelling in general · CPC title

• E21B49/087

  Well testing, e.g. testing for reservoir productivity or formation parameters · CPC title

• G06Q50/02

  Agriculture; Fishing; Forestry; Mining · CPC title

• E21B2200/20

  Computer models or simulations, e.g. for reservoirs under production, drill bits · CPC title

• E21B2200/22

  Fuzzy logic, artificial intelligence, neural networks or the like · CPC title