CN117252374A - Multi-layer production yield distribution method for tight gas-combined production well based on machine learning - Google Patents

Multi-layer production yield distribution method for tight gas-combined production well based on machine learning Download PDF

Info

Publication number
CN117252374A
CN117252374A CN202311232163.8A CN202311232163A CN117252374A CN 117252374 A CN117252374 A CN 117252374A CN 202311232163 A CN202311232163 A CN 202311232163A CN 117252374 A CN117252374 A CN 117252374A
Authority
CN
China
Prior art keywords
production
layer
model
gas
distribution
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202311232163.8A
Other languages
Chinese (zh)
Inventor
冯国庆
韩达
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Southwest Petroleum University
Original Assignee
Southwest Petroleum University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Southwest Petroleum University filed Critical Southwest Petroleum University
Priority to CN202311232163.8A priority Critical patent/CN117252374A/en
Publication of CN117252374A publication Critical patent/CN117252374A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/0499—Feedforward networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/067—Enterprise or organisation modelling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/02—Agriculture; Fishing; Forestry; Mining

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • General Physics & Mathematics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Software Systems (AREA)
  • Marketing (AREA)
  • Mathematical Physics (AREA)
  • Game Theory and Decision Science (AREA)
  • Quality & Reliability (AREA)
  • Development Economics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Operations Research (AREA)
  • General Engineering & Computer Science (AREA)
  • Computing Systems (AREA)
  • Evolutionary Computation (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • Artificial Intelligence (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Computational Linguistics (AREA)
  • Molecular Biology (AREA)
  • Educational Administration (AREA)
  • Primary Health Care (AREA)
  • Mining & Mineral Resources (AREA)
  • Marine Sciences & Fisheries (AREA)
  • Animal Husbandry (AREA)
  • Agronomy & Crop Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a multi-layer production yield distribution method of a tight gas well based on machine learning, and provides a method for establishing a multi-layer production yield distribution evaluation model of a tight gas well based on an artificial neural network. The invention selects factors affecting multi-layer production, such as: the porosity, permeability, formation pressure, perforation thickness and reservoir thickness are taken as input data of the model, the distribution coefficient calculated by a digital-analog mechanism model is taken as output, meanwhile, the data measured by the gas production profile are combined, so that sample data are more abundant, the defect of single sample data structure and the like is avoided, an input-output sample set is built, the sample set is input into a machine learning model, and the machine learning model is built by learning through an artificial neural network method. By using the artificial neural network to establish a yield distribution model, the multi-layer production yield distribution of the tight gas-gas production well can be better solved, the production condition of each production layer can be more conveniently, rapidly, accurately and effectively known, and better and more data support and technical support are provided for reservoir description and later production measure implementation.

Description

Multi-layer production yield distribution method for tight gas-combined production well based on machine learning
Technical Field
The invention relates to the technical field of oil and gas field development.
Background
In oil and gas resource development, because of the complex distribution of gas layer positions, multiple strata are often required to be simultaneously mined through a compact gas-tight gas production well so as to improve the mining effect and the resource utilization rate. However, there are differences in the production, physical parameters, and costs of different formations, so a reasonable production distribution model needs to be established to optimize the production configuration of tight gas production wells. In order to solve the problem of multi-layer yield distribution of the tight gas production well, a corresponding model needs to be constructed by using a mathematical modeling method. Common modeling methods include linear models, nonlinear models, dynamic models, optimization algorithm models, and the like. The production distribution problem of tight gas reservoirs often involves complex nonlinear relationships such as seepage rules, pressure distribution, etc., and conventional linear models are difficult to describe accurately. The traditional linear model can only describe a simple linear relation, and cannot accurately model nonlinear factor influence, so that the model prediction accuracy is not high. There are interactions and complex correlations between wells in multiple layers of gas production wells in the production distribution problem. The conventional linear model often assumes that layers are independent of each other, and the interaction cannot be fully considered in modeling, so that the model cannot fully and accurately describe the actual situation and is relatively simple to process data, but is also relatively sensitive to the existence of abnormal values. However, in multi-layer production distribution of gas-well production, the productivity of each layer of wells may change with time, geological conditions and other factors, and the model needs to have flexibility and adaptability to perform real-time adjustment and optimization. In summary, the conventional linear model has the disadvantages of being unable to capture the nonlinear relationship, neglecting interaction and complex association, being sensitive to abnormal values, and limiting flexibility and adaptability in the multi-layer production yield distribution problem of the gas-producing well. To more accurately build a yield distribution model, nonlinear models such as artificial neural networks can provide better solutions.
In order to solve the problems, the method combines the numerical simulation of the oil and gas reservoir with the machine learning method based on the basic seepage equation. The method for establishing the multi-layer production yield distribution model of the tight gas-combined production well based on the artificial neural network is provided, the problem that the multi-layer production yield distribution of the tight gas-combined production well has multiple variables and uncertainty is solved, and the artificial neural network can learn and adjust in real time according to actual data, so that the adaptability and accuracy of the model are improved.
In conclusion, the artificial neural network has obvious advantages in the aspect of establishing a multi-layer production yield distribution model of the tight gas-gas production well, and can improve the accuracy, adaptability and calculation efficiency of the model.
The invention of domestic related production distribution of the combined production well is as follows:
the invention discloses a multi-layer gas production well yield splitting method and system, and relates to the technical field of oil and gas field development. By adopting the method and the system provided by the invention, the splitting result of the gas well yield can be accurately and effectively determined.
The invention relates to the technical field of oil and gas field development, in particular to a multi-layer gas production well yield splitting method. According to the invention, yield splitting is carried out according to actual shaft pressure data and a gas well pipe flow equation, so that the theoretical basis is more sufficient, and the calculation accuracy is more accurate.
The embodiment of the invention discloses an intelligent yield splitting method and system for a multi-layer gas production well, which realize yield splitting of the multi-layer gas production well by using a trained machine learning model, can conveniently, quickly, accurately and effectively know the production conditions of each production layer, and provides more data support and technical reference for fine description of reservoirs, residual gas distribution research and later production measure implementation.
Compared with the invention of the method and the system for splitting the yield of the multi-layer gas recovery well and the method for splitting the yield of the multi-layer gas recovery well, the method for splitting the yield of the multi-layer gas recovery well adopts a nonlinear artificial neural network mode to predict the yield splitting, and improves the accuracy, the adaptability and the calculation efficiency of a model. Compared with the intelligent yield splitting method and system of the multi-layer gas production well, the method and the system not only utilize the data such as permeability, porosity, stratum coefficient and the like to perform splitting prediction, but also further research related distribution coefficients through the distribution coefficients and yield distribution results obtained by fusing perforation thickness, dynamic parameters, stratum pressure and the gas production profile of the production well, and perform accurate and efficient yield distribution.
The invention provides a method for establishing a multi-layer production yield distribution model of a tight gas-gas production well aiming at the defects of the existing method.
Disclosure of Invention
The invention provides a method for establishing a multi-layer production yield distribution model of a tight gas-gas production well aiming at the defects of the existing method.
The method of the invention is based on the following principle:
the artificial neural network takes neurons as a basic structure and consists of an input layer, a hidden layer and an output layer. The learning process of the artificial neural network algorithm comprises forward propagation and reverse propagation, wherein in the forward propagation process, input information is input from an input layer, processed by a hidden layer and then transmitted to an output layer, when the error of the output layer is larger than the minimum error, the learning process enters the reverse propagation process, an error signal primary path returns, the weights of neurons of each layer are optimized, then the learning process enters the forward propagation process again, the two processes are circularly reciprocated until the error of the output layer is smaller than the minimum error, and the final result is output.
According to the above description, the method for describing the multi-layer production yield distribution of the tight gas production well is obtained by the following specific steps:
(1) According to a plurality of groups of parameter values such as porosity, permeability, formation pressure, reservoir thickness, perforation thickness and the like of different reservoirs, corresponding gas yield ratio is obtained by numerical model software:
(1) data collection and arrangement: and collecting relevant geology, stratum parameters, bottom hole flow, production data, porosity, permeability, stratum pressure, reservoir thickness, perforation thickness, distribution coefficients obtained by production logging of a production well, yield distribution results and the like. Ensuring the accuracy and the integrity of data and screening.
(2) Reservoir description and model setting: and carrying out reservoir description on each layer section related to the tight gas production well, wherein the reservoir description comprises parameters such as reservoir thickness, porosity, permeability, distribution coefficient of a gas production profile, yield distribution result and the like. And determining the model setting of the tight gas production well according to the actual conditions and the requirements.
(3) Yield distribution model selection: and selecting a proper yield distribution model according to the characteristics of the tight gas production well.
(4) Model parameter estimation and optimization: parameters in the selected model are estimated and optimized. This may involve methods such as fitting of historical data, analog computation, or numerical optimization. And the fitting degree of the model and the actual data is optimized by adjusting the parameter value, so that the accuracy and the reliability of the model are improved.
(2) The porosity, the permeability, the pressure, the reservoir thickness, the perforation thickness and the gas production ratio of the co-production layer are taken as input variables, the data measured by combining the gas production profile are input into a machine learning model for learning, wherein 70% is a training set, 30% is a verification set, and the gas production ratio is predicted by introducing an artificial neural network;
(3) And comparing and verifying the predicted daily yield ratio of the natural gas with the daily yield ratio of the original natural gas, and carrying out error analysis.
Compared with the prior art, the invention has the advantages that: avoiding complex calculation, reducing time consumption and reducing cost.
Drawings
FIG. 1 is a schematic diagram of a multi-layer digital-analog model of a tight gas production well according to the present invention
FIG. 2 is a network structure diagram of a multi-layer production distribution model of a tight gas production well according to the invention;
FIG. 3 is a graph of the correlation of the porosity, permeability, pressure, reservoir thickness, perforation thickness, yield distribution results and daily gas production distribution ratio of the reservoirs obtained according to the present invention;
Detailed Description
The present invention will be described in further detail with reference to the following examples, in order to make the objects, studies and advantages of the present invention more apparent.
A multi-layer production yield distribution method of a dense gas-combined production well based on machine learning comprises the following steps:
and according to a plurality of groups of porosity, permeability, pressure, reservoir thickness and perforation thickness parameter values of different reservoirs, taking the values as input data, taking a yield comparison result of numerical simulation calculation as output data, and constructing a sample database by combining distribution coefficients obtained by a gas production profile.
The input data porosity, permeability, formation pressure, perforation thickness, reservoir thickness, and output data yield ratios in the sample set are shown in table 1.
Table 1 randomly generated porosity ratio, permeability ratio, pressure ratio, perforation thickness ratio, reservoir thickness ratio, yield ratio table
FIG. 2 is a diagram of a neural network structure of a multi-layer production distribution model of a tight gas production well.
The porosity, the permeability, the pressure, the reservoir thickness, the perforation thickness and the gas production ratio of the gas production layer are taken as input variables, the gas production section measured data are combined, the gas production section measured data are input into a machine learning model for learning, 70% of the gas production section is a training set, 30% of the gas production section is a verification set, and an artificial neural network is introduced to predict distribution ratios of different reservoir yields of the tight gas production well;
three sets of models are built under different hidden layers at different reservoir parameters, respectively, as shown in Table 2
Table 2 model network architecture
And comparing and verifying the predicted gas yield distribution ratios of different reservoirs of the tight gas production well with the gas yield distribution ratios of different reservoirs of the tight gas production well, and carrying out error analysis.
Fig. 3 is a graph comparing the distribution ratio of the gas production of different reservoirs of the tight gas production well tested by the obtained model with the distribution ratio of the gas production of different reservoirs of the original tight gas production well, wherein the fitting rate of the tight gas production well test result and the original data in the graph reaches 93%, and most of the obtained curve is high in coincidence with the original curve.
Those of ordinary skill in the art will appreciate that the embodiments described herein are intended to aid the reader in understanding the practice of the invention and that the scope of the invention is not limited to such specific statements and embodiments. Those of ordinary skill in the art can make various other specific modifications and combinations from the teachings of the present disclosure without departing from the spirit thereof, and such modifications and combinations remain within the scope of the present disclosure.

Claims (1)

1. Multi-layer production yield distribution method for tight gas-combined production well based on machine learning
The method for establishing the multi-layer production distribution model of the tight gas-production well comprises the following specific steps of:
(1) According to a plurality of groups of parameter values such as porosity, permeability, pressure, thickness, perforation thickness and the like of different reservoirs, corresponding natural gas yield of each reservoir is obtained through simulation by numerical model software;
(2) The porosity, the permeability, the pressure, the reservoir thickness, the perforation thickness and the gas production ratio of the co-production layer are taken as input variables, the data measured by combining the gas production profile are input into a machine learning model for learning, wherein 70% is a training set, 30% is a verification set, and the gas production ratio is predicted by introducing an artificial neural network;
(3) And after the model reaches the convergence condition, analyzing the result of the verification set data, and when the coincidence rate of the predicted capacity distribution coefficient and the actual distribution coefficient reaches more than 90%, calculating and finishing outputting the network structure parameter, otherwise, adjusting the neural network structure parameter until the coincidence rate meets the requirement.
CN202311232163.8A 2023-09-22 2023-09-22 Multi-layer production yield distribution method for tight gas-combined production well based on machine learning Pending CN117252374A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311232163.8A CN117252374A (en) 2023-09-22 2023-09-22 Multi-layer production yield distribution method for tight gas-combined production well based on machine learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311232163.8A CN117252374A (en) 2023-09-22 2023-09-22 Multi-layer production yield distribution method for tight gas-combined production well based on machine learning

Publications (1)

Publication Number Publication Date
CN117252374A true CN117252374A (en) 2023-12-19

Family

ID=89136377

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311232163.8A Pending CN117252374A (en) 2023-09-22 2023-09-22 Multi-layer production yield distribution method for tight gas-combined production well based on machine learning

Country Status (1)

Country Link
CN (1) CN117252374A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117522173A (en) * 2024-01-04 2024-02-06 山东科技大学 Natural gas hydrate depressurization exploitation productivity prediction method based on deep neural network

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117522173A (en) * 2024-01-04 2024-02-06 山东科技大学 Natural gas hydrate depressurization exploitation productivity prediction method based on deep neural network
CN117522173B (en) * 2024-01-04 2024-04-26 山东科技大学 Natural gas hydrate depressurization exploitation productivity prediction method based on deep neural network

Similar Documents

Publication Publication Date Title
CN109447532B (en) Oil reservoir inter-well connectivity determination method based on data driving
CA2640725C (en) Methods, systems, and computer-readable media for fast updating of oil and gas field production models with physical and proxy simulators
CN110400006B (en) Oil well yield prediction method based on deep learning algorithm
US8504341B2 (en) Methods, systems, and computer readable media for fast updating of oil and gas field production models with physical and proxy simulators
CN109543828B (en) Water absorption profile prediction method based on small sample condition
CN109214026A (en) A kind of shale gas horizontal well initial productivity prediction technique
CN117252374A (en) Multi-layer production yield distribution method for tight gas-combined production well based on machine learning
CN112308269A (en) Yield prediction method and device for low-permeability oil and gas reservoir
CN108595803A (en) Shale gas well liquid loading pressure prediction method based on recurrent neural network
Lü et al. Productivity matching and quantitative prediction of coalbed methane wells based on BP neural network
US20230358123A1 (en) Reinforcement learning-based decision optimization method of oilfield production system
CN112541571A (en) Injection-production connectivity determination method based on machine learning of double parallel neural networks
CN115563856A (en) XGboost and CV cross function-based multilayer commingled production fluid profile splitting method
CN111339719A (en) Fracture-cavity type oil reservoir water injection numerical simulation automatic history fitting method
CN110988997A (en) Hydrocarbon source rock three-dimensional space distribution quantitative prediction technology based on machine learning
CN116415394A (en) Two-loop equipment model simulation method integrating machine learning and mechanism model
CN115618750B (en) Underground oil-water seepage agent model based on coupling neural network
CN112302594A (en) Method, device and equipment for determining connection structure of heterogeneous water-drive reservoir
CN117390381A (en) Underground diaphragm wall joint seam leakage prediction method and device based on deep learning
CN108765562B (en) Oil-gas productivity evaluation method based on three-dimensional geological model
CN112052977A (en) Oil reservoir reserve prediction method based on deep space-time attention network
CN110486009A (en) A kind of automatic reverse method of parameter and system on infinity stratum
WO2019216892A1 (en) Optimized methodology for automatic history matching of a petroleum reservoir model with ensemble kalman filter
CN113705917A (en) Method for predicting initial productivity of fractured horizontal well of tight reservoir
CN113762559A (en) Intelligent yield splitting method and system for multilayer commingled production gas well

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination