CN115204064B - Gas injection huff-puff parameter optimization method and system in shale oil exploitation - Google Patents
Gas injection huff-puff parameter optimization method and system in shale oil exploitation Download PDFInfo
- Publication number
- CN115204064B CN115204064B CN202211133564.3A CN202211133564A CN115204064B CN 115204064 B CN115204064 B CN 115204064B CN 202211133564 A CN202211133564 A CN 202211133564A CN 115204064 B CN115204064 B CN 115204064B
- Authority
- CN
- China
- Prior art keywords
- process parameter
- sample set
- objective function
- offspring
- child
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 417
- 238000005457 optimization Methods 0.000 title claims abstract description 98
- 239000003079 shale oil Substances 0.000 title claims abstract description 65
- 238000002347 injection Methods 0.000 title claims abstract description 54
- 239000007924 injection Substances 0.000 title claims abstract description 54
- 238000012549 training Methods 0.000 claims abstract description 27
- 238000004422 calculation algorithm Methods 0.000 claims description 53
- 238000004088 simulation Methods 0.000 claims description 43
- 238000004519 manufacturing process Methods 0.000 claims description 15
- 238000011084 recovery Methods 0.000 claims description 10
- 238000012935 Averaging Methods 0.000 claims description 5
- 238000003066 decision tree Methods 0.000 claims description 5
- 238000012706 support-vector machine Methods 0.000 claims description 5
- 238000013528 artificial neural network Methods 0.000 claims description 4
- 238000005516 engineering process Methods 0.000 claims description 4
- 239000000126 substance Substances 0.000 claims description 4
- 230000000694 effects Effects 0.000 abstract description 7
- 230000006870 function Effects 0.000 description 102
- 238000011161 development Methods 0.000 description 11
- 239000003921 oil Substances 0.000 description 8
- 239000000243 solution Substances 0.000 description 7
- 238000005070 sampling Methods 0.000 description 6
- 238000003860 storage Methods 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 4
- 239000010779 crude oil Substances 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 230000009191 jumping Effects 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000015654 memory Effects 0.000 description 2
- 239000002245 particle Substances 0.000 description 2
- 238000010276 construction Methods 0.000 description 1
- 238000005553 drilling Methods 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 238000013386 optimize process Methods 0.000 description 1
- 239000011148 porous material Substances 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000001902 propagating effect Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 238000002791 soaking Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/06—Multi-objective optimisation, e.g. Pareto optimisation using simulated annealing [SA], ant colony algorithms or genetic algorithms [GA]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- Biophysics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Business, Economics & Management (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Physiology (AREA)
- Molecular Biology (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Genetics & Genomics (AREA)
- Agronomy & Crop Science (AREA)
- Animal Husbandry (AREA)
- Marine Sciences & Fisheries (AREA)
- Mining & Mineral Resources (AREA)
- Computing Systems (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a gas injection huff and puff parameter optimization method and a gas injection huff and puff parameter optimization system in shale oil exploitation, which belong to the technical field of optimization of shale oil exploitation means and comprise the following steps: constructing an optimized sample set; respectively training a plurality of pre-established regression prediction models by utilizing the optimized sample set to obtain a plurality of trained regression prediction models; selecting an initial population in the optimized sample set; iteratively updating the initial population to obtain a target population; constructing a progeny estimation sample set in the process; judging whether an optimization ending condition is reached; if so, taking the offspring process parameter group with the maximum objective function value in the target population as the optimal process parameter group; otherwise, updating the optimized sample set according to the child estimated sample set, and retraining a plurality of regression prediction models. According to the method, a plurality of groups of process parameters are generated in the value range of the process parameters, corresponding objective function values are obtained, an optimized sample set is constructed, and the problem that the training effect on the model is poor due to the fact that the data quantity can be obtained is effectively avoided.
Description
Technical Field
The invention relates to the technical field of optimization of shale oil exploitation means, in particular to a method and a system for optimizing gas injection huff and puff parameters in shale oil exploitation.
Background
The realization of the efficient development of unconventional oil and gas resources such as shale oil reservoirs and the like is an important measure for ensuring the balance of supply and demand of the crude oil market and guaranteeing the national energy safety. But the method is limited by the complex pore structure and seepage characteristics of the shale oil reservoir, the primary recovery rate of crude oil is extremely low, and effective measures for improving the recovery rate are required to be developed. CO 2 2 Huff and puff development is an effective means for improving the recovery ratio of shale oil reservoir, and determining huff and puff technological parameters such as reasonable measure implementation time, gas injection rate and the like is to ensure CO 2 The throughput measures are economicalThe precondition of effectiveness is also the precondition of ensuring the improvement of the shale oil reservoir recovery ratio.
However, most of the existing shale oil reservoir gas injection throughput process parameter optimization methods are economic benefits which can be achieved by analyzing a set of process parameters through a numerical simulation method, but the method is difficult to obtain a global optimal result. With the popularization of the regression prediction algorithm, the relationship between the process parameters and the economic benefits predicted by the regression prediction model also appears in the prior art, but in the method, a large amount of training data needs to be acquired for training the regression prediction model, otherwise, the accuracy of the regression prediction model cannot be guaranteed, so that the optimization effect on the gas injection huff-puff process parameters cannot be guaranteed.
Disclosure of Invention
The invention aims to provide a gas injection throughout parameter optimization method and system in shale oil exploitation, which improve the accuracy of prediction of objective function values of process parameter sets, thereby ensuring the reliability of selection of the optimal process parameter sets.
In order to achieve the purpose, the invention provides the following scheme:
a gas injection huff and puff parameter optimization method in shale oil exploitation is used for optimizing each process parameter of a gas injection huff and puff process, and comprises the following steps:
constructing an optimized sample set; the optimization sample set comprises a plurality of process parameter sets and objective function values corresponding to the process parameter sets;
selecting a plurality of process parameter sets with the maximum objective function values and corresponding objective function values in the optimized sample set as initial populations of the evolutionary algorithm;
taking a process parameter group as input, taking an objective function value corresponding to the process parameter group as target output, and respectively training a plurality of regression prediction models established in advance to obtain a plurality of trained regression prediction models; the regression prediction models are different in structure;
iteratively updating the initial population by using an evolutionary algorithm and a plurality of trained regression prediction models to obtain a target population;
constructing a child estimation sample set according to child process parameter sets of all intermediate populations generated in the process of iteratively updating the initial population and estimation objective function values corresponding to the child process parameter sets;
judging whether an optimization ending condition is reached; if so, taking the offspring process parameter set with the maximum objective function value in the target population as the optimal process parameter set of the target shale oil well;
otherwise, re-determining an initial population according to the offspring estimation sample set, updating the optimization sample set, and jumping to the step of training a plurality of pre-established regression prediction models respectively by taking a process parameter set as input and taking an objective function value corresponding to the process parameter set as target output to obtain a plurality of trained regression prediction models.
Optionally, the constructing an optimized sample set specifically includes:
randomly generating a plurality of process parameter sets according to the constraint conditions of the process parameters;
aiming at any process parameter group, carrying out numerical simulation according to the process parameter group by using a numerical simulation model of a target shale oil well to obtain a target function value corresponding to the process parameter group; and the process parameter sets and the objective function values corresponding to the process parameter sets form an optimization sample set.
Optionally, the iteratively updating the initial population by using the evolutionary algorithm and the trained multiple regression prediction models to obtain a target population specifically includes:
carrying out cross variation operation on each process parameter group in the initial population to obtain an intermediate population; the intermediate population comprises offspring process parameter sets corresponding to the process parameter sets of the initial population;
determining the estimated objective function value of each filial generation process parameter set by using the trained multiple regression prediction models;
comparing the objective function value of each process parameter set with the estimated objective function value of the corresponding filial generation process parameter set, and updating the process parameter set in the initial population;
judging whether an iteration ending condition is met, if not, skipping to the step of carrying out cross variation operation on each process parameter group in the initial population to obtain an intermediate population; and the intermediate population comprises a offspring process parameter group corresponding to each process parameter group of the initial population.
Optionally, the determining the estimated objective function value of each parameter set of the offspring by using the trained multiple regression prediction models specifically includes:
respectively inputting the offspring process parameter sets into a plurality of trained parameter optimization models to obtain a plurality of objective function values of the offspring process parameter sets;
and averaging the plurality of objective function values of the filial generation process parameter set to obtain an estimated objective function value corresponding to the filial generation process parameter set.
Optionally, the re-determining the initial population according to the child estimation sample set, and updating the optimized sample set specifically includes:
calculating an uncertainty index of any child process parameter group; the uncertainty index is a variance between a plurality of objective function values of the set of descendant process parameters;
calculating a distance index of any child process parameter group; the distance index is an average value of Euclidean distances between the offspring process parameter set and other offspring process parameter sets in the offspring estimated sample set;
sorting the offspring process parameter sets in a descending order according to the uncertainty indexes of the offspring process parameter sets, and before selection, selectingp 1 The group of offspring process parameters of (2) is taken as a first sample set;
removing the first sample set from the child estimation sample set, performing descending sorting according to the objective function value of each child process parameter set, and selecting the first sample set before selectionp 2 The offspring technological parameter group of (2) is used as a second sample set;
removing the first sample set and the second sample set from the child estimation sample set, sorting in descending order according to the distance index of each child process parameter set, and selectingFront sidep 3 The offspring technological parameter group of (2) is used as a third sample set;
merging the first sample set, the second sample set and the third sample set to obtain a merged sample set;
performing numerical simulation on each filial generation process parameter set in the combined sample set by using a numerical simulation model of the target shale oil well to obtain an objective function value corresponding to each filial generation process parameter set;
and taking each filial generation process parameter group of the merged sample set and the corresponding objective function value as a new initial population, and adding the new initial population into the optimized sample set.
Optionally, the uncertainty index of the set of descendant process parameters is calculated according to the following formula:
wherein, the first and the second end of the pipe are connected with each other,s 2 (x) Is an uncertainty index of the set of descendant process parameters, x is the set of descendant process parameters,mas to the number of values of the objective function,
for the first of the child process parameter setsiThe value of each of the objective function values,
and the estimated objective function value of the filial generation process parameter set is obtained.
Optionally, the distance index of the group of child process parameters is calculated according to the following formula:
wherein the content of the first and second substances,dist(x) A distance index for the set of progeny process parameters,nestimating the number of sets of offspring process parameters in the sample set for the offspring,Dthe number of the process parameters in the process parameter set is shown,x j for the child process parameter groupjThe value of the individual process parameters is,x i,j is a firstiSetting the parameter group of the filial generationjThe value of the individual process parameters.
Optionally, the objective function value corresponding to the process parameter set is an accumulated oil yield or an economic net present value of the target shale oil well when the gas injection throughput process is implemented according to the process parameter set.
Optionally, the regression prediction model is any one of a deep neural network, an integrated decision tree, and a support vector machine.
Corresponding to the gas injection huff and puff parameter optimization method in shale oil exploitation, the invention also provides a gas injection huff and puff parameter optimization system in shale oil exploitation, which is used for optimizing each process parameter of a gas injection huff and puff process, and the gas injection huff and puff parameter optimization system comprises:
the optimized sample set constructing module is used for constructing an optimized sample set; the optimization sample set comprises a plurality of process parameter sets and objective function values corresponding to the process parameter sets;
an initial population determining module, configured to select, in the optimized sample set, a plurality of process parameter sets with a largest objective function value and corresponding objective function values as an initial population of an evolutionary algorithm;
the multi-model training module is used for taking a process parameter group as input, taking an objective function value corresponding to the process parameter group as target output, and respectively training a plurality of pre-established regression prediction models to obtain a plurality of trained regression prediction models; the structures of the regression prediction models are different from each other;
the population iterative evolution module is used for iteratively updating the initial population by utilizing an evolution algorithm and a plurality of trained regression prediction models to obtain a target population;
the offspring estimation sample set building module is used for taking the offspring process parameter sets of all the intermediate populations and the estimation objective function values corresponding to the offspring process parameter sets in the process of iteratively updating the initial population as an offspring estimation sample set;
the first judgment module is used for judging whether the optimization ending condition is reached or not to obtain a first judgment result;
the optimized sample set updating module is used for re-determining the initial population according to the child estimated sample set when the first judgment result is negative, updating the optimized sample set and calling the multi-model training module;
and the optimal process parameter determining module is used for taking the process parameter group with the maximum objective function value in the target population as the optimal process parameter group of the target shale oil well when the first judgment result is yes.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects:
the invention provides a gas injection huff and puff parameter optimization method and a gas injection huff and puff parameter optimization system in shale oil exploitation, wherein the gas injection huff and puff parameter optimization method comprises the following steps: constructing an optimized sample set; respectively training a plurality of regression prediction models which are established in advance by taking the process parameter set as input and the objective function value corresponding to the process parameter set as target output to obtain a plurality of trained regression prediction models; selecting a plurality of process parameter sets with the maximum objective function values and corresponding objective function values in the optimized sample set as initial populations of the evolutionary algorithm; iteratively updating the initial population by using an evolutionary algorithm and a plurality of trained regression prediction models to obtain a target population; constructing a child estimated sample set according to child process parameter sets of all intermediate populations in the process of iteratively updating the initial population and estimated objective function values corresponding to the child process parameter sets; judging whether an optimization ending condition is reached or not; if so, taking the offspring process parameter group with the maximum objective function value in the target population as the optimal process parameter group of the target shale oil well; otherwise, updating the optimized sample set according to the estimated offspring sample set, and retraining the multiple regression prediction models. According to the method, a plurality of groups of process parameters are generated in the value range of the process parameters, corresponding objective function values are obtained through numerical simulation, an optimized sample set is constructed, and the problem that the training effect on the model is poor due to the fact that the data quantity can be obtained is effectively avoided.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without creative efforts.
Fig. 1 is a flowchart of a method for optimizing gas injection throughput parameters in shale oil production according to embodiment 1 of the present invention;
FIG. 2 is a schematic plan view of a numerical simulation model in the method according to embodiment 1 of the present invention;
FIG. 3 is an iterative graph of net present economic value versus number of numerical simulations in the method provided in example 1 of the present invention;
FIG. 4 is a graph comparing the production dynamic curve of the optimization scheme and the conventional optimization scheme in the method provided in example 1 of the present invention;
FIG. 5 is a comparison graph of the extraction degree curves of the optimization scheme and the conventional optimization scheme in the method provided by example 1 of the present invention;
fig. 6 is a schematic structural diagram of a gas injection throughput parameter optimization system for shale oil production according to embodiment 2 of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The currently common process parameter optimization methods include the following three methods:
(1) And analyzing the oil reservoir development effect when different process parameters are different in value by combining methods such as numerical simulation and physical simulation experiment, and performing scheme optimization (CN 202011244646.6). The method can only carry out simple comparison and selection on the selected scheme, cannot find a globally optimal development scheme, and has poor effect.
(2) Indexes such as economic net present values and the like are used as optimization objective functions, values of the objective functions are calculated through numerical simulation, optimization algorithms such as genetic algorithms and the like are called to optimize process parameters, the method can obtain better optimization solutions, but repeated numerical reservoir simulation needs to be carried out in the operation process of the optimization algorithms, the needed time and the needed computing resources are large, and the algorithm efficiency is low.
(3) Compared with the method of directly calling numerical simulation, the method can save calculation resources to a certain extent, but the accuracy of the established proxy model can influence the reliability of an optimization result, and the method can improve the precision of the proxy model only by increasing the number of numerical simulation generated samples, so that the consumption of calculation resources can be increased, and the efficiency of the algorithm can be reduced. Therefore, the shale oil well gas injection huff and puff process parameter optimization method capable of ensuring the accuracy of the regression prediction model and improving the calculation efficiency has important significance.
The invention aims to provide a gas injection throughout parameter optimization method and system in shale oil exploitation, which improve the accuracy of prediction of objective function values of process parameter sets, and thus ensure the reliability of selection of the optimal process parameter sets.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
Example 1:
the embodiment provides a gas injection throughput parameter optimization method for shale oil exploitation, which comprises the following steps of:
s1, constructing an optimized sample set; the optimization sample set comprises a plurality of process parameter sets and objective function values corresponding to the process parameter sets; and the objective function value corresponding to the process parameter set is the accumulated oil yield or the economic net current value of the target shale oil well when the gas injection throughput process is implemented according to the process parameter set. The step S1 specifically includes:
and S11, randomly generating a plurality of process parameter sets according to the constraint conditions of the process parameters.
S12, aiming at any process parameter set, carrying out numerical simulation according to the process parameter set by using a numerical simulation model of the target shale oil well to obtain an objective function value corresponding to the process parameter set; and the process parameter sets and the objective function values corresponding to the process parameter sets form an optimization sample set.
S2, determining an initial population of the evolutionary algorithm; and selecting a plurality of process parameter sets with the maximum objective function values and corresponding objective function values in the optimized sample set as initial populations of the evolutionary algorithm.
S3, training the multiple regression prediction models; taking a process parameter group as input, taking an objective function value corresponding to the process parameter group as target output, and respectively training a plurality of regression prediction models established in advance to obtain a plurality of trained regression prediction models; the regression prediction models are different in structure; the regression prediction model is any one of a deep neural network, an integrated decision tree and a support vector machine.
S4, iteratively updating the initial population by using an evolutionary algorithm and a plurality of trained regression prediction models to obtain a target population; the method specifically comprises the following steps:
s41, carrying out cross variation operation on each process parameter group in the initial population to obtain an intermediate population; and the intermediate population comprises child process parameter sets corresponding to the process parameter sets of the initial population.
S42, determining the estimated objective function value of each filial generation process parameter set by using the trained multiple regression prediction models; the method specifically comprises the following steps:
and S421, respectively inputting the filial generation process parameter sets into the trained parameter optimization models to obtain a plurality of objective function values of the filial generation process parameter sets.
And S422, averaging the multiple objective function values of the offspring process parameter set to obtain the estimated objective function value corresponding to the offspring process parameter set.
S43, comparing the objective function values of the process parameter sets with the estimated objective function values of the corresponding filial process parameter sets, and updating the process parameter sets in the initial population; and comparing the parent process parameter set with the child process parameter set according to the requirement of the differential evolution algorithm, and taking the process parameter set with the high objective function value as a new parent process parameter set.
S44, judging whether an iteration end condition is reached, if not, jumping to the step of performing cross variation operation on each process parameter group in the initial population to obtain an intermediate population; and the intermediate population comprises a offspring process parameter group corresponding to each process parameter group of the initial population. In other embodiments, other population-based group intelligence optimization algorithms such as a particle swarm algorithm may be adopted, and the iteration end condition may be a preset maximum iteration step number.
S5, constructing a child estimation sample set; and constructing a child estimation sample set according to child process parameter sets of all intermediate populations generated in the process of iteratively updating the initial population and the estimation objective function values corresponding to the child process parameter sets.
And S6, judging whether the optimization finishing condition is reached, if so, executing a step S7, otherwise, executing a step S8. The optimization ending condition may be that the number of times of the numerical simulation reaches a preset maximum number of times or the regression prediction model reaches a preset minimum precision, and the like.
S7, determining an optimal process parameter set; and taking the offspring process parameter set with the maximum objective function value in the target population as the optimal process parameter set of the target shale oil well.
And S8, re-determining the initial population according to the child estimated sample set, updating the optimized sample set, and skipping to the step S3. Step S8 specifically includes:
before step S81, the number of sample sets selected by different sample update strategies is setp 1 ，p 2 ，p 3 Whereinp 1 ，p 2 ，p 3 The addition is equal to the size of the initial population.
S81, aiming at any filial generation process parameter group, calculating an uncertainty index of the filial generation process parameter group; the uncertainty index is a variance between a plurality of objective function values of the set of descendant process parameters; in this embodiment, the uncertainty index of the offspring process parameter set is calculated according to the following formula:
wherein, the first and the second end of the pipe are connected with each other,s 2 (x) Is an uncertainty index for the set of descendant process parameters, x is the set of descendant process parameters,mas to the number of values of the objective function,
set forth for the child process parameteriThe value of each of the objective functions is,
and the estimated objective function value of the filial generation process parameter set is obtained.
S82, calculating a distance index of any filial generation process parameter group aiming at the filial generation process parameter group; the distance index is an average value of Euclidean distances between the child generation process parameter set and other child generation process parameter sets in the child generation prediction sample set; in this embodiment, the distance index of the child process parameter set is calculated according to the following formula:
wherein the content of the first and second substances,dist(x) A distance index for the set of progeny process parameters,nestimating the number of sets of offspring process parameters in the sample set for the offspring,Dindicates the number of the process parameters in the process parameter set,x j set forth the parameters for the child processjThe value of the individual process parameters is,x i,j is as followsiSet forth the parameters of the offspring technologyjValues of individual process parameters.
S83, sorting the offspring process parameter sets in a descending order according to the uncertainty indexes of the offspring process parameter sets, and before selection, selecting the offspring process parameter setsp 1 The group of the child process parameters of the person serves as a first sample set.
S84, removing the first sample set from the child estimated sample set, sorting in a descending order according to the objective function value of each child process parameter set, and selecting the first sample set beforep 2 The child set of process parameters of the individuals serves as a second sample set.
S85, removing the first sample set and the second sample set from the child estimated sample set, sorting in descending order according to the distance index of each child process parameter set, and before selectionp 3 And taking the child process parameter group of the person as a third sample set.
S86, merging the first sample set, the second sample set and the third sample set to obtain a merged sample set.
And S87, performing numerical simulation on each filial generation process parameter set in the combined sample set by using a numerical simulation model of the target shale oil well to obtain an objective function value corresponding to each filial generation process parameter set.
And S88, taking the parameter sets of all filial generation process parameters of the merged sample set and the corresponding objective function values as new initial populations, and adding the new initial populations into the optimized sample set.
The method for optimizing gas injection throughput parameters in shale oil exploitation provided in embodiment 1 of the present invention is described below with reference to a specific example, where before optimizing the gas injection throughput parameters in shale oil exploitation, a gas injection throughput development numerical simulation model of a target shale oil well needs to be established according to shale oil geology, engineering data, and production development data of the target shale oil well to be optimized; specifically, the shale oil geology, engineering data and development data can be obtained from geological exploration data, well logging curves and other data of a target well, and the specific obtaining mode can be determined according to actual conditions. In this example, the basic parameters for establishing the numerical simulation model are shown in table 1, and a schematic plan view of the model is shown in fig. 2.
TABLE 1 numerical simulation model basis parameters
After a target shale oil well gas injection huff-puff development numerical simulation model is established, determining a target function of a process parameter group, process parameters and constraint conditions of the process parameters, selecting a basic optimization algorithm, and setting basic optimization algorithm control parameters and optimization method iteration stop conditions;
specifically, the objective function may be one of the accumulated oil amount or the economic net present value when the gas injection throughput development is performed according to the set of process parameters; in this example, the economic net present value of the target shale oil well is taken as the optimization objective function, and the maximized economic net present value is taken as the optimization objective, and the specific calculation formula of the economic net present value is shown as the following formula:
wherein u represents an optimized variable vector;N t the total number of time steps is indicated,nrepresenting the current time step number;t n the time of the current production time difference from the beginning of production is expressed in the unit of month;bthe annual interest rate;oilpriceis the oil price, the unit is Yuan,
the crude oil yield in m for the current production time 3 ；
For the operating cost, the unit is Yuan/Fang;
、
purchasing CO separately for injection 2 And recovery of CO 2 The unit of the cost is yuan/ton;FC、C well 、C fracture_k respectively setting the fixed cost, the drilling and completion cost and the cost of fracturing a single crack, wherein the unit is yuan;n f the number of press cracks.
The technological parameter can be one or any combination of gas injection throughput opportunity, throughput cycle number, single-cycle gas injection time, gas injection rate, single-cycle soaking time and single-cycle production time; the constraint conditions of the process parameters comprise upper and lower limits of the values of the process parameters and other constraint conditions, wherein the other constraint conditions can comprise that the wellhead pressure is not more than the maximum pressure provided by the gas injection compressor during gas injection, or the accumulated oil yield is not less than the preset minimum yield when the economic net present value is taken as an optimization objective function, and the like.
The selected process parameter group in the example is shown in table 2, the selected constraint conditions respectively comprise the upper limit and the lower limit of each process parameter, which are also shown in table 2, and the special constraint is that the bottom hole pressure does not exceed 85MPa;
TABLE 2 set of Process parameters and constraints
Specifically, the basic optimization algorithm may be a population-based group intelligence optimization algorithm such as a differential evolution algorithm, a particle swarm algorithm, and the like; the algorithm control parameters are control parameters corresponding to the selected basic optimization algorithm, for example, when the basic optimization algorithm is selected as a differential evolution algorithm, the basic optimization algorithm control parameters include population size, variation factors, cross probability, maximum iteration step number and the like; in this example, a differential evolution algorithm is selected as a basic optimization algorithm, and specific optimization algorithm control parameters are shown in table 3.
TABLE 3 optimization algorithm control parameters
The iteration stop condition of the optimization method can be that the preset maximum numerical simulation times are reached or the regression prediction model reaches the preset minimum precision and the like.
In the foregoing S1, an optimized sample set is constructed, specifically in this example, as follows: developing a numerical simulation model based on the gas injection throughput of the target shale oil well, sampling to generate an optimized sample set comprising a plurality of process parameter sets, and determining an initial population of a basic optimization algorithm;
specifically, sampling to generate an optimized sample set including a plurality of process parameter sets may include: determining the upper and lower value limits of each process parameter according to the constraint conditions of the process parameters; randomly sampling within the upper limit and the lower limit of the value of each process parameter by using a Monte Carlo or Latin hypercube sampling method to form a group of process parameter groups, and repeatedly performing random sampling according to preset times; and carrying out numerical simulation according to each process parameter set by using the gas injection throughput development numerical simulation model of the target shale oil well to obtain an objective function value corresponding to each process parameter set, wherein all the process parameter sets and the corresponding objective function values jointly form an optimized sample set.
In the embodiment, the total accumulated sampling is carried out for 150 times, 150 groups of process parameter sets are generated, and numerical simulation is respectively carried out on each process parameter set by utilizing a numerical simulation model to determine the corresponding objective function value, so that an optimized sample set is formed; then selecting 25 groups of samples with the maximum objective function value as an initial population of a basic optimization algorithm;
the step S2 determines an initial population of the evolutionary algorithm, and the step of determining the initial population may specifically include: sequencing according to the size of the objective function value of each process parameter set according to the optimization sample set; and selecting a corresponding number of samples with the maximum objective function from the optimized sample set according to the set population size as an initial population.
Step S3 is to train a plurality of regression prediction models; before training, a regression prediction model group should be established in advance based on the optimization sample set;
specifically, the number m of regression prediction models is set, the regression prediction model group is a model group formed by m regression prediction models, each regression prediction model can be any regression prediction model such as a deep neural network, an integrated decision tree, a support vector machine and the like, and the structures of all the models are different; m is a positive integer greater than or equal to 3 and can be set arbitrarily; in this example, the regression prediction model group is composed of 3 different regression prediction models, which respectively include a deep confidence network model, an integrated decision tree model and a regression support vector machine model; and (3) taking the value of each process parameter set in the optimized sample set as characteristic input, taking the corresponding objective function value as label output, and respectively training each regression prediction model to form a regression prediction model group of the target shale oil well gas injection huff and puff process.
Respectively selecting a specific algorithm adopted by each regression prediction model; respectively training each regression prediction model by using the process parameter sets in the optimized sample set as input characteristic data and the objective function value corresponding to each process parameter set as target output label data; and the trained regression prediction models form the regression prediction model group.
And S4, iteratively updating the initial population by using an evolutionary algorithm and the trained multiple regression prediction models to obtain a target population. Specifically, in the embodiment, a regression prediction model group and an initial population of the target shale oil well are utilized, a basic optimization algorithm is called for optimization solution, a process parameter group and a corresponding objective function value of each generation of intermediate population in optimization iteration are recorded, and a target population and a offspring estimation sample set obtained by the optimization are obtained; the specific steps can be subdivided into:
calculating and acquiring a child process parameter group corresponding to each process parameter group in the initial population according to the initial population by using the basic optimization algorithm;
predicting m estimated objective function values of each filial generation process parameter set by using m regression prediction models in the regression prediction model group, and averaging the m estimated objective function values to serve as the estimated objective function values of each filial generation process parameter set;
sequentially comparing the objective function value of each process parameter set with the estimated objective function value of the corresponding filial generation process parameter set, and selecting a more optimal individual as a next generation individual to replace the current population;
judging whether the maximum iteration step number of the basic optimization algorithm is met, if so, selecting the optimized final generation target population as the optimization result to be output, and if not, calculating according to the current intermediate population to obtain the next generation population and continuing iteration;
in this example, in the process of executing the step of calculating the estimated objective function value of each offspring process parameter set, the step of S5 constructing an offspring estimated sample set is completed at the same time: and recording the m estimated objective function values and the estimated objective function values of all the offspring process parameter sets into an offspring estimated sample set.
Step S6, judging whether the iteration stop condition of the optimization method is met currently, if so, determining the process parameters when the objective function of the shale oil well is optimal according to the objective population of the optimization result, otherwise, estimating a sample set according to the offspring, updating the optimized sample set by using a multi-sample updating strategy, determining a new initial population of the optimization algorithm, and retraining a regression prediction model group; specifically, the multiple sample updating strategy comprises an active learning strategy, an optimal individual strategy and a farthest individual strategy.
The estimating a sample set according to the child, updating the optimized sample set by using a multiple sample updating strategy, and determining the content of the initial population of the new optimization algorithm, that is, the step S8 specifically includes:
corresponding to the step S81, respectively calculating each offspring process parameter set according to the offspring estimated sample setmThe variance between the estimated objective function values is used as an uncertainty index of each filial generation process parameter set; in this embodiment, the uncertainty index of the offspring process parameter set is calculated according to the following formula:
wherein the content of the first and second substances,s 2 (x) Is an uncertainty index for the set of descendant process parameters, x is the set of descendant process parameters,mas to the number of values of the objective function,
set forth for the child process parameteriThe value of each of the objective functions is,
and the estimated objective function value of the filial generation process parameter set is obtained.
Corresponding to the step S82, respectively calculating the euclidean distance between each process parameter of each offspring process parameter set and each process parameter of the rest offspring process parameter sets according to the offspring estimated sample set, and averaging the euclidean distances to serve as the distance index of each offspring process parameter set; in this embodiment, the distance index of the child process parameter set is calculated according to the following formula:
wherein, the first and the second end of the pipe are connected with each other,dist(x) A distance index for the set of descendant process parameters,nestimating the number of sets of child process parameters in the sample set for the child,Dthe number of the process parameters in the process parameter set is shown,x j set forth the parameters for the child processjThe value of the individual process parameters is,x i,j is as followsiSet forth the parameters of the offspring technologyjThe value of the individual process parameters.
Corresponding to the steps S83 to S88:
sorting the filial generation process parameter sets in a descending order according to the uncertainty indexes of the filial generation process parameter sets, and before selection, selectingp 1 The child process parameter group of individuals is used as a first sample set;
removing the first sample set from the offspring estimated sample set, and processing parameters according to each offspringThe objective function values of the array are sorted in descending order before being selectedp 2 Using the offspring technological parameter group of the individuals as a second sample set;
removing the first sample set and the second sample set from the child estimation sample set, sorting in descending order according to the distance index of each child process parameter set, and selecting the previous sample setp 3 Using the offspring technological parameter group of the individuals as a third sample set;
merging the first sample set, the second sample set and the third sample set to obtain a merged sample set;
performing numerical simulation on each filial generation process parameter set in the combined sample set by using a numerical simulation model of the target shale oil well to obtain an objective function value corresponding to each filial generation process parameter set;
and taking the parameter sets of all filial generation process parameters of the merged sample set and the corresponding objective function values as new initial populations, and adding the new initial populations into the optimized sample set. In the embodiment, the precision of the proxy model is continuously improved through an active learning strategy in the multiple sample updating, the reliability of an optimization result is ensured, the development of a possible optimal solution in the optimization algorithm optimization process and the exploration of a far position in an optimized solution space are balanced by using an optimal sample strategy and a farthest distance individual strategy, and the capability of optimally jumping out of a local optimal solution can be effectively improved on the premise of not losing the solving efficiency, so that the more rapid and effective shale oil well gas injection huff-puff process parameter optimization is realized.
In this specific embodiment, a total of 5 sample update processes are performed, each sample update process is performed for 25 times of value simulation, and the number of numerical simulation times required for constructing an initial optimized sample set is also included, the whole optimization process is cumulatively performed for 275 times of value simulation, an iteration curve of the optimal economic net present value in the optimization process increasing along with the number of numerical simulation times is shown in fig. 3, and the obtained optimal process parameter set is shown in table 4.
TABLE 4 set of finalized optimal process parameters
Comparing the optimal process parameter set obtained by the optimization method provided by the embodiment of the invention with the optimal scheme obtained by the traditional orthogonal test design, a comparison graph of the production dynamic curve of the target well under the optimal process parameter scheme obtained by the optimization method provided by the embodiment of the invention with the production dynamic curve of the target well under the optimal process parameter scheme designed by the orthogonal test is shown in fig. 4, and a comparison graph of the extraction degree curve is shown in fig. 5. Compared with the optimal scheme of orthogonal test design, the method for optimizing the process parameters provided by the embodiment of the invention has the advantages that the accumulated oil yield of the target well under the optimal process parameter scheme is improved by 49.13%, the extraction degree is improved by 3.8 percentage points, and the economic net present value is improved by 35.66%.
In this embodiment, in step S11, a plurality of sets of process parameters are generated within the value range of the process parameters, and then the corresponding objective function values are obtained through numerical simulation, so as to construct an optimized sample set, thereby effectively avoiding the problem of poor model training effect caused by insufficient obtainable data volume.
In step S3, a plurality of regression prediction models with different structures are trained through the optimized sample set generated by simulation, and compared with a method for predicting objective function values by using a single model in the prior art, the method provided by the invention can analyze the characteristics of training data in different layers.
Meanwhile, in the optimization process of the step S4, a differential evolution algorithm is used for carrying out cross variation on the process parameter sets in the initial population, a descendant estimated sample set is constructed, the content of the step S8 is executed when the optimization end condition is not reached, the optimized sample set is updated through the descendant estimated sample set, and the regression prediction models are trained again, so that the data volume participating in model training can be promoted again, the accuracy degree of the estimated objective function value determined through the regression prediction models is ensured, and the reliability of optimization of the gas injection handling process parameters is ensured.
Example 2:
as shown in a schematic structural diagram of fig. 6, in correspondence to the method for optimizing throughput parameters in shale oil production provided in embodiment 1, this embodiment provides a system for optimizing throughput parameters in shale oil production, including:
the optimized sample set constructing module is used for constructing an optimized sample set; the optimization sample set comprises a plurality of process parameter sets and objective function values corresponding to the process parameter sets;
an initial population determining module, configured to select, in the optimized sample set, a plurality of process parameter sets with a largest objective function value and corresponding objective function values as an initial population of an evolutionary algorithm;
the multi-model training module is used for respectively training a plurality of pre-established regression prediction models by taking a process parameter group as input and taking an objective function value corresponding to the process parameter group as target output to obtain a plurality of trained regression prediction models; the structures of the regression prediction models are different from each other;
the population iterative evolution module is used for iteratively updating the initial population by utilizing a differential evolution algorithm and a plurality of trained regression prediction models to obtain a target population;
the child estimation sample set construction module is used for taking the child process parameter sets of all the intermediate populations and the estimation objective function values corresponding to the child process parameter sets in the process of iteratively updating the initial population as child estimation sample sets;
the first judgment module is used for judging whether the optimization ending condition is reached or not to obtain a first judgment result;
the optimized sample set updating module is used for updating the optimized sample set according to the child estimated sample set when the first judgment result is negative; re-determining the initial population, and calling the multi-model training module;
and the optimal process parameter determining module is used for taking the process parameter group with the maximum objective function value in the target population as the optimal process parameter group of the target shale oil well when the first judgment result is yes.
Portions of the technology may be considered "articles of manufacture" or "articles of manufacture" in the form of executable code and/or associated data embodied in or carried out by a computer readable medium. Tangible, non-transitory storage media may include memory or storage for use by any computer, processor, or similar device or associated module. For example, various semiconductor memories, tape drives, disk drives, or any similar device capable of providing a storage function for software.
All or a portion of the software may sometimes communicate over a network, such as the internet or other communication network. Such communication may load software from one computer device or processor to another. For example: from a server or host computer of the video object detection device to a hardware platform of a computer environment, or other computer environment implementing a system, or similar functionality related to providing information needed for object detection. Thus, another medium capable of transferring software elements may also be used as a physical connection between local devices, such as optical, electrical, electromagnetic waves, etc., propagating through cables, optical cables, air, etc. The physical medium used for the carrier wave, such as an electrical, wireless connection, or optical cable, etc., can also be considered to be the medium carrying the software. As used herein, unless limited to a tangible "storage" medium, other terms referring to a computer or machine "readable medium" refer to media that participate in the execution of any instructions by a processor.
Although specific examples are employed herein, the foregoing description is only illustrative of the principles and implementations of the present invention, and the following examples are provided only to facilitate the understanding of the method and its core concepts; those skilled in the art will appreciate that the modules or steps of the invention described above can be implemented using general purpose computing apparatus, or alternatively, they can be implemented using program code executable by computing apparatus, such that it is executed by computing apparatus when stored in a storage device, or separately fabricated into integrated circuit modules, or multiple modules or steps thereof can be fabricated into a single integrated circuit module. The present invention is not limited to any specific combination of hardware and software.
Meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.
Claims (8)
1. A gas injection huff and puff parameter optimization method in shale oil exploitation is characterized by being used for optimizing various process parameters of a gas injection huff and puff process, and the gas injection huff and puff parameter optimization method in shale oil exploitation comprises the following steps:
constructing an optimized sample set; the optimization sample set comprises a plurality of process parameter sets and objective function values corresponding to the process parameter sets; the objective function value corresponding to the process parameter set is the accumulated oil yield or the economic net present value of the target shale oil well when the gas injection huff and puff process is implemented according to the process parameter set;
selecting a plurality of process parameter sets with the maximum objective function values and corresponding objective function values in the optimized sample set as initial populations of the evolutionary algorithm;
taking a process parameter group as input, taking an objective function value corresponding to the process parameter group as target output, and respectively training a plurality of regression prediction models established in advance to obtain a plurality of trained regression prediction models; the regression prediction models are different in structure;
iteratively updating the initial population by using an evolutionary algorithm and a plurality of trained regression prediction models to obtain a target population;
constructing a child estimation sample set according to child process parameter sets of all intermediate populations generated in the process of iteratively updating the initial population and estimation objective function values corresponding to the child process parameter sets;
judging whether an optimization ending condition is reached; if so, taking the offspring process parameter set with the maximum objective function value in the target population as the optimal process parameter set of the target shale oil well;
otherwise, re-determining the initial population according to the child estimated sample set, and updating the optimized sample set; skipping to the step of taking a process parameter group as input, taking an objective function value corresponding to the process parameter group as target output, and respectively training a plurality of pre-established regression prediction models to obtain a plurality of trained regression prediction models;
the re-determining the initial population according to the child estimated sample set and updating the optimized sample set specifically includes:
calculating an uncertainty index of any offspring process parameter set; the uncertainty index is a variance between a plurality of objective function values of the set of descendant process parameters;
calculating a distance index of any child process parameter group; the distance index is an average value of Euclidean distances between the offspring process parameter set and other offspring process parameter sets in the offspring estimated sample set;
sorting the offspring process parameter sets in a descending order according to the uncertainty indexes of the offspring process parameter sets, and before selection, selectingp 1 The group of offspring process parameters of (2) is used as a first sample set;
removing the first sample set from the child estimation sample set, performing descending sorting according to the objective function value of each child process parameter set, and selecting the first sample set before selectionp 2 The offspring technological parameter group of (2) is used as a second sample set;
removing the first sample set and the second sample set from the child estimation sample set, sorting in descending order according to the distance index of each child process parameter set, and selecting the previous sample setp 3 The descendant process parameter group of (2) is used as a third sample set;
merging the first sample set, the second sample set and the third sample set to obtain a merged sample set;
performing numerical simulation on each filial generation process parameter set in the combined sample set by using a numerical simulation model of the target shale oil well to obtain an objective function value corresponding to each filial generation process parameter set;
and taking each filial generation process parameter group of the merged sample set and the corresponding objective function value as a new initial population, and adding the new initial population into the optimized sample set.
2. The shale oil recovery gas injection throughput parameter optimization method of claim 1, wherein the constructing an optimized sample set specifically comprises:
randomly generating a plurality of process parameter sets according to the constraint conditions of the process parameters;
aiming at any process parameter set, carrying out numerical simulation according to the process parameter set by using a numerical simulation model of a target shale oil well to obtain an objective function value corresponding to the process parameter set; and the process parameter sets and the objective function values corresponding to the process parameter sets form an optimization sample set.
3. The method for optimizing injection gas throughput parameters in shale oil recovery according to claim 1, wherein the initial population is iteratively updated by using an evolutionary algorithm and a plurality of trained regression prediction models to obtain a target population, and specifically comprises:
carrying out cross variation operation on each process parameter group in the initial population to obtain an intermediate population; the intermediate population comprises child process parameter sets corresponding to the process parameter sets of the initial population;
determining the estimated objective function value of each filial generation process parameter set by using the trained multiple regression prediction models;
comparing the objective function value of each process parameter set with the estimated objective function value of the corresponding filial generation process parameter set, and updating the process parameter set in the initial population;
judging whether an iteration ending condition is met, if not, skipping to the step of carrying out cross variation operation on each process parameter group in the initial population to obtain an intermediate population; and the intermediate population comprises child process parameter sets corresponding to the process parameter sets of the initial population.
4. The method for optimizing gas injection throughput parameters in shale oil production according to claim 3, wherein the determining the estimated objective function value of each offspring process parameter set by using a plurality of trained regression prediction models specifically comprises:
respectively inputting the offspring process parameter sets into a plurality of trained parameter optimization models to obtain a plurality of objective function values of the offspring process parameter sets;
and averaging the plurality of objective function values of the filial generation process parameter set to obtain an estimated objective function value corresponding to the filial generation process parameter set.
5. The method of optimizing injection gas throughput parameters in shale oil recovery of claim 1, wherein the uncertainty index of the set of offspring process parameters is calculated according to the following formula:
wherein, the first and the second end of the pipe are connected with each other,s 2 (x) Is an uncertainty index of the set of descendant process parameters, x is the set of descendant process parameters,mas to the number of values of the objective function,
for the first of the child process parameter setsiThe value of each of the objective function values,
and the estimated objective function value of the filial generation process parameter set is obtained.
6. The method of optimizing injection gas throughput parameters in shale oil recovery of claim 1, wherein the distance index of the set of offspring process parameters is calculated according to the following formula:
wherein the content of the first and second substances,dist(x) A distance index for the set of progeny process parameters,nestimating offspring process parameters for the offspring in the sample setThe number of the groups is such that,Dindicates the number of the process parameters in the process parameter set,x j for the child process parameter groupjThe value of the individual process parameters is,x i,j is as followsiSet forth the parameters of the offspring technologyjThe value of the individual process parameters.
7. The method of optimizing injection gas throughput parameters in shale oil recovery according to claim 1, wherein the regression prediction model is any one of a deep neural network, an integrated decision tree and a support vector machine.
8. The gas injection huff and puff parameter optimizing system for shale oil exploitation is used for optimizing various process parameters of a gas injection huff and puff process, and comprises the following components:
the optimized sample set constructing module is used for constructing an optimized sample set; the optimization sample set comprises a plurality of process parameter sets and objective function values corresponding to the process parameter sets; the objective function value corresponding to the process parameter set is the accumulated oil yield or the economic net present value of the target shale oil well when the gas injection huff and puff process is implemented according to the process parameter set;
an initial population determining module, configured to select, in the optimized sample set, a plurality of process parameter sets with a largest objective function value and corresponding objective function values as an initial population of an evolutionary algorithm;
the multi-model training module is used for taking a process parameter group as input, taking an objective function value corresponding to the process parameter group as target output, and respectively training a plurality of pre-established regression prediction models to obtain a plurality of trained regression prediction models; the regression prediction models are different in structure;
the population iterative evolution module is used for iteratively updating the initial population by utilizing an evolution algorithm and a plurality of trained regression prediction models to obtain a target population;
the offspring estimation sample set building module is used for taking the offspring process parameter sets of all the intermediate populations and the estimation objective function values corresponding to the offspring process parameter sets in the process of iteratively updating the initial population as an offspring estimation sample set;
the first judgment module is used for judging whether an optimization ending condition is met or not to obtain a first judgment result;
an optimized sample set updating module, configured to, when the first determination result is negative, re-determine an initial population according to the child estimated sample set, update the optimized sample set, and invoke the multi-model training module;
the optimized sample set updating module is specifically configured to: calculating an uncertainty index of any child process parameter group; the uncertainty index is a variance between a plurality of objective function values of the set of descendant process parameters; calculating a distance index of any child process parameter group; the distance index is an average value of Euclidean distances between the offspring process parameter set and other offspring process parameter sets in the offspring estimated sample set; sorting the filial generation process parameter sets in a descending order according to the uncertainty indexes of the filial generation process parameter sets, and before selection, selectingp 1 The group of offspring process parameters of (2) is used as a first sample set; removing the first sample set from the child estimation sample set, performing descending sorting according to the objective function value of each child process parameter set, and selecting the first sample set before selectionp 2 The offspring process parameter group of (2) is used as a second sample set; removing the first sample set and the second sample set from the child estimation sample set, sorting in descending order according to the distance index of each child process parameter set, and selecting the previous sample setp 3 The descendant process parameter group of (2) is used as a third sample set; merging the first sample set, the second sample set and the third sample set to obtain a merged sample set; carrying out numerical simulation on each filial generation process parameter set in the combined sample set by using a numerical simulation model of a target shale oil well to obtain an objective function value corresponding to each filial generation process parameter set; taking each filial generation process parameter group of the merged sample set and the corresponding objective function value as a new initial population, and adding the new initial population into the optimized sample set;
and the optimal process parameter determining module is used for taking the process parameter group with the maximum objective function value in the target population as the optimal process parameter group of the target shale oil well when the first judgment result is yes.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211133564.3A CN115204064B (en) | 2022-09-19 | 2022-09-19 | Gas injection huff-puff parameter optimization method and system in shale oil exploitation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211133564.3A CN115204064B (en) | 2022-09-19 | 2022-09-19 | Gas injection huff-puff parameter optimization method and system in shale oil exploitation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115204064A CN115204064A (en) | 2022-10-18 |
| CN115204064B true CN115204064B (en) | 2022-12-02 |
Family
ID=83572580
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211133564.3A Active CN115204064B (en) | 2022-09-19 | 2022-09-19 | Gas injection huff-puff parameter optimization method and system in shale oil exploitation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115204064B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115470910A (en) * | 2022-10-20 | 2022-12-13 | 晞德软件(北京)有限公司 | Automatic parameter adjusting method based on Bayesian optimization and K-center sampling |
| CN116401236B (en) * | 2023-06-07 | 2023-08-18 | 瀚高基础软件股份有限公司 | Method and equipment for adaptively optimizing database parameters |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105050132A (en) * | 2015-08-10 | 2015-11-11 | 北京邮电大学 | Method for estimating extreme value throughput capacity of cell |
| CN112379594A (en) * | 2020-11-09 | 2021-02-19 | 武汉理工大学 | Data-driven carbon fiber stock solution temperature control parameter optimization method |
Family Cites Families (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105303251A (en) * | 2015-10-10 | 2016-02-03 | 中国矿业大学 | Interval support vector machine model for solar seawater desalination problem, and optimized design thereof |
| CN108960486B (en) * | 2018-06-12 | 2021-10-29 | 郑州航空工业管理学院 | Interactive set evolution method for predicting adaptive value based on gray support vector regression |
| CN109325616A (en) * | 2018-09-01 | 2019-02-12 | 哈尔滨工程大学 | It is a kind of to be returned based on Gaussian process and the fine particle that combines of glowworm swarm algorithm is predicted and source tracing method |
| US11151175B2 (en) * | 2018-09-24 | 2021-10-19 | International Business Machines Corporation | On-demand relation extraction from text |
| CN109236258B (en) * | 2018-10-27 | 2019-07-23 | 中国地质大学（北京） | A kind of compact oil reservoir pressure break horizontal well optimization method based on Adaptive proxy model |
| CN112152267B (en) * | 2019-06-27 | 2022-07-29 | 浙江工业大学 | Power grid random reactive power optimization scheduling method considering source load uncertainty |
| CN110782366B (en) * | 2019-10-12 | 2022-09-06 | 大连理工大学 | Water supply pipe network multi-objective optimization regulation and control method based on scene uncertainty |
| US11574703B2 (en) * | 2019-12-23 | 2023-02-07 | TeselaGen Biotechnology Inc. | Method, apparatus, and computer-readable medium for efficiently optimizing a phenotype with a combination of a generative and a predictive model |
| CN111861774B (en) * | 2020-06-22 | 2021-06-04 | 中国石油大学(华东) | Oil reservoir production machine learning method based on parallel agent model |
| CN114580256A (en) * | 2020-12-01 | 2022-06-03 | 中国石油化工股份有限公司 | High-water-content oil reservoir well position optimization method based on seepage characteristic zone |
| CN113610229A (en) * | 2021-08-10 | 2021-11-05 | 白雪云 | Active learning method based on norm uncertainty index |
| CN114519274A (en) * | 2022-02-21 | 2022-05-20 | 中国石油大学(华东) | Gas drive reservoir injection-production parameter step-by-step optimization method |
| CN114595608B (en) * | 2022-03-11 | 2023-06-27 | 中国石油大学(华东) | Fracturing construction parameter and working system parameter optimization method and system |
| CN114759590A (en) * | 2022-04-29 | 2022-07-15 | 武汉大学 | Converter station steady-state reactive power coordination strategy optimization method and device based on data driving |
| CN114969953B (en) * | 2022-07-28 | 2023-02-03 | 华中科技大学 | Optimized shield underpass tunnel design method and equipment based on Catboost-NSGA-III |
-
2022
- 2022-09-19 CN CN202211133564.3A patent/CN115204064B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105050132A (en) * | 2015-08-10 | 2015-11-11 | 北京邮电大学 | Method for estimating extreme value throughput capacity of cell |
| CN112379594A (en) * | 2020-11-09 | 2021-02-19 | 武汉理工大学 | Data-driven carbon fiber stock solution temperature control parameter optimization method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115204064A (en) | 2022-10-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115204064B (en) | Gas injection huff-puff parameter optimization method and system in shale oil exploitation | |
| Shi et al. | Development of subsurface geological cross-section from limited site-specific boreholes and prior geological knowledge using iterative convolution XGBoost | |
| CN109611087B (en) | Volcanic oil reservoir parameter intelligent prediction method and system | |
| CN109416769B (en) | Computer-implemented method of generating an oil Field Development Plan (FDP) for producing hydrocarbon reservoirs | |
| Hutahaean et al. | On optimal selection of objective grouping for multiobjective history matching | |
| CN113359212B (en) | Reservoir characteristic prediction method and model based on deep learning | |
| CN112761628B (en) | Shale gas yield determination method and device based on long-term and short-term memory neural network | |
| CN114723095A (en) | Missing well logging curve prediction method and device | |
| Raji et al. | A new approach in well placement optimization using metaheuristic algorithms | |
| Alpak et al. | Biobjective optimization of well placement: algorithm, validation, and field testing | |
| CN111058840A (en) | Organic carbon content (TOC) evaluation method based on high-order neural network | |
| CN116665067A (en) | Ore finding target area optimization system and method based on graph neural network | |
| CN113065279A (en) | Method, device, equipment and storage medium for predicting total organic carbon content | |
| Brown et al. | Machine learning on Crays to optimize petrophysical workflows in oil and gas exploration | |
| Du et al. | Production forecasting with the interwell interference by integrating graph convolutional and long short-term memory neural network | |
| Qi et al. | An evolutionary sequential transfer optimization algorithm for well placement optimization based on task characteristics | |
| CN117473305A (en) | Method and system for predicting reservoir parameters enhanced by neighbor information | |
| Ma et al. | Design of steam alternating solvent process operational parameters considering shale heterogeneity | |
| CN116927771A (en) | Method, device, equipment and medium for predicting total organic carbon data of shale reservoir | |
| CN108133286B (en) | Underground water multi-target calculation method based on ground settlement substitution model | |
| Arouri et al. | Adaptive Rank-Based Selection of Geological Realizations for Optimum Field Development Planning | |
| Hamdi et al. | Gaussian process for uncertainty quantification of reservoir models | |
| Manasipov et al. | Physics informed machine learning for production forecast | |
| CN112749082B (en) | Test case generation method and system based on DE-TH algorithm | |
| Juda et al. | An attempt to boost posterior population expansion using fast machine learning algorithms |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |