CN113761613A - Drilling fluid design method and system - Google Patents
Drilling fluid design method and system Download PDFInfo
- Publication number
- CN113761613A CN113761613A CN202010503095.4A CN202010503095A CN113761613A CN 113761613 A CN113761613 A CN 113761613A CN 202010503095 A CN202010503095 A CN 202010503095A CN 113761613 A CN113761613 A CN 113761613A
- Authority
- CN
- China
- Prior art keywords
- drilling fluid
- new
- library
- similar
- new problem
- 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
Links
- 238000005553 drilling Methods 0.000 title claims abstract description 176
- 239000012530 fluid Substances 0.000 title claims abstract description 168
- 238000013461 design Methods 0.000 title claims abstract description 124
- 238000000034 method Methods 0.000 title claims abstract description 47
- 238000013528 artificial neural network Methods 0.000 claims description 36
- 230000015572 biosynthetic process Effects 0.000 claims description 27
- 238000012423 maintenance Methods 0.000 claims description 11
- 238000005457 optimization Methods 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 9
- 238000011282 treatment Methods 0.000 claims description 9
- 238000012937 correction Methods 0.000 claims description 8
- 239000011148 porous material Substances 0.000 claims description 7
- 230000035945 sensitivity Effects 0.000 claims description 6
- VNWKTOKETHGBQD-UHFFFAOYSA-N methane Chemical compound C VNWKTOKETHGBQD-UHFFFAOYSA-N 0.000 abstract description 6
- 239000003345 natural gas Substances 0.000 abstract description 3
- 238000011161 development Methods 0.000 abstract description 2
- 239000003209 petroleum derivative Substances 0.000 abstract description 2
- 239000003921 oil Substances 0.000 description 12
- 239000013598 vector Substances 0.000 description 10
- 238000010276 construction Methods 0.000 description 9
- 238000005516 engineering process Methods 0.000 description 7
- 230000008569 process Effects 0.000 description 7
- 238000004364 calculation method Methods 0.000 description 6
- 230000006870 function Effects 0.000 description 6
- 239000007789 gas Substances 0.000 description 6
- 238000004422 calculation algorithm Methods 0.000 description 5
- 238000012549 training Methods 0.000 description 5
- 239000000203 mixture Substances 0.000 description 4
- 238000003860 storage Methods 0.000 description 4
- 238000000605 extraction Methods 0.000 description 3
- 238000009472 formulation Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000007789 sealing Methods 0.000 description 3
- 239000002734 clay mineral Substances 0.000 description 2
- 238000012938 design process Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 239000004973 liquid crystal related substance Substances 0.000 description 2
- 238000011418 maintenance treatment Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 239000003129 oil well Substances 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 239000004215 Carbon black (E152) Substances 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 150000001875 compounds Chemical class 0.000 description 1
- 239000000470 constituent Substances 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 239000010779 crude oil Substances 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 229930195733 hydrocarbon Natural products 0.000 description 1
- 150000002430 hydrocarbons Chemical class 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000006698 induction Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012067 mathematical method Methods 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000004806 packaging method and process Methods 0.000 description 1
- 230000035699 permeability Effects 0.000 description 1
- 239000003208 petroleum Substances 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000012552 review Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/13—Architectural design, e.g. computer-aided architectural design [CAAD] related to design of buildings, bridges, landscapes, production plants or roads
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Geometry (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Computer Hardware Design (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Architecture (AREA)
- Civil Engineering (AREA)
- Structural Engineering (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a drilling fluid design method and a drilling fluid design system, and belongs to the field of petroleum and natural gas exploration and development and computer science. The method comprises the following steps: (1) establishing a case library; (2) acquiring each characteristic attribute of the new problem; (3) searching the most similar example in the example library as a similar example according to each characteristic attribute of the new problem; (4) judging whether the similar examples meet the requirements, if so, turning to the step (7), and if not, turning to the step (5); (5) correcting and optimizing the similar examples to obtain a solution of a new problem, and forming the new problem and the solution of the new problem into a new example; (6) inputting the new case into the case library, and then returning to the step (3): (7) and outputting the design scheme of the drilling fluid. The invention effectively improves the scientificity and the rapidness of drilling fluid design based on historical data, effectively reduces the complex risk of underground engineering and avoids reservoir pollution to the maximum extent.
Description
Technical Field
The invention belongs to the fields of petroleum and natural gas exploration and development and computer science, and particularly relates to a drilling fluid design method and system.
Background
The design of the drilling fluid plays a significant role in drilling engineering. Research finds that oil companies or drilling fluid technical service companies mainly rely on indoor tests and expert experience for drilling fluid design, and the design content comprises the selection of a drilling fluid system, the control of drilling fluid performance parameters, the type and the amount of drilling fluid materials, field maintenance treatment measures and the like. The unreasonable design of the drilling fluid can cause serious engineering complexity, such as well leakage, well wall collapse, drill sticking and drill bit mud bags, which increases the construction cost or causes reservoir pollution to influence the achievement of the oilfield capacity construction target.
In recent years, with the continuous improvement of the informatization level of oil fields, on the basis of a drilling engineering database, people explore a lot in the aspect of optimizing drilling fluid design, introduce the relevant theory and technology of an expert system into the drilling fluid design, analyze and establish the framework of the drilling fluid design expert system from the aspects of knowledge acquisition, knowledge representation and the like, such as drilling engineering design software developed by Changqing oil field companies, a drilling fluid expert system developed by Shengli oil field companies, and a drilling fluid formula intelligent designer disclosed by Chinese patent publication CN1760854 (by applying data preprocessing and algorithm, a drilling fluid formula with specified performance is designed, and comprises a data memory, a central processing unit, an algorithm program module, a shockproof medium, a liquid crystal display screen, an external packaging box data memory, the central processing unit and the algorithm program module which are sealed in a sealing box, the sealing box is filled with the shockproof medium, the sealing box is positioned below the liquid crystal display screen, and the system starting button and the system function input selection button are arranged on the function panel in parallel. ) Etc., but these systems provide only a simple data-based query and do not lend themselves to decision-making opinions on design projects.
Also, the conventional expert system has the following problems: a large amount of expert knowledge needs to be collected, a system frame needs to be established, time and labor are wasted in later maintenance, scientificity and rapidness are poor, risks caused by complex underground engineering cannot be effectively reduced, and reservoir pollution cannot be avoided.
Disclosure of Invention
The invention aims to solve the problems in the prior art and provides a drilling fluid design method and a drilling fluid design system, which are used for designing a drilling fluid based on the example reasoning of a twin neural network and improving the accuracy and convenience of the design of the conventional drilling fluid.
The invention is realized by the following technical scheme:
in a first aspect of the invention, there is provided a drilling fluid design method, the method comprising:
(1) establishing a case library;
(2) acquiring each characteristic attribute of the new problem;
(3) searching the most similar example in the example library as a similar example according to each characteristic attribute of the new problem;
(4) judging whether the similar examples meet the requirements, if so, turning to the step (7), and if not, turning to the step (5);
(5) correcting and optimizing the similar examples to obtain a solution of a new problem, and forming the new problem and the solution of the new problem into a new example;
(6) inputting the new case into the case library, and then returning to the step (3):
(7) and outputting the design scheme of the drilling fluid.
The operation of the step (1) comprises the following steps:
inputting examples of known drilling fluid design schemes into an example library one by one;
example information for each example and a drilling fluid design corresponding to the example information are stored in the example library.
The example information includes: basic information, formation lithology, pore medium characteristics, formation fluid characteristics, well sections, three pressure profiles, drilling risk types, formation sensitivity, well bore enlargement rate and a drilling fluid system;
the drilling fluid design scheme comprises: the drilling fluid comprises a drilling fluid formula, drilling fluid performance parameters, drilling fluid maintenance and treatment key points, oil and gas reservoir protection requirements and a well body structure.
The operation of the step (2) comprises the following steps:
inputting example information of a new question;
and extracting each numerical characteristic parameter in the input example information of the new question as each characteristic attribute of the new question.
The operation of the step (3) comprises:
the following processing is respectively carried out on each example in the example library: according to the characteristic attributes of the new problem and the characteristic attributes of the examples in the example library, utilizing a twin neural network to obtain the similarity between the new problem and the examples in the example library;
and finding the example with the maximum similarity, wherein the example with the maximum similarity is the similar example.
Obtaining a solution of the new problem in the step (5), namely obtaining a new drilling fluid design scheme;
the operation of forming the new problem and the solution of the new problem into a new example in the step (5) comprises the following steps: and forming a new example by using the input example information of the new problem and the new drilling fluid design scheme obtained through correction and optimization.
The operation of inputting the new case into the case library in the step (6) comprises:
the new paradigm is entered into the paradigm library using a twin neural network.
The operation of the step (7) comprises the following steps:
the drilling fluid design of the similar example is output.
In a second aspect of the invention, there is provided a drilling fluid design system, the system comprising:
a paradigm repository configured to store paradigms;
a case base establishing unit: connected to the example library, configured to input examples of known drilling fluid designs into the example library one by one;
a new question input unit: configured to obtain respective feature attributes of the new question;
the retrieval unit is respectively connected with the example library and the new problem input unit and is configured to retrieve an example which is most similar to the new problem in the example library as a similar example according to each characteristic attribute of the new problem;
the judging unit is respectively connected with the searching unit, the correcting and optimizing unit and the output unit and is configured to judge whether the similar examples meet the requirements or not, if yes, the similar examples are sent to the output unit, and if not, the similar examples are sent to the correcting and optimizing unit;
the correcting and optimizing unit is connected with the judging unit and the new example input unit and is configured to correct and optimize the similar examples, obtain a solution of a new problem, form the new problem and the solution of the new problem into a new example and send the new example to the new example input unit;
the new example input unit is respectively connected with the example library and the correction and optimization unit and is configured to input the new example into the example library;
and the output unit is connected with the judging unit and is configured to output the drilling fluid design scheme of the similar example.
The example library stores example information of each example and drilling fluid design schemes corresponding to the example information.
The example information includes: basic information, formation lithology, pore medium characteristics, formation fluid characteristics, well sections, three pressure profiles, drilling risk types, formation sensitivity, well bore enlargement rate and a drilling fluid system;
the drilling fluid design scheme comprises: the drilling fluid comprises a drilling fluid formula, drilling fluid performance parameters, drilling fluid maintenance and treatment key points, oil and gas reservoir protection requirements and a well body structure.
The new question input unit inputs example information of a new question;
and extracting each numerical characteristic parameter in the input example information of the new question as each characteristic attribute of the new question.
The retrieval unit respectively performs the following processing on each example in the example library: according to the characteristic attributes of the new problem and the characteristic attributes of the examples in the example library, utilizing a twin neural network to obtain the similarity between the new problem and the examples in the example library;
and finding the example with the maximum similarity, wherein the example with the maximum similarity is the similar example.
The new example input unit inputs a new example into the example library using the twin neural network.
Compared with the prior art, the invention has the beneficial effects that: the method utilizes a similar solving mode to carry out feature extraction, similar example matching, correction and optimization on new problems to form a new example continuously-expanded example library, so that the collection and storage of drilling fluid design knowledge are relatively simple, the application and maintenance of the knowledge in the example library are easier, the scientificity and the rapidness of drilling fluid design based on historical data are effectively improved, the complex risk of underground engineering is effectively reduced, and the reservoir pollution is avoided to the greatest extent.
Drawings
FIG. 1 is a block diagram of the steps of the method of the present invention;
FIG. 2 is a similarity calculation schematic;
FIG. 3 example drilling fluid design input parameters for an embodiment of the present invention;
FIG. 4 example drilling fluid design input parameters for an embodiment of the present invention;
FIG. 5 shows the output of the drilling fluid design case in an embodiment of the present invention;
FIG. 6 shows the output of the drilling fluid design case in the embodiment of the present invention.
Detailed Description
The invention is described in further detail below with reference to the accompanying drawings:
the drilling fluid design process relies on expert experience as part of the drilling engineering design. The artificial intelligence technology has unique advantages in solving the reasoning problem depending on experience thinking, and the expert system based on rule reasoning is successfully researched and applied in the early stage, but the expert system has some defects, for example, a great amount of expert knowledge needs to be collected to fill a rule base of the expert system, which is very time-consuming and difficult to realize, so the problem limits the application field of the expert system based on rule reasoning.
The design of the drilling fluid is an important component of the drilling engineering design, and the design quality is directly related to the stability and quality of a well bore in the drilling process and the realization of the production capacity construction target of an oil-gas well. For years, the drilling digital construction lays a foundation for drilling fluid design based on historical data, but the traditional retrieval or reasoning method is difficult to realize drilling fluid scheme decision design based on a drilling fluid case base due to the difficulty in high-dimensional information retrieval and matching. The invention provides an example reasoning method based on a twin neural network, which mainly comprises a construction form of an example library, a drilling fluid example matching architecture based on the twin neural network and an example reasoning algorithm (the construction form of the example library refers to the description content in the following table 1, the drilling fluid example matching architecture based on the twin neural network refers to the architecture shown in fig. 2, and the example reasoning algorithm refers to the process shown in fig. 1), can obviously improve the drilling fluid matching accuracy based on the example library, is suitable for intelligent recommendation under the condition that the example library or input parameters are incomplete (the same as the complete example library or the input parameters but larger errors exist in similarity matching due to the incomplete example library or the incomplete input parameters), and is favorable for improving the scientificity and the convenience of drilling fluid design.
The example reasoning technology based on the twin neural network is used as a quick method for solving similar problems, and the core idea is to use historical examples and practical experience to solve the new problems facing now. The basic idea is as follows: firstly, extracting characteristic information capable of characterizing the problem according to the description of the problem, then comparing the characteristic information with examples in an example library to retrieve similar examples, and finally training, integrating or modifying the retrieved similar examples according to the specific requirements of the new problem to obtain a scheme for solving the current problem. If the content of the example library is rich enough, the user has an opportunity to find the same example as the current problem to be solved and directly obtain a solution, however, in most cases, the search result is partially similar to the current problem to be solved, so that the similar example needs to be changed according to the actual situation and special requirements of the problem to be solved, and the process is called the example modification and optimization. The modified result of the similar example is a full solution to the new problem, and the new problem and the full solution are combined into a new example, and the machine learning is completed by screening and determining whether to store the new problem and the full solution in the source example library. The learning process enables the machine to continuously enhance the capability of processing problems and the solving process becomes easier.
The invention relates to a drilling fluid design method based on a twin neural network paradigm inference technology, and the core idea is to solve the problem of new drilling fluid design by using a once paradigm and practical experience. The example library is a storage unit of a historical drilling fluid design scheme and is a data base for solving the drilling fluid design scheme. From fig. 1, it can be seen that the drilling fluid design flow based on example reasoning is: automatically extracting characteristic attributes from design well basic data input by a user, starting an example retrieval mechanism, starting to search a drilling fluid example library for a drilling fluid design scheme similar to the design well basic data, outputting the similar example as a design result if the searched similar example can meet the design requirement of the design well, otherwise, correcting and optimizing the similar example according to the specific construction requirement of the design well, and finishing the correction of the similar example by utilizing an integrated learning model (the integrated learning is an existing learning mode and is obtained by weighted superposition of a neural network model, a rule inference model and a support vector machine and is not repeated here) based on a neural network, a rule inference system and a support vector machine in a correction and optimization part so that the design result meets the construction requirement of the design well. And finally, according to a certain screening principle, whether the new example formed by the output drilling fluid design scheme is added into the source example library or not is selected to achieve the purpose of system learning.
In the twin neural network-based example reasoning technology, the complicated knowledge engineering is simplified to be represented in the form of the example base, so that the collection and storage of the drilling fluid design knowledge are relatively simple, and the application and maintenance of the knowledge in the example base are easier.
(1) Exemplary content
The inclusion of a drilling fluid design paradigm is limited, so the present invention takes the form of a representation of the characteristics of the drilling fluid design, with the paradigm being represented by a series of characteristics that make up it. When a user inputs a drilling fluid design problem into the system, the system extracts corresponding characteristic attributes (namely, the information in table 1) from the problem description, and then compares the characteristic attributes with the data in the example library to find out similar examples.
According to the design requirements of the drilling fluid, the drilling fluid mainly comprises the following components: example names, oil fields, blocks, well types, well completion depths, formation types, lithology descriptions, formation clay mineral information, formation shale physicochemical information, downhole complex conditions and treatment measures, potential accident types and response methods, drilling fluid system names, drilling fluid formulations, drilling fluid performance parameters, drilling fluid maintenance treatment points, hydrocarbon reservoir protection requirements, well body structures and other information form a drilling fluid design example.
(2) Example presentation Structure
A complete drilling fluid example of the present invention contains example information and drilling fluid design. The example information includes: oil field name, block name, well type, surface type, formation (group/section), formation lithology 1, formation lithology 2, formation lithology 3, formation lithology 4, other lithology, permeability, pore space, fracture, karst cave, crude oil type zone water type, natural gas type, CO type, oil well type, oil well type, well type2、H2S, opening times, top depth, bottom depth, whether a reservoir is reserved, pore pressure coefficient, collapse pressure coefficient, fracture pressure coefficient, well leakage, well collapse, stuck drilling, blowout, stratum sensitivity, well diameter expansion rate, drilling fluid system and the like. The drilling fluid design scheme comprises information on aspects such as a drilling fluid formula, drilling fluid performance parameters, drilling fluid maintenance and treatment key points, oil-gas reservoir protection requirements, a well body structure and the like. In the drilling fluid design process, aiming atWith "example information" of a certain characteristic, the system goes through a series of similarity analyses to solve for a drilling fluid design for a new well.
(3) Example framework representation Structure example
The drilling fluid design scheme of a certain well is subjected to feature extraction processing according to a frame representation method, the structural hierarchy of the drilling fluid design scheme is shown in table 1, grooves 1 to 8 in table 1 represent 8 text boxes capable of inputting example information, information of the groove is input under each groove, for example, the groove 1 is basic information, and information needing to be input comprises 'oil field name', 'block name', 'well type', 'surface type', 'stratum (group/segment)'. When the invention is used, users need to fill the information in each groove one by one. The "feature attribute extraction" in fig. 1 refers to obtaining each piece of information in table 1.
TABLE 1
Essentially, the twin neural network based paradigm inference technique is based on using past experience to obtain a reasoning model of the current problem solution. The first step of solving the problem by using the example inference method based on the twin neural network is to describe the problem to be solved according to the example representation form in the example library, so that the similar example can be found in the example library. The problem description information that needs to be input by the user in the present invention is the aforementioned "exemplary information" (referring to all information of slot 1 to slot 8 in table 1). The essence of the example retrieval is to compare the similarity of the current problem to be solved with the past problems, and after the example similarities in the example library are sorted, find out one or more historical problems with higher similarity to the current problem, i.e. find out one or more drilling fluid design schemes with higher similarity.
In the research of the example search technology, the similarity between the components of two different examples needs to be researched, that is, the similarity between the characteristics of the examples represented by the frame structure is essentially to find the similarity calculation model between the characteristic attributes. The basic idea is to calculate the similarity between the characteristic attributes of the constituent examples, and then synthesize the similarity of the characteristic attributes to obtain the similarity between the two examples. It can therefore be seen that similarity computation between example feature attributes plays a very critical role in designing example-based reasoning.
As can be seen from table 1, there are values in each of the properties representing an example of a drilling fluid, so the overall similarity of the example is calculated according to the value classes.
Currently, examples of search strategies mainly include a neural network, a nearest neighbor method, a K-nearest neighbor method, a fuzzy matching method, a induction method, a knowledge guidance method, a template search method, and the like. In the example matching numerical algorithm of the present invention, the similarity algorithm selects a twin neural network (the twin neural network is introduced in the step of "search mechanism" in fig. 1, and the twin neural network is used to calculate the similarity between the new problem and the example, so as to obtain a plurality of similar examples with similarity arranged in order from large to small), the twin neural network is a framework of the existing neural network, and is not a specific network, like seq2seq, and RNN or CNN may be used in specific implementation. The basic idea is as follows: the numerical characteristic parameters extracted in case 1 and case 2 are used as initial data of example retrieval to respectively extract characteristic vectors through a network, and then a similarity loss function is carried out on the two characteristic vectors in the last layer for network training. And finally, the similarity between the numerical characteristic attribute in the new problem and each original example numerical characteristic attribute is obtained. The similarity calculation principle is shown in fig. 2.
Research finds that for the same initial condition data, if two groups of different parameter weights are adopted to carry out example similarity calculation, the obtained results are greatly different, and whether a group of reasonable parameter weights can be found is crucial to obtaining an accurate retrieval result. The method for selecting the parameter weight value by expert review and scoring is commonly adopted, and a twin neural network method is adopted in the invention.
The numerical characteristic attribute similarity calculation function in the retrieval adopts a formula:
in the formula (I), the compound is shown in the specification,
representing the distance, Sim, between vector X and vector Y after the iterative training of the twin networknum(X, Y) is the similarity between the two examples;
x — a vector of initial condition feature parameter values, (X1, X2, …, xn), which refers to a vector of initial features of the new question entered;
y-a vector of parameter values corresponding to the initial condition parameters in the source example, Y ═ (Y1, Y2, …, yn), a vector of initial features of existing examples in the example library;
xi-the value of the ith target condition characteristic parameter;
yiith Source instance corresponds to xiA value of a condition characteristic parameter;
Simnum(X, Y) -the similarity function (distance obtained by using the twin neural network) of the vector X and the vector Y in the iteration of the twin network (the loss function of the twin neural network reaches the minimum value in the iteration process)
Then, the similarity Sim is obtained by the calculation of the formulanum(X,Y))；
wiThe weight of the ith parameter, the value of W is continuously updated during the twin network training (W represents the weight, b represents the deviation, W and b will change during the neural network training process due to the iterative search of the optimal solution, W is different for each layer of the network)
X1, x2 in fig. 2 represent one parameter in correspondence X, Y;
a rich drilling fluid design scheme example library is the key to the success of the system design. Therefore, when a drilling fluid design for a new well is completed, the design scheme should be added to the example library in time to enrich the system. However, if the new design coincides with or closely resembles an existing example in the example library, the design does not have to be added to the example library. When the similar examples searched out can only meet partial requirements in the actual design, namely, the example library does not have the drilling fluid design scheme directly identical to the design well, the examples need to be integrated or modified, and the formed new design scheme is added into the example library.
The examples of the invention are as follows:
[ EXAMPLES one ]
As shown in fig. 1, the method of the present invention comprises:
(1) creating a paradigm library using a paradigm of known drilling fluid design scenarios
Examples of known drilling fluid designs are first individually entered into the example library. Example information and drilling fluid design schemes corresponding to the example information are stored in the example library. The example information includes all information in table 1, such as example name, field name, block name, well type, completed well depth, formation type, lithology description, formation clay mineral information, formation shale physicochemical information, downhole complexity and treatment, type and response to potential accidents, drilling fluid system name, and the like. The drilling fluid design scheme comprises information on aspects such as a drilling fluid formula, drilling fluid performance parameters, drilling fluid maintenance and treatment key points, oil-gas reservoir protection requirements, a well body structure and the like.
(2) Acquiring each characteristic attribute of the new question: extracting the characteristic attributes of the new problems to obtain each characteristic attribute of the new problems;
first, a new problem (i.e., a problem to be solved) is described according to an example representation form in an example library, the example uses a frame structure shown in table 1 to represent and then constitutes each feature of the example, a numerical feature parameter in example information is extracted as each feature attribute of the new problem, and the numerical feature parameter in the example information in table 1 is each feature attribute of the new problem.
(3) Comparing each characteristic attribute of the new problem with each characteristic attribute of each example in the example library one by one, and searching the example which is most similar to the new problem in the example library as a similar example;
specifically, a twin neural network is adopted to obtain the similarity between the new problem and each example in the example library, and the example with the maximum similarity is found, namely the example is the similar example;
(4) judging whether the similar samples meet the requirements (judging whether the design requirements of the designed well are met according to expert experience), if so, turning to the step (7), and if not, turning to the step (5);
(5) correcting and optimizing the similar examples (according to the specific construction requirements of the designed well, correcting and optimizing the system selection and the formula composition of the drilling fluid in the similar examples, the specific method adopts the existing method, and the details are not repeated here), obtaining the solution of new problems, namely obtaining a new drilling fluid design scheme, and taking the example information of the new problems and the new drilling fluid design scheme as the new examples;
(6) inputting a new case into the case library, and then returning to the step (3):
in this embodiment, the twin neural network is used to input the new paradigm into the paradigm library, and the specific operations include: the new paradigm is input into the twin neural network in turn and in pairs with each paradigm in the paradigm library, and the loop iterates until the twin neural network converges or the number of loops is less than a specified value, which may be specified by the user, typically a larger integer. The twin neural network is used to input the new paradigm into the paradigm library in order to enable the new paradigm to correct itself according to the existing paradigm in the paradigm library, and further meet the requirement of building the library, that is, the new paradigm adjusts itself to make it belong to the same distribution space as the existing paradigm, that is, to make it the same type of data, which is the existing mathematical method and is not described herein again.
(7) Outputting a drilling fluid design scheme: the drilling fluid design of the similar example is output.
[ example two ]
The invention also provides a drilling fluid design system, comprising:
a paradigm repository configured to store paradigms; in this embodiment, the example library stores example information of each example and a drilling fluid design scheme corresponding to the example information. The example information includes: basic information, formation lithology, pore medium characteristics, formation fluid characteristics, well sections, three pressure profiles, drilling risk types, formation sensitivity, well bore enlargement rate and a drilling fluid system; the drilling fluid design scheme comprises: the drilling fluid comprises a drilling fluid formula, drilling fluid performance parameters, drilling fluid maintenance and treatment key points, oil and gas reservoir protection requirements and a well body structure.
A case base establishing unit: connected to the example library, configured to input examples of known drilling fluid designs into the example library one by one;
a new question input unit: configured to obtain respective feature attributes of the new question; in this embodiment, the new question input unit inputs example information of a new question; and extracting each numerical characteristic parameter in the input example information of the new question as each characteristic attribute of the new question.
The retrieval unit is respectively connected with the example library and the new problem input unit and is configured to retrieve an example which is most similar to the new problem in the example library as a similar example according to each characteristic attribute of the new problem; in this embodiment, the search unit performs the following processing on each instance in the instance library respectively: according to the characteristic attributes of the new problem and the characteristic attributes of the examples in the example library, utilizing a twin neural network to obtain the similarity between the new problem and the examples in the example library;
and finding the example with the maximum similarity, wherein the example with the maximum similarity is the similar example.
The judging unit is respectively connected with the searching unit, the correcting and optimizing unit and the output unit and is configured to judge whether the similar examples meet the requirements or not, if yes, the similar examples are sent to the output unit, and if not, the similar examples are sent to the correcting and optimizing unit;
the correcting and optimizing unit is connected with the judging unit and the new example input unit and is configured to correct and optimize the similar examples, obtain a solution of a new problem, form the new problem and the solution of the new problem into a new example and send the new example to the new example input unit;
the new example input unit is respectively connected with the example library and the correction and optimization unit and is configured to input the new example into the example library; in this embodiment, the new example input unit inputs the new example into the example library using the twin neural network.
And the output unit is connected with the judging unit and is configured to output the drilling fluid design scheme of the similar example.
In order to verify the effect of the present invention, the framework hierarchy of the drilling fluid example is shown in table 1 in more than 30 oil fields, more than 800 wells, more than 180 drilling fluid systems, more than 1800 formulations, and 3857 drilling fluid design cases using the present invention.
Firstly, inputting real drilling fluid related parameters such as well type, well depth, underground complex conditions of adjacent wells and the like, and respectively inputting parameters of two wells in fig. 3 and 4. Take two wells in a certain oil field as an example.
And clicking the next step after the input is finished, and performing case matching. Similar cases are then calculated, as shown in fig. 5 and 6, where fig. 5 and 6 are the output of two wells, respectively, with similar cases ranging from high to low for each initial instance.
By adopting a cross validation mode, the average accuracy of the obtained drilling fluid matching reaches 81.2%. Compared with the traditional Euclidean distance and the weighted Euclidean distance, the accuracy rate of the method is respectively improved by 13.9% and 3.1%.
In the twin neural network-based example reasoning technology, the complicated knowledge engineering is simplified to be represented by the formation of the example base, so that the collection and storage of the drilling fluid design knowledge are relatively simple, and the application and maintenance of the knowledge in the example base are easier. The achievement can be popularized and applied in the petroleum industry, the problem of drilling fluid design reasoning can be effectively solved, a basis can be provided for the optimization of drilling fluid formula design and the reasonable formulation of implementation cases, and the achievement has a very wide market prospect.
The above-described embodiment is only one embodiment of the present invention, and it will be apparent to those skilled in the art that various modifications and variations can be easily made based on the application and principle of the present invention disclosed in the present application, and the present invention is not limited to the method described in the above-described embodiment of the present invention, so that the above-described embodiment is only preferred, and not restrictive.
Claims (14)
1. A drilling fluid design method is characterized in that: the method comprises the following steps:
(1) establishing a case library;
(2) acquiring each characteristic attribute of the new problem;
(3) searching the most similar example in the example library as a similar example according to each characteristic attribute of the new problem;
(4) judging whether the similar examples meet the requirements, if so, turning to the step (7), and if not, turning to the step (5);
(5) correcting and optimizing the similar examples to obtain a solution of a new problem, and forming the new problem and the solution of the new problem into a new example;
(6) inputting the new case into the case library, and then returning to the step (3):
(7) and outputting the design scheme of the drilling fluid.
2. The drilling fluid design method of claim 1, wherein: the operation of the step (1) comprises the following steps:
inputting examples of known drilling fluid design schemes into an example library one by one;
example information for each example and a drilling fluid design corresponding to the example information are stored in the example library.
3. The drilling fluid design method of claim 2, wherein: the example information includes: basic information, formation lithology, pore medium characteristics, formation fluid characteristics, well sections, three pressure profiles, drilling risk types, formation sensitivity, well bore enlargement rate and a drilling fluid system;
the drilling fluid design scheme comprises: the drilling fluid comprises a drilling fluid formula, drilling fluid performance parameters, drilling fluid maintenance and treatment key points, oil and gas reservoir protection requirements and a well body structure.
4. The drilling fluid design method of claim 3, wherein: the operation of the step (2) comprises the following steps:
inputting example information of a new question;
and extracting each numerical characteristic parameter in the input example information of the new question as each characteristic attribute of the new question.
5. The drilling fluid design method of claim 4, wherein: the operation of the step (3) comprises:
the following processing is respectively carried out on each example in the example library: according to the characteristic attributes of the new problem and the characteristic attributes of the examples in the example library, utilizing a twin neural network to obtain the similarity between the new problem and the examples in the example library;
and finding the example with the maximum similarity, wherein the example with the maximum similarity is the similar example.
6. The drilling fluid design method of claim 5, wherein: obtaining a solution of the new problem in the step (5), namely obtaining a new drilling fluid design scheme;
the operation of forming the new problem and the solution of the new problem into a new example in the step (5) comprises the following steps: and forming a new example by using the input example information of the new problem and the new drilling fluid design scheme obtained through correction and optimization.
7. The drilling fluid design method of claim 6, wherein: the operation of inputting the new case into the case library in the step (6) comprises:
the new paradigm is entered into the paradigm library using a twin neural network.
8. The drilling fluid design method of claim 5, wherein: the operation of the step (7) comprises the following steps:
the drilling fluid design of the similar example is output.
9. A drilling fluid design system, characterized by: the system comprises:
a paradigm repository configured to store paradigms;
a case base establishing unit: connected to the example library, configured to input examples of known drilling fluid designs into the example library one by one;
a new question input unit: configured to obtain respective feature attributes of the new question;
the retrieval unit is respectively connected with the example library and the new problem input unit and is configured to retrieve an example which is most similar to the new problem in the example library as a similar example according to each characteristic attribute of the new problem;
the judging unit is respectively connected with the searching unit, the correcting and optimizing unit and the output unit and is configured to judge whether the similar examples meet the requirements or not, if yes, the similar examples are sent to the output unit, and if not, the similar examples are sent to the correcting and optimizing unit;
the correcting and optimizing unit is connected with the judging unit and the new example input unit and is configured to correct and optimize the similar examples, obtain a solution of a new problem, form the new problem and the solution of the new problem into a new example and send the new example to the new example input unit;
the new example input unit is respectively connected with the example library and the correction and optimization unit and is configured to input the new example into the example library;
and the output unit is connected with the judging unit and is configured to output the drilling fluid design scheme of the similar example.
10. The drilling fluid design system of claim 9, wherein: the example library stores example information of each example and drilling fluid design schemes corresponding to the example information.
11. The drilling fluid design system of claim 10, wherein: the example information includes: basic information, formation lithology, pore medium characteristics, formation fluid characteristics, well sections, three pressure profiles, drilling risk types, formation sensitivity, well bore enlargement rate and a drilling fluid system;
the drilling fluid design scheme comprises: the drilling fluid comprises a drilling fluid formula, drilling fluid performance parameters, drilling fluid maintenance and treatment key points, oil and gas reservoir protection requirements and a well body structure.
12. The drilling fluid design system of claim 11, wherein: the new question input unit inputs example information of a new question;
and extracting each numerical characteristic parameter in the input example information of the new question as each characteristic attribute of the new question.
13. The drilling fluid design system of claim 12, wherein: the retrieval unit respectively performs the following processing on each example in the example library: according to the characteristic attributes of the new problem and the characteristic attributes of the examples in the example library, utilizing a twin neural network to obtain the similarity between the new problem and the examples in the example library;
and finding the example with the maximum similarity, wherein the example with the maximum similarity is the similar example.
14. The drilling fluid design system of claim 12, wherein: the new example input unit inputs a new example into the example library using the twin neural network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010503095.4A CN113761613A (en) | 2020-06-05 | 2020-06-05 | Drilling fluid design method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010503095.4A CN113761613A (en) | 2020-06-05 | 2020-06-05 | Drilling fluid design method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113761613A true CN113761613A (en) | 2021-12-07 |
Family
ID=78783875
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010503095.4A Pending CN113761613A (en) | 2020-06-05 | 2020-06-05 | Drilling fluid design method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113761613A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116682499A (en) * | 2023-04-23 | 2023-09-01 | 常州大学 | Auxiliary method for drilling fluid formula design based on stratum information and well information |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1760854A (en) * | 2005-09-09 | 2006-04-19 | 冶金自动化研究设计院 | Intelligent designer for prescription of drilling fluid |
| US20120094876A1 (en) * | 2010-10-19 | 2012-04-19 | Dale Jamison | Designed drilling fluids for ecd management and exceptional fluid performance |
| CN104794528A (en) * | 2015-01-05 | 2015-07-22 | 胡宝清 | Rocky desertification governance model case library management system and rocky desertification governance model case library management method |
| CN110032635A (en) * | 2019-04-22 | 2019-07-19 | 齐鲁工业大学 | One kind being based on the problem of depth characteristic fused neural network to matching process and device |
| CN110046240A (en) * | 2019-04-16 | 2019-07-23 | 浙江爱闻格环保科技有限公司 | In conjunction with the target domain question and answer method for pushing of keyword retrieval and twin neural network |
-
2020
- 2020-06-05 CN CN202010503095.4A patent/CN113761613A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1760854A (en) * | 2005-09-09 | 2006-04-19 | 冶金自动化研究设计院 | Intelligent designer for prescription of drilling fluid |
| US20120094876A1 (en) * | 2010-10-19 | 2012-04-19 | Dale Jamison | Designed drilling fluids for ecd management and exceptional fluid performance |
| CN104794528A (en) * | 2015-01-05 | 2015-07-22 | 胡宝清 | Rocky desertification governance model case library management system and rocky desertification governance model case library management method |
| CN110046240A (en) * | 2019-04-16 | 2019-07-23 | 浙江爱闻格环保科技有限公司 | In conjunction with the target domain question and answer method for pushing of keyword retrieval and twin neural network |
| CN110032635A (en) * | 2019-04-22 | 2019-07-19 | 齐鲁工业大学 | One kind being based on the problem of depth characteristic fused neural network to matching process and device |
Non-Patent Citations (1)
| Title |
|---|
| 胡茂焱: "钻井液设计系统的研究", 《中国博士学位论文全文数据库 基础科学辑》, pages 57 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116682499A (en) * | 2023-04-23 | 2023-09-01 | 常州大学 | Auxiliary method for drilling fluid formula design based on stratum information and well information |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11746650B2 (en) | Method for calculating single-well controlled reserve of low-permeability/tight gas reservoir and analyzing residual gas thereof | |
| CN108388921B (en) | Overflow loss real-time identification method based on random forest | |
| Guyaguler et al. | Optimization of well placement in a Gulf of Mexico waterflooding project | |
| US6236894B1 (en) | Petroleum production optimization utilizing adaptive network and genetic algorithm techniques | |
| US20230266502A1 (en) | Collaborative optimization method for gas injection huff-n-puff parameters in tight oil reservoirs | |
| CN114117881A (en) | Sand production risk prediction method and system | |
| Park et al. | Improved decision making with new efficient workflows for well placement optimization | |
| CN106250984A (en) | Method and device for judging oil-water relation mode of oil well | |
| Lee et al. | Regeneration of channelized reservoirs using history-matched facies-probability map without inverse scheme | |
| CN113608265A (en) | Reservoir prediction method based on deep hybrid neural network | |
| Tang et al. | Use of low-fidelity models with machine-learning error correction for well placement optimization | |
| CN113761613A (en) | Drilling fluid design method and system | |
| CN118036477A (en) | Well position and well control parameter optimization method based on space-time diagram neural network | |
| Ariadji et al. | A novel tool for designing well placements by combination of modified genetic algorithm and artificial neural network | |
| CN117743978A (en) | Rock mass integrity identification method based on gradient lifting decision tree and while-drilling parameters | |
| CN111488384B (en) | Intelligent drilling scheme recommendation method and system | |
| CN113095352A (en) | Oil-gas-containing property evaluation method for fracture-cavity type oil reservoir semi-supervised learning reservoir fused with static and dynamic characteristics | |
| CN116796237A (en) | Oil-water well communication relation identification method based on improved graph neural network | |
| Ariadji et al. | Optimization of vertical well placement for oil field development based on basic reservoir rock properties using a genetic algorithm | |
| Kravis et al. | A case based system for oil and gas well design with risk assessment | |
| Huang et al. | A comprehensive machine learning approach for quantitatively analyzing development performance and optimization for a heterogeneous carbonate reservoir in Middle East | |
| Naufal et al. | A digital oilfield comprehensive study: Automated intelligent production network optimization | |
| Li et al. | Flow field characterization and evaluation method based on unsupervised machine learning | |
| Saradva et al. | Exploring the Potential of Machine Learning Technology to Generate Productivity Maps Using Offset Field Data for a Strategic Coiled Tubing Drilling Program-Onshore Sharjah, UAE | |
| Temizel et al. | Production Forecasting in Conventional Oil Reservoirs Using Deep Learning |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20211207 |