CN113962148A - Yield prediction method, device and equipment based on convolutional coding dynamic sequence network - Google Patents
Yield prediction method, device and equipment based on convolutional coding dynamic sequence network Download PDFInfo
- Publication number
- CN113962148A CN113962148A CN202111219923.2A CN202111219923A CN113962148A CN 113962148 A CN113962148 A CN 113962148A CN 202111219923 A CN202111219923 A CN 202111219923A CN 113962148 A CN113962148 A CN 113962148A
- Authority
- CN
- China
- Prior art keywords
- data
- horizontal well
- dynamic
- static
- target horizontal
- 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.)
- Granted
Links
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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/22—Yield analysis or yield optimisation
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Medical Informatics (AREA)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
- Compression Of Band Width Or Redundancy In Fax (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The embodiment of the specification provides a yield prediction method, a device and equipment based on a convolutional coding dynamic sequence network, and the method comprises the following steps: acquiring static data and dynamic data corresponding to each target horizontal well; the dynamic data comprise yield data corresponding to each time point in preset time; dividing the dynamic data of each target horizontal well into first dynamic data and second dynamic data based on the predicted time point to obtain a sample data set; wherein the predicted time point is within the preset time; and training the convolutional coding dynamic sequence neural network by using the sample data set to obtain a target horizontal well yield prediction model. By the aid of the method and the device, the production prediction accuracy of the fractured horizontal well can be improved.
Description
Technical Field
The application relates to the technical field of shale gas exploration and development, in particular to a yield prediction method, device and equipment based on a convolutional coding dynamic sequence network.
Background
Shale gas development gradually becomes a new hotspot of world energy development in recent years, and shale gas mainly exists in shale rich in organic matters and interlayers to exist in adsorbed gas and free gas, so that the shale gas has great guiding significance for accurate yield prediction of fractured horizontal wells and shale gas exploration and development.
In the prior art, because the long-short term memory neural network has a unique structure and can 'remember' the previous information, the long-short term memory neural network is generally adopted for predicting the time sequence yield. However, since the long-term and short-term memory neural network cannot consider some parameters which have a large influence on the yield and do not change with time, the yield prediction result obtained by using the long-term and short-term memory neural network has a large deviation from the actual situation.
Therefore, there is a need in the art for a solution to the above problems.
Disclosure of Invention
The embodiment of the specification provides a yield prediction method, a yield prediction device and yield prediction equipment based on a convolutional coding dynamic sequence network, which can more fully utilize the existing data, realize the combination of static data and dynamic data and improve the yield prediction accuracy of a fractured horizontal well.
The method, the device and the equipment for predicting the yield based on the convolutional coding dynamic sequence network are realized in the following mode.
A yield prediction method based on a convolutional coding dynamic sequence network comprises the following steps: acquiring static data and dynamic data corresponding to each target horizontal well; the dynamic data comprise yield data corresponding to each time point in preset time; dividing the dynamic data of each target horizontal well into first dynamic data and second dynamic data based on the predicted time point to obtain a sample data set; wherein the predicted time point is within the preset time; and training a convolutional coding dynamic sequence neural network by using the sample data set to obtain a target horizontal well yield prediction model.
A yield prediction apparatus based on a convolutional encoded dynamic sequence network, comprising: the acquisition module is used for acquiring static data and dynamic data corresponding to each target horizontal well; the dynamic data comprise yield data corresponding to each time point in preset time; the dividing module is used for dividing the dynamic data of each target horizontal well into first dynamic data and second dynamic data based on the predicted time point to obtain a sample data set; wherein the predicted time point is within the preset time; and the training module is used for training the convolutional coding dynamic sequence neural network by using the sample data set to obtain a target horizontal well yield prediction model.
A yield prediction apparatus based on a convolutional encoded dynamic sequence network, comprising a processor and a memory for storing executable instructions, which when executed by the processor, implement the steps of any one of the method embodiments of the present specification.
A computer readable storage medium having stored thereon computer instructions which, when executed, implement the steps of any one of the method embodiments in the present specification.
The yield prediction method, device and equipment based on the convolutional coding dynamic sequence network can acquire static data and dynamic data corresponding to each target horizontal well, wherein the dynamic data comprise yield data corresponding to each time point within preset time, the dynamic data of each target horizontal well are divided into first dynamic data and second dynamic data based on prediction time points, and a sample data set is obtained; and training a convolutional coding dynamic sequence neural network by using the sample data set within the preset time to obtain a target horizontal well yield prediction model. Because the static parameters of the fractured horizontal well are considered when the convolutional coding dynamic sequence neural network is trained, the existing data can be utilized more fully, the combination of the static data and the dynamic data is realized, the yield prediction accuracy of the fractured horizontal well is improved, and the yield sequence can be predicted for a long time by the yield sequence, so that the method is more suitable for practical application and better guides the production and development of oil fields.
Drawings
The accompanying drawings, which are included to provide a further understanding of the specification, are incorporated in and constitute a part of this specification, and are not intended to limit the specification. In the drawings:
fig. 1 is a schematic flowchart of a yield prediction method based on a convolutional coding dynamic sequence network according to an embodiment of the present disclosure;
FIG. 2 is a schematic diagram of the predicted results of a horizontal well yield prediction model and a long-short term memory neural network model provided in the embodiments of the present disclosure on the 100 th and 450 th samples in a test set;
FIG. 3 is a histogram of cumulative distribution of relative errors predicted for 750 samples by a horizontal well production prediction model and a long-short term memory neural network model provided in an embodiment of the present disclosure;
fig. 4 is a schematic block diagram of a yield prediction apparatus based on a convolutional coding dynamic sequence network according to an embodiment of the present disclosure;
fig. 5 is a block diagram of a hardware structure of a yield prediction server based on a convolutional coding dynamic sequence network according to an embodiment of the present disclosure.
Detailed Description
In order to make those skilled in the art better understand the technical solutions in the present specification, the technical solutions in the embodiments of the present specification will be clearly and completely described below with reference to the drawings in the embodiments of the present specification, and it is obvious that the described embodiments are only a part of the embodiments in the present specification, and not all of the embodiments. All other embodiments that can be obtained by a person skilled in the art on the basis of one or more embodiments of the present description without inventive step shall fall within the scope of protection of the embodiments of the present description.
The following describes an embodiment of the present disclosure with a specific application scenario as an example. Specifically, fig. 1 is a schematic flowchart of a yield prediction method based on a convolutional coding dynamic sequence network according to an embodiment of the present disclosure. Although the present specification provides the method steps or apparatus structures as shown in the following examples or figures, more or less steps or modules may be included in the method or apparatus structures based on conventional or non-inventive efforts.
One embodiment provided by the present specification can be applied to a client, a server, and the like. The client may include a terminal device, such as a smart phone, a tablet computer, and the like. The server may include a single computer device, or may include a server cluster formed by a plurality of servers, or a server structure of a distributed system, and the like.
It should be noted that the following description of the embodiments does not limit the technical solutions in other extensible application scenarios based on the present specification. In an embodiment of the method for predicting the yield based on the convolutional code dynamic sequence network, as shown in fig. 1, the method may include the following steps.
S0: acquiring static data and dynamic data corresponding to each target horizontal well; and the dynamic data comprise yield data corresponding to each time point in preset time.
The target horizontal well may include one or more wells. The target horizontal well can be a fractured horizontal well and can also be other wells for producing oil and gas, such as shale gas horizontal wells and the like. The static data may be data corresponding to the static parameter, and the static data is a fixed value and does not change with time. The dynamic data may be data corresponding to dynamic parameters, and the dynamic data may change with time.
In some embodiments, the static parameters may include geological parameters, fracturing parameters, and the like. In some embodiments, the static parameters may include: reservoir thickness, initial formation pressure, matrix permeability, matrix porosity, fracture modification volumetric region permeability, fracture modification volumetric region porosity, fracture half-length, fracture permeability, fracture number, fracture conductivity, well length, and the like. The dynamic parameters may include production data, etc., such as annual average gas production, annual average water production, etc.
In some embodiments, the dynamic data may include production data corresponding to each point in time within a predetermined time. The preset time may be preset according to an actual scene, for example, 10 years, 17 years, and the like. Each time point may also be set according to an actual scene, for example, one year may correspond to one time point, 2 years may correspond to one time point, half a year corresponds to one time point, and the like. The interval corresponding to the preset time may be set according to an actual scenario, for example, if the target horizontal well starts to produce from 2011, 10 years between 2011 and 2020 may be used as the preset time, and if the target horizontal well starts to produce from 2000, 17 years between 2005 and 2021 may be used as the preset time. In the embodiment of the present specification, an example is given by taking a year as an example, and other scenarios are similar and may be referred to each other.
In some embodiments, the obtaining the static data and the dynamic data corresponding to each target horizontal well may include: determining the value range of each static parameter of the target horizontal well; based on the value range of each static parameter, generating static data corresponding to each target horizontal well by using a Latin hypercube sampling method; and respectively inputting the static data of each target horizontal well into shale gas reservoir numerical simulation software to obtain the dynamic data corresponding to each target horizontal well.
In some embodiments, although the values of the static parameters of different target horizontal wells may differ, each static parameter corresponds to a reasonable value range due to the limitation of actual geological conditions, and therefore, the value range of each static parameter of a target horizontal well can be determined.
In some embodiments, the determining the value range of each static parameter of the target horizontal well may include: the method comprises the steps of determining from information recorded in a preset database, or determining through receiving value ranges of various static parameters input by a user, and the like. It is understood that, the value range of each static parameter may also be determined in other possible manners, for example, the value range of each static parameter is searched in a web page according to a certain search condition, which may be determined specifically according to an actual situation, and this is not limited in this embodiment of the present specification.
In some embodiments, a Latin Hypercube Sampling (LHS) is a method of approximate random Sampling from multivariate parameter distributions, belonging to hierarchical Sampling techniques. In statistical sampling, a latin square refers to a square containing only one sample per row and column. The Latin hypercube is the popularization of a Latin square matrix in multiple dimensions, and each hyperplane vertical to an axis contains at most one sample. Assuming that there are N variables (dimensions), each variable can be divided into M intervals with the same probability. At this time, M sample points satisfying the latin hypercube condition may be selected. It should be noted that Latin hypercube sampling requires the same number of partitions M per variable. Suppose we want to extract m samples in an n-dimensional vector space. The Latin hypercube sampling step is as follows: (1) dividing each dimension into m intervals which do not overlap each other, so that each interval has the same probability (usually considering a uniform distribution, so that the intervals have the same length); (2) randomly extracting a point in each interval in each dimension; (3) randomly extracting points selected in the step (2) from each dimension, and forming vectors by the points to form sample data; wherein M, N, M and N are positive integers.
In some embodiments, after the value range of each static parameter of the target horizontal well is determined, the static data of each target horizontal well can be generated by using a latin hypercube sampling method based on the value range of each static parameter, and then the static data of each target horizontal well is respectively input into shale gas reservoir numerical simulation software to obtain the dynamic data corresponding to each target horizontal well. The shale gas reservoir numerical simulation software can be used for carrying out numerical simulation to obtain simulation data. For example, static data corresponding to 5000 wells may be input into the numerical simulation software unconng for numerical simulation, and corresponding dynamic data, such as production data, may be obtained. unconng is an oil reservoir numerical simulator, can adopt an embedded discrete fracture method to simulate a fracturing fracture and a natural fracture, and can be used for work such as production history fitting, production prediction, well completion mode and production system optimization.
In some embodiments, the dynamic data corresponding to each target horizontal well output by the shale gas reservoir numerical simulation software may be output in a form of a table, or may be an image or a file, which may be determined specifically according to an actual situation, and the application does not limit this.
In some embodiments, in the case that the dynamic data corresponding to each target horizontal well is output in the form of a table, the table may include a plurality of time points and production data corresponding to each time point.
In some embodiments, static data and dynamic data corresponding to multiple target horizontal wells may also be obtained by collecting field data.
In some embodiments, after obtaining the static data and the dynamic data of each target horizontal well, the static data and the dynamic data may be further preprocessed. Wherein the pre-treatment may comprise at least one of: deleting abnormal values, filling missing values, normalizing and the like.
In some embodiments, static data and dynamic data of each target horizontal well obtained through shale gas reservoir numerical simulation software are relatively complete and have no missing values, so that only normalization can be performed on preprocessing, that is, the static data and the dynamic data can be respectively normalized, so that the influences of dimension and data magnitude can be removed, and a neural network can be better trained.
In some embodiments, the normalization formula that may be employed is as follows:
wherein x ismaxMaximum value, x, representing a certain class of dataminRepresenting the minimum value, x, of a certain class of dataoldRepresents any number, x, of this class of datanewRepresents a pair xoldNormalized data, xnewIs taken as value of [0, 1]。
S2: dividing the dynamic data of each target horizontal well into first dynamic data and second dynamic data based on the predicted time point to obtain a sample data set; wherein the predicted time point is within the preset time.
In some embodiments, after the static data and the dynamic data corresponding to each target horizontal well are obtained, the dynamic data of each target horizontal well may be divided into first dynamic data and second dynamic data based on a predicted time point, so as to obtain a sample data set. Wherein the predicted time point is within the preset time. The first dynamic data and the second dynamic data respectively comprise yield data corresponding to different time points. The predicted time point may be set according to an actual scene, which is not limited in this specification. The sample data set may include multiple sets of sample data, each set of sample data including static data, first dynamic data, and second dynamic data.
For example, in some implementation scenarios, the acquired dynamic data of each target horizontal well includes an annual average gas production rate corresponding to each year in 17 years, and the prediction time point is the 9 th year, the dynamic data of each target horizontal well in 17 years can be divided into two parts based on the prediction time point: the first part is the annual average gas production from year 1 to year 9 and the second part is the annual average gas production from year 10 to year 17.
In some embodiments, the first dynamic data is a pre-fraction of the existing production for each target horizontal well. For example, if one well produces gas (oil) for 10 years, the annual average gas (oil) production of the 10 years can be obtained, and if the annual average gas (oil) production of each year of the next 5 years is desired to be known, then for the dynamic data of the existing multiple wells, the annual average gas (oil) production of the previous 10 years can be used as the first dynamic data (also referred to as the previous sequence), and the annual average gas production of the next 5 years can be used as the second dynamic data (also referred to as the next sequence), so that the previous sequence can predict the next sequence. It should be noted that the dynamic data may be divided in a manner similar to the division of the dynamic data including only the production data, and may be referred to each other. For example, the dynamic data may further include water production, and in this case, the annual average gas production of the next 5 years may be predicted by using the previous 10 annual average gas production + annual average water production, except that the dimension of the input data is changed from 1 to 2, and the dimension of the output data is unchanged.
In some embodiments, after the sample data set is obtained, the data in the sample data set may be stored into a data table. Where each row of the data table may represent a set of sample data.
S4: and training a convolutional coding dynamic sequence neural network by using the sample data set to obtain a target horizontal well yield prediction model.
In some embodiments, after obtaining the sample data set, the convolutional coding dynamic sequence neural network may be trained using the sample data set to obtain a target horizontal well production prediction model. The target horizontal well yield prediction model can be used for predicting the subsequent yield of the target horizontal well.
In some embodiments, the convolutional encoded dynamic sequence neural network may include a convolutional neural network portion, an information fusion layer, an encoder portion, and a decoder portion. The convolutional neural network part can be used for performing feature extraction on the static data of each target horizontal well to obtain the static feature data of each target horizontal well. The information fusion layer can be used for fusing the static characteristic data and the first dynamic data of each target horizontal well to obtain fusion data corresponding to each target horizontal well. The encoder part can be used for encoding fusion data corresponding to each target horizontal well to obtain an intermediate vector. The decoder portion may be configured to decode the intermediate vector to obtain an output result. In some implementation scenarios, the encoder portion and the decoder portion may be long-short term memory neural networks.
In some embodiments, training a convolutional coding dynamic sequence neural network by using the sample data set to obtain a target horizontal well yield prediction model may include: dividing the sample data set into a training set and a test set; taking the static data and the first dynamic data in the training set as input and the second dynamic data as output, and training the convolutional encoding dynamic sequence neural network to obtain a first horizontal well yield prediction model; inputting the static data and the first dynamic data in the test set into the first horizontal well yield prediction model to obtain a first prediction result; and under the condition that the first prediction result and the second dynamic data in the test set meet preset conditions, taking the first horizontal well yield prediction model as a target horizontal well yield prediction model.
In some embodiments, after the sample data set is obtained, the sample data set may be divided into a training set and a test set. The training set can be used for training a convolutional coding dynamic sequence neural network, and the test set can be used for testing a model obtained by training. In some implementation scenarios, the sample data set may be divided into a training set and a test set according to a preset ratio. The preset division ratio may be set according to an actual scene, and may be, for example, 7: 3. 6: 4, etc., which are not limited in this specification.
In some embodiments, after the sample data set is obtained, the sample data set may be further divided into a training set, a validation set, and a test set. For example, the ratio of 6: 2: 2, dividing the sample data set into a training set, a verification set and a test set. It should be noted that, in the embodiments of the present specification, the division into the training set and the test set is taken as an example for illustration, and other division manners are similar and may be referred to each other.
In some embodiments, after obtaining the training set and the test set, the convolutional coding dynamic sequence neural network may be trained by using the static data and the first dynamic data in the training set as inputs and the second dynamic data as outputs to obtain a first horizontal well yield prediction model.
In some embodiments, training the convolutional encoded dynamic sequence neural network with the static data and the first dynamic data in the training set as inputs and the second dynamic data as outputs may include: performing feature extraction on the static data of each target horizontal well based on the convolutional neural network part to obtain the static feature data of each target horizontal well; fusing the static characteristic data and the first dynamic data of each target horizontal well based on the information fusion layer to obtain fusion data corresponding to each target horizontal well; encoding fusion data corresponding to each target horizontal well based on the encoder part to obtain an intermediate vector; decoding the intermediate vector based on the decoder portion to obtain an output result. Where the intermediate vectors are the hidden and unit states of the encoder portion.
In some embodiments, before the static characteristic data and the first dynamic data of each target horizontal well are fused based on the information fusion layer, the dimensions of the static characteristic data and the dimensions of the first dynamic data may be matched, and in the case of dimension matching, information fusion is performed. In some implementation scenarios, the dimension of the extracted static feature data may be adjusted to match the dimension of the first dynamic data by changing the number of neurons in the network layer. The information fusion layer may also be referred to as a network layer. Of course, the above description is only exemplary, the way of matching the dimensions is not limited to the above examples, and other modifications are possible for those skilled in the art in light of the technical spirit of the present application, and all that can be achieved is intended to be covered by the scope of the present application as long as the achieved functions and effects are the same as or similar to the present application.
In some embodiments, after the dimensions of the static characteristic data and the first dynamic data are matched, the static characteristic data and the first dynamic data of each target horizontal well may be fused in the following manner:
g＝σ(Wcxc+Wlxl+b)
xcombined＝g·xc+(1-g)·xl
wherein g represents weight in information fusion, sigma represents sigoid activation function, and xcRepresenting static characteristic data, xlRepresenting first dynamic data, Wc、WlRespectively representing the weights of the network layers, b representing the offsets of the network layers, xcombinedRepresenting the fused data. Matrix Wc、WlIs [0, 1 ]]，Wc、WlAnd b, trainable parameters of the convolutional coding dynamic sequence neural network are parameters determined by the network after the convolutional coding dynamic sequence neural network is trained, the values can be randomly assigned during initialization, and the final values are determined by the network. The mathematical form of σ can be expressed as:
the value of the function value is (0, 1). g may be used to control the ratio of static characteristic data and first dynamic data. x is the number ofcombinedIs the data that is ultimately input to the convolutional encoded dynamic sequence neural network encoder portion.
In some embodiments, after the intermediate vector is decoded based on the decoder portion and the output result is obtained, the output result may be compared with the corresponding second dynamic data, and in a case that the output result and the second dynamic data do not satisfy the preset condition, the trainable parameters of the network may be adjusted according to the second dynamic data in the training set, so that the output result of the network approaches the corresponding second dynamic data. The preset condition may include, but is not limited to, that a relative error between the output result and the corresponding second dynamic data is smaller than a preset value.
In some embodiments, in the process of training the convolutional encoding dynamic sequence neural network by using the training set, a Teacher Forcing strategy can be adopted. The working principle of the Teacher Forcing strategy is as follows: at time t of the training process, the expected or actual outputs of the training set are used as inputs for the next time step, rather than the model-generated outputs. The second dynamic data in each set of sample data in the training set may be understood as an expected output or an actual output, i.e. a label, corresponding to the set of sample data. The output generated using the model may be understood to be the output result corresponding to the decoder portion of the convolutional encoded dynamic sequence neural network.
In some embodiments, after a convolutional encoding dynamic sequence neural network is trained by using a training set to obtain a first horizontal well yield prediction model, static data and first dynamic data in a test set can be input into the first horizontal well yield prediction model to obtain a first prediction result; and under the condition that the first prediction result and the second dynamic data in the test set meet preset conditions, taking the first horizontal well yield prediction model as a target horizontal well yield prediction model.
Because the test set is data never seen in the first horizontal well yield prediction model, if the first horizontal well yield prediction model can perform better prediction on the test set, the model can be shown to have better performance and strong generalization capability, so in some embodiments, static data and first dynamic data in the test set can be input into the first horizontal well yield prediction model for prediction to obtain a first prediction result. Further, whether the first prediction result and the second dynamic data corresponding to the test set meet preset conditions or not can be judged, and the first horizontal well yield prediction model can be used as a target horizontal well yield prediction model under the condition that the preset conditions are met. The preset condition may include that a relative error between the first prediction result and the corresponding second dynamic data in the test set is smaller than a preset value. The preset value may be set according to an actual scene, which is not limited in this specification. The first prediction result may be understood as a predicted value, and the corresponding second dynamic data in the test set may be understood as a true value. Of course, the preset conditions may also include others, which are not limited in this specification.
In some embodiments, the relative error of the first prediction result with the corresponding second dynamic data in the test set may be calculated by:
wherein r istIn order to be a relative error,
is the true value, ytIs a predicted value.
In some embodiments, the target horizontal well production prediction model may be understood as the final training outcome is the model. Because the horizontal well yield prediction model is obtained based on convolutional coding dynamic sequence neural network training, the horizontal well yield prediction model can also comprise four parts: the device comprises a convolutional neural network part, an information fusion layer, an encoder part and a decoder part.
The embodiment of the specification can be applied to oilfield production prediction, such as shale gas production prediction and the like.
In the embodiment of the specification, because the static parameters of the fractured horizontal well are considered when the convolutional coding dynamic sequence neural network is trained, the data can be more fully utilized, and the effect of the yield decrement prediction of the fractured horizontal well can be greatly improved.
It is to be understood that the foregoing is only exemplary, and the embodiments of the present disclosure are not limited to the above examples, and other modifications may be made by those skilled in the art within the spirit of the present disclosure, and the scope of the present disclosure is intended to be covered by the claims as long as the functions and effects achieved by the embodiments are the same as or similar to the present disclosure.
The method is described below with reference to specific embodiments, however, it should be noted that, for better explaining the present application, the following specific embodiments take the static data and the dynamic data of the fractured horizontal well as examples, and do not limit other application scenarios of the present application.
Because of the lack of actual data, in this embodiment, the numerical simulation software is used to generate static data and dynamic data corresponding to 5000 wells.
In this embodiment, before the static data and the dynamic data corresponding to 5000 wells are generated by using the numerical simulation software, a shale gas reservoir numerical simulation model may be established first. The established shale gas reservoir numerical simulation model comprises the following steps: the number of grids in the X direction is 150, the number of grids in the Y direction is 60, and the length of each unit grid in the X direction and the Y direction is 10 m; a horizontal well is arranged in the model, and the length of the horizontal well in the X horizontal direction is 1000 m. The model simulates a shale gas reservoir by considering Langmuir (Langmuir) adsorption resolution, where the limiting adsorption concentration (mass of gas adsorbed per mass of solid) is set at 0.0019362, the Langmuir pressure is set at 18.549bar, and the critical desorption pressure is set at 400 bar. The model sets constant pressure 300bar production and simulates the shale gas reservoir production process of 17 years.
In this embodiment, after the shale gas reservoir numerical simulation model is established, the value range of each static parameter of the shale gas reservoir fractured horizontal well can be determined. In this embodiment, the selected static parameters include: reservoir thickness, taking the value as [15m, 100m ]; initial formation pressure, taking the value of [350bar, 550bar ]; the matrix permeability is [0.000001md, 0.001md ]; the porosity of the matrix is [0.054, 0.08 ]; SRV (Stimulated Reservoir Volume) zone permeability, taking the value of [0.001md, 0.1md ]; SRV region porosity [0.08, 0.15 ]; the half-length of the crack is [60m, 280m ]; the crack permeability is [1000md, 5000md ]; the number of cracks was found to be [10, 30 ].
In this embodiment, after the value ranges of the static parameters of the shale gas reservoir fractured horizontal well are determined, the parameters may be sampled by a latin hypercube sampling method, 5000 samples, that is, static data corresponding to 5000 wells, are extracted, and then the 5000 wells are numerically simulated by numerical simulation software unconng to obtain dynamic data corresponding to 5000 wells, so that the static data and the dynamic data of the 5000 wells are obtained. It should be noted that, here, the dynamic data only considers the production data, and the production data is simulated 17-year data, where one year corresponds to one data point, that is, each well corresponds to 17 production data points, and each data point represents the annual average gas production rate.
Because the data generated by the numerical simulation software is relatively perfect and has no missing value, only the data needs to be normalized, and the influence of dimension and the magnitude of the data is removed.
Further, the production data for each well may be divided into two parts, the first part being the annual average gas production from year 1 to year 9 (which may be referred to as the first dynamic data or production series of the first part), and the second part being the annual average gas production from year 10 to year 17 (which may be referred to as the second dynamic data or production series of the second part), thereby obtaining a sample data set. The sample data set comprises 5000 groups of sample data, each group of sample data corresponds to one well, and each group of sample data comprises static data, first dynamic data and second dynamic data.
In this embodiment, after the sample data set is obtained, 4000 sets of sample data may be extracted from 5000 sets of sample data as a training set to train the convolutional coding dynamic sequence neural network, 250 sets of sample data may be extracted as a verification set to detect whether an overfitting phenomenon exists in the model in the training process, and 750 sets of sample data may be extracted as a test set to verify the accuracy of the model obtained by training.
Specifically, in the training process, the convolutional coding dynamic sequence neural network can be trained by taking the static data and the first dynamic data in the training set as input and the second dynamic data as output, so as to obtain a first horizontal well yield prediction model. The convolutional coding dynamic sequence neural network comprises a convolutional neural network part, an information fusion layer, an encoder part and a decoder part, so that in the training process by using a training set, static data can be firstly input into the convolutional neural network part for feature extraction to obtain static feature data, then the static feature data and the first dynamic data are input into the information fusion layer for fusion to obtain fusion data, further, the fusion data is input into the encoder part for encoding to obtain an intermediate variable, and finally, the second dynamic data in the training set is used as a label to decode the intermediate variable by using the decoder part to obtain an output result.
In this embodiment, after the first horizontal well yield prediction model is obtained, the static data and the first dynamic data in the test set may be input into the first horizontal well yield prediction model to obtain a first prediction result, and the first horizontal well yield prediction model is used as the target horizontal well yield prediction model when the first prediction result and the second dynamic data in the test set satisfy the preset condition.
Because the target horizontal well yield prediction model finally obtained by the method is obtained based on the static data and the yield sequence training of the first part, in order to verify the prediction effect of the target horizontal well yield prediction model obtained by the method (namely, the effect of predicting the yield sequence of the second part by using the static data and the yield sequence of the first part), the method also performs yield prediction by using a dynamic sequence model (namely, predicting the yield sequence of the second part by using the yield sequence of the first part). The dynamic sequence model can be a long-term and short-term memory neural network model and the like.
As shown in fig. 2, fig. 2 is a schematic diagram of the predicted results of the horizontal well yield prediction model and the long-short term memory neural network model provided in the embodiment of the present specification on the 100 th and 450 th samples on the test set. The left graph is the prediction result of the horizontal well yield prediction model, the right graph is the prediction result of the long-short term memory neural network model, the horizontal coordinate represents the number of years, the vertical coordinate represents the gas production rate/relative error, the error represents the relative error, the numerical simulation value represents the yield sequence of the second part in the test set, namely the true value, and the prediction error represents the average relative error of 750 samples in the test set.
As shown in fig. 3, fig. 3 is a relative error cumulative distribution histogram predicted by a horizontal well yield prediction model and a long-short term memory neural network model for 750 samples according to an embodiment of the present disclosure, where a left side graph is the relative error cumulative distribution histogram predicted by the horizontal well yield prediction model for 750 samples, a right side graph is the relative error cumulative distribution histogram predicted by the long-short term memory neural network model for 750 samples, an abscissa represents a relative error, and an ordinate represents a sample proportion. As can be seen from fig. 3, for the horizontal well yield prediction model, 55.47% of the sample relative error is less than 0.5%, which is about 15% more than that of the long-short term memory neural network model, the minimum error predicted by the horizontal well yield prediction model is within 10%, and the minimum error of the long-short term memory neural network model is more than 10%. Therefore, the prediction effect of the horizontal well yield prediction model obtained by combining with the static data training is superior to that of a long-term and short-term memory neural network model obtained without combining with the static data training.
Therefore, the horizontal well yield prediction model obtained by training can greatly improve the efficiency of yield prediction and the accuracy of prediction, and has strong practicability.
In the embodiment of the specification, a horizontal well yield prediction model established by using static data and dynamic data of each fractured horizontal well can be used for capacity decrement prediction.
From the above description, it can be seen that the static data and the dynamic data corresponding to each target horizontal well can be obtained in the embodiment of the application, wherein the dynamic data includes yield data corresponding to each time point within a preset time, and the dynamic data of each target horizontal well is divided into first dynamic data and second dynamic data based on a predicted time point to obtain a sample data set; and training a convolutional coding dynamic sequence neural network by using the sample data set within the preset time to obtain a target horizontal well yield prediction model. Because the static parameters of the fractured horizontal well are considered when the convolutional coding dynamic sequence neural network is trained, the existing data can be utilized more fully, the combination of the static data and the dynamic data is realized, the yield prediction accuracy of the fractured horizontal well is improved, and the yield sequence can be predicted for a long time by the yield sequence, so that the method is more suitable for practical application and better guides the production and development of oil fields.
In the present specification, each embodiment of the method is described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from other embodiments. Reference is made to the description of the method embodiments.
Based on the yield prediction method based on the convolutional encoding dynamic sequence network, one or more embodiments of the present specification further provide a yield prediction apparatus based on the convolutional encoding dynamic sequence network. The apparatus may include systems (including distributed systems), software (applications), modules, components, servers, clients, etc. that use the methods described in the embodiments of the present specification in conjunction with any necessary apparatus to implement the hardware. Based on the same innovative conception, embodiments of the present specification provide an apparatus as described in the following embodiments. Since the implementation scheme of the apparatus for solving the problem is similar to that of the method, the specific implementation of the apparatus in the embodiment of the present specification may refer to the implementation of the foregoing method, and repeated details are not repeated. As used hereinafter, the term "unit" or "module" may be a combination of software and/or hardware that implements a predetermined function. Although the means described in the embodiments below are preferably implemented in software, an implementation in hardware, or a combination of software and hardware is also possible and contemplated.
Specifically, fig. 4 is a schematic block structure diagram of a yield prediction apparatus based on a convolutional coding dynamic sequence network according to an embodiment of the present disclosure, and as shown in fig. 4, the yield prediction apparatus based on a convolutional coding dynamic sequence network according to the present disclosure may include: an acquisition module 120, a partitioning module 122, and a training module 124.
An obtaining module 120, configured to obtain static data and dynamic data corresponding to each target horizontal well; the dynamic data comprise yield data corresponding to each time point in preset time;
the dividing module 122 is configured to divide the dynamic data of each target horizontal well into first dynamic data and second dynamic data based on the predicted time point, and obtain a sample data set; wherein the predicted time point is within the preset time;
and the training module 124 is used for training the convolutional coding dynamic sequence neural network by using the sample data set to obtain a target horizontal well yield prediction model.
It should be noted that the above-mentioned description of the apparatus according to the method embodiment may also include other embodiments, and specific implementation manners may refer to the description of the related method embodiment, which is not described herein again.
The present specification also provides an embodiment of a yield prediction apparatus based on a convolutional encoded dynamic sequence network, including a processor and a memory for storing processor-executable instructions, where the processor executes the instructions to implement any one of the above method embodiments. For example, the instructions when executed by the processor implement steps comprising: acquiring static data and dynamic data corresponding to each target horizontal well; the dynamic data comprise yield data corresponding to each time point in preset time; dividing the dynamic data of each target horizontal well into first dynamic data and second dynamic data based on the predicted time point to obtain a sample data set; wherein the predicted time point is within the preset time; and training a convolutional coding dynamic sequence neural network by using the sample data set to obtain a target horizontal well yield prediction model.
It should be noted that the above-mentioned apparatuses may also include other embodiments according to the description of the method or apparatus embodiments. The specific implementation manner may refer to the description of the related method embodiment, and is not described in detail herein.
The method embodiments provided in the present specification may be executed in a mobile terminal, a computer terminal, a server or a similar computing device. Taking an example of the above operation on a server, fig. 5 is a block diagram of a hardware structure of a yield prediction server based on a convolutional coding dynamic sequence network according to an embodiment of the present disclosure, where the server may be a yield prediction apparatus based on a convolutional coding dynamic sequence network or a yield prediction system based on a convolutional coding dynamic sequence network according to the above embodiment. As shown in fig. 5, the server 10 may include one or more (only one shown) processors 100 (the processors 100 may include, but are not limited to, a processing device such as a microprocessor MCU or a programmable logic device FPGA, etc.), a memory 200 for storing data, and a transmission module 300 for communication functions. It will be understood by those skilled in the art that the structure shown in fig. 5 is only an illustration and is not intended to limit the structure of the electronic device. For example, the server 10 may also include more or fewer components than shown in FIG. 5, and may also include other processing hardware, such as a database or multi-level cache, a GPU, or have a different configuration than shown in FIG. 5, for example.
The memory 200 may be used to store software programs and modules of application software, such as program instructions/modules corresponding to the yield prediction method based on the convolutional encoding dynamic sequence network in the embodiment of the present specification, and the processor 100 executes various functional applications and data processing by executing the software programs and modules stored in the memory 200. Memory 200 may include high speed random access memory and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, memory 200 may further include memory located remotely from processor 100, which may be connected to a computer terminal through a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
The transmission module 300 is used for receiving or transmitting data via a network. Specific examples of the network described above may include a wireless network provided by a communication provider of the computer terminal. In one example, the transmission module 300 includes a Network adapter (NIC) that can be connected to other Network devices through a base station so as to communicate with the internet. In one example, the transmission module 300 may be a Radio Frequency (RF) module, which is used for communicating with the internet in a wireless manner.
The foregoing description has been directed to specific embodiments of this disclosure. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.
The method or apparatus provided by the present specification and described in the foregoing embodiments may implement service logic through a computer program and record the service logic on a storage medium, where the storage medium may be read and executed by a computer, so as to implement the effect of the solution described in the embodiments of the present specification. The storage medium may include a physical device for storing information, and typically, the information is digitized and then stored using an electrical, magnetic, or optical media. The storage medium may include: devices that store information using electrical energy, such as various types of memory, e.g., RAM, ROM, etc.; devices that store information using magnetic energy, such as hard disks, floppy disks, tapes, core memories, bubble memories, and usb disks; devices that store information optically, such as CDs or DVDs. Of course, there are other ways of storing media that can be read, such as quantum memory, graphene memory, and so forth.
The embodiment of the method or the apparatus for predicting yield based on the convolutional coding dynamic sequence network provided in this specification may be implemented in a computer by executing corresponding program instructions by a processor, for example, implemented in a PC end using a c + + language of a windows operating system, implemented in a linux system, or implemented in an intelligent terminal using, for example, android, an iOS system programming language, implemented in processing logic based on a quantum computer, and the like.
It should be noted that descriptions of the apparatus, the device, and the system described above according to the related method embodiments may also include other embodiments, and specific implementations may refer to descriptions of corresponding method embodiments, which are not described in detail herein.
The embodiments in the present application are described in a progressive manner, and the same and similar parts among the embodiments can be referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the hardware + program class embodiment, since it is substantially similar to the method embodiment, the description is simple, and the relevant points can be referred to the partial description of the method embodiment.
Claims (10)
1. A yield prediction method based on a convolutional coding dynamic sequence network is characterized by comprising the following steps:
acquiring static data and dynamic data corresponding to each target horizontal well; the dynamic data comprise yield data corresponding to each time point in preset time;
dividing the dynamic data of each target horizontal well into first dynamic data and second dynamic data based on the predicted time point to obtain a sample data set; wherein the predicted time point is within the preset time;
and training a convolutional coding dynamic sequence neural network by using the sample data set to obtain a target horizontal well yield prediction model.
2. The method of claim 1, wherein the obtaining of the static data and the dynamic data corresponding to each target horizontal well comprises:
determining the value range of each static parameter of the target horizontal well;
based on the value range of each static parameter, generating static data corresponding to each target horizontal well by using a Latin hypercube sampling method;
and respectively inputting the static data of each target horizontal well into shale gas reservoir numerical simulation software to obtain the dynamic data corresponding to each target horizontal well.
3. The method of claim 2, wherein the static parameters comprise: reservoir thickness, initial formation pressure, matrix permeability, matrix porosity, fracture modification volumetric region permeability, fracture modification volumetric region porosity, fracture half-length, fracture permeability, fracture number, and fracture conductivity.
4. The method of claim 1, wherein training a convolutional encoding dynamic sequence neural network with the sample data set to obtain a target horizontal well production prediction model comprises:
dividing the sample data set into a training set and a test set;
taking the static data and the first dynamic data in the training set as input and the second dynamic data as output, and training the convolutional encoding dynamic sequence neural network to obtain a first horizontal well yield prediction model;
inputting the static data and the first dynamic data in the test set into the first horizontal well yield prediction model to obtain a first prediction result;
and under the condition that the first prediction result and the second dynamic data in the test set meet preset conditions, taking the first horizontal well yield prediction model as a target horizontal well yield prediction model.
5. The method of claim 4, wherein the convolutionally encoded dynamic sequence neural network comprises a convolutional neural network portion, an information fusion layer, an encoder portion, and a decoder portion.
6. The method of claim 5, wherein training the convolutionally encoded dynamic sequence neural network with static data and first dynamic data in the training set as inputs and second dynamic data as outputs comprises:
performing feature extraction on the static data of each target horizontal well based on the convolutional neural network part to obtain the static feature data of each target horizontal well;
fusing the static characteristic data and the first dynamic data of each target horizontal well based on the information fusion layer to obtain fusion data corresponding to each target horizontal well;
encoding fusion data corresponding to each target horizontal well based on the encoder part to obtain an intermediate vector;
decoding the intermediate vector based on the decoder portion to obtain an output result.
7. The method of claim 6, wherein the static characteristic data and the first dynamic data for each target horizontal well are fused by:
g＝σ(Wcxc+Wlxl+b)
xcombined＝g·xc+(1-g)·xl
wherein g represents weight in information fusion, sigma represents sigoid activation function, and xcRepresenting static characteristic data, xlRepresenting first dynamic data, Wc、WlRespectively representing the weights of the network layers, b representing the offsets of the network layers, xcombinedRepresenting the fused data.
8. An apparatus for predicting yield based on a convolutional encoded dynamic sequence network, comprising:
the acquisition module is used for acquiring static data and dynamic data corresponding to each target horizontal well; the dynamic data comprise yield data corresponding to each time point in preset time;
the dividing module is used for dividing the dynamic data of each target horizontal well into first dynamic data and second dynamic data based on the predicted time point to obtain a sample data set; wherein the predicted time point is within the preset time;
and the training module is used for training the convolutional coding dynamic sequence neural network by using the sample data set to obtain a target horizontal well yield prediction model.
9. A yield prediction device based on a convolutional encoded dynamic sequence network, comprising at least one processor and a memory storing computer executable instructions, which when executed by the processor implement the steps of the method of any one of claims 1 to 7.
10. A computer-readable storage medium having stored thereon computer instructions which, when executed, implement the steps of the method of any one of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111219923.2A CN113962148B (en) | 2021-10-20 | 2021-10-20 | Yield prediction method, device and equipment based on convolutional coding dynamic sequence network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111219923.2A CN113962148B (en) | 2021-10-20 | 2021-10-20 | Yield prediction method, device and equipment based on convolutional coding dynamic sequence network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113962148A true CN113962148A (en) | 2022-01-21 |
| CN113962148B CN113962148B (en) | 2022-09-13 |
Family
ID=79465582
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111219923.2A Active CN113962148B (en) | 2021-10-20 | 2021-10-20 | Yield prediction method, device and equipment based on convolutional coding dynamic sequence network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113962148B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115169761A (en) * | 2022-09-08 | 2022-10-11 | 中科数智能源科技(深圳)有限公司 | GNN and LSTM based complex well pattern oil field yield prediction method |
| CN115929289A (en) * | 2022-12-05 | 2023-04-07 | 西南石油大学 | Shale gas yield prediction method and device based on time sequence |
| CN116151480A (en) * | 2023-04-04 | 2023-05-23 | 北京科技大学 | Shale oil well yield prediction method and device |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112228054A (en) * | 2020-11-06 | 2021-01-15 | 中国石油大学(北京) | Method, device and equipment for determining shale gas yield based on convolutional neural network |
| CN112761628A (en) * | 2021-01-25 | 2021-05-07 | 中国石油大学(北京) | Shale gas yield determination method and device based on long-term and short-term memory neural network |
-
2021
- 2021-10-20 CN CN202111219923.2A patent/CN113962148B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112228054A (en) * | 2020-11-06 | 2021-01-15 | 中国石油大学(北京) | Method, device and equipment for determining shale gas yield based on convolutional neural network |
| CN112761628A (en) * | 2021-01-25 | 2021-05-07 | 中国石油大学(北京) | Shale gas yield determination method and device based on long-term and short-term memory neural network |
Non-Patent Citations (1)
| Title |
|---|
| 谷建伟等: "基于数据挖掘的长短期记忆网络模型油井产量预测方法", 《特种油气藏》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115169761A (en) * | 2022-09-08 | 2022-10-11 | 中科数智能源科技(深圳)有限公司 | GNN and LSTM based complex well pattern oil field yield prediction method |
| CN115169761B (en) * | 2022-09-08 | 2022-12-13 | 中科数智能源科技(深圳)有限公司 | GNN and LSTM based complex well pattern oil field yield prediction method |
| CN115929289A (en) * | 2022-12-05 | 2023-04-07 | 西南石油大学 | Shale gas yield prediction method and device based on time sequence |
| CN115929289B (en) * | 2022-12-05 | 2024-05-28 | 西南石油大学 | Shale gas yield prediction method and device based on time sequence |
| CN116151480A (en) * | 2023-04-04 | 2023-05-23 | 北京科技大学 | Shale oil well yield prediction method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113962148B (en) | 2022-09-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113962148B (en) | Yield prediction method, device and equipment based on convolutional coding dynamic sequence network | |
| CN113052371B (en) | Residual oil distribution prediction method and device based on deep convolutional neural network | |
| CN112228054B (en) | Method, device and equipment for determining shale gas yield based on convolutional neural network | |
| CN112761628B (en) | Shale gas yield determination method and device based on long-term and short-term memory neural network | |
| CN111832227B (en) | Shale gas saturation determination method, device and equipment based on deep learning | |
| CN112766319B (en) | Dialogue intention recognition model training method, device, computer equipment and medium | |
| RU2577256C1 (en) | System and method for reservoir simulation using requested data | |
| CN109598334B (en) | Sample generation method and device | |
| CN115186936B (en) | Optimal well pattern construction method for oil field based on GNN model | |
| CN110414592A (en) | A kind of Digital verification code recognition methods based on multi-task learning | |
| CN113779881A (en) | Method, device and equipment for predicting capacity of dense water-containing gas reservoir | |
| CN112016774A (en) | Distribution network running state identification method and system based on data enhancement technology | |
| CN110486009B (en) | Automatic parameter reverse solving method and system for infinite stratum | |
| US20230140905A1 (en) | Systems and methods for completion optimization for waterflood assets | |
| CN116861187A (en) | Lithology recognition model training method, lithology recognition method and related devices | |
| CN113447997B (en) | Reservoir fracture identification method, identification device and identification system | |
| Barros et al. | Clustering techniques for value-of-information assessment in closed-loop reservoir management | |
| CN112419493B (en) | Shale reservoir three-dimensional attribute model building method and device | |
| CN115130268A (en) | Oil reservoir injection-production scheme design method and system, storage medium and electronic equipment | |
| CN112834404A (en) | Method and device for determining gas production capacity of shale reservoir | |
| CN114154435B (en) | Method, device and equipment for determining porosity of formation fluid | |
| CN117744354B (en) | Simulation method and device for capacity data distribution | |
| CN118411043A (en) | Unconventional shale gas reservoir productivity prediction method and device based on Bayesian optimization | |
| Horváth | Defining depositional environment by using neural network | |
| CN118506908A (en) | Method and device for generating small molecular compound based on condition generation model of protein-small molecular interaction fingerprint |
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 |