CN113627607A - Carbonate reservoir sedimentary facies identification method and device, electronic equipment and medium - Google Patents
Carbonate reservoir sedimentary facies identification method and device, electronic equipment and medium Download PDFInfo
- Publication number
- CN113627607A CN113627607A CN202010378218.6A CN202010378218A CN113627607A CN 113627607 A CN113627607 A CN 113627607A CN 202010378218 A CN202010378218 A CN 202010378218A CN 113627607 A CN113627607 A CN 113627607A
- Authority
- CN
- China
- Prior art keywords
- neural network
- training
- time
- network model
- long
- 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
- 208000035126 Facies Diseases 0.000 title claims abstract description 84
- 238000000034 method Methods 0.000 title claims abstract description 56
- BVKZGUZCCUSVTD-UHFFFAOYSA-L Carbonate Chemical compound [O-]C([O-])=O BVKZGUZCCUSVTD-UHFFFAOYSA-L 0.000 title claims abstract description 37
- 238000012549 training Methods 0.000 claims abstract description 149
- 238000003062 neural network model Methods 0.000 claims abstract description 98
- 230000015654 memory Effects 0.000 claims abstract description 82
- 238000013528 artificial neural network Methods 0.000 claims abstract description 47
- 230000007787 long-term memory Effects 0.000 claims abstract description 35
- 230000006403 short-term memory Effects 0.000 claims abstract description 23
- 238000004364 calculation method Methods 0.000 claims description 20
- 238000012360 testing method Methods 0.000 claims description 20
- 238000007781 pre-processing Methods 0.000 claims description 16
- 230000008569 process Effects 0.000 claims description 15
- 238000001914 filtration Methods 0.000 claims description 11
- 230000006870 function Effects 0.000 claims description 7
- 238000011478 gradient descent method Methods 0.000 claims description 7
- 238000004590 computer program Methods 0.000 claims description 5
- 238000010586 diagram Methods 0.000 description 14
- 239000011159 matrix material Substances 0.000 description 10
- 230000001537 neural effect Effects 0.000 description 10
- 230000000694 effects Effects 0.000 description 6
- 210000002569 neuron Anatomy 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 238000010606 normalization Methods 0.000 description 3
- 230000000644 propagated effect Effects 0.000 description 3
- 238000013145 classification model Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 1
- 238000007621 cluster analysis Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000000704 physical effect Effects 0.000 description 1
- 238000012847 principal component analysis method Methods 0.000 description 1
- 239000011435 rock Substances 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A carbonate reservoir sedimentary facies identification method, apparatus, electronic device and medium are disclosed. The method can comprise the following steps: dividing a reservoir into a plurality of sedimentary facies, and carrying out pretreatment aiming at logging data to obtain the pretreated logging data; establishing a standard training library according to the preprocessed logging data of the multiple sedimentary facies; constructing a long-time memory neural network model, and setting initial parameters of the neural network; inputting samples of the standard training library into the long-term and short-term memory neural network model for learning training, and adjusting neural network parameters to obtain a final long-term and short-term memory neural network model; and inputting the preprocessed logging data of the target well into a final long-term memory neural network model, and predicting the reservoir sedimentary facies type of the target well. According to the method, the sedimentary facies of the reservoir can be automatically divided through the long-time memory network, the automatic identification precision is high, the manual workload can be reduced, and the working efficiency is improved.
Description
Technical Field
The invention relates to the field of intelligent interpretation of a logging curve of an oil reservoir, in particular to a carbonate reservoir sedimentary facies identification method, a device, electronic equipment and a medium.
Background
Carbonate reservoirs are a hot spot field for oil and gas exploration and development at home and abroad. Along with the expansion of exploration and development to the deep layer and the deep sea, the number of drilled wells is reduced, the data quality is poor, the sedimentary facies accuracy requirement of a condition constrained well is higher in geological modeling and reserve calculation of a carbonate reservoir, the rock core data is limited, the standards for manually distinguishing the sedimentary facies of a non-coring well are not uniform, and the challenge is brought to the recognition of the carbonate sedimentary facies by the traditional method.
The logging curve can provide the change of reservoir information of the oil reservoir along with the depth, is a real reflection of the physical property of the reservoir of the oil reservoir, and can be used for judging the type of a sedimentary facies. However, the interpretation of the current logging data mainly involves manual qualitative interpretation, and although computer automatic interpretation by a cluster analysis method, a principal component analysis method, a support vector machine method and the like has been attempted, the effect of dividing the sedimentary facies by using the logging curves is not satisfactory due to the complexity of geological problems and the limitations of the methods.
The neural network can establish a structural complex mapping relation between a logging phase and a geological phase, and complex lithology can be identified on a complex lithology section, but the traditional neural network method depends on the accuracy of a sample and prediction data, and the effect of dividing reservoir lithology is not ideal. Compared with the traditional neural network, the circular neural network allows data to flow between layer units, solves the problem of time sequence, but has the problem of long-distance dependence.
In summary, the existing method for identifying reservoir sedimentary facies according to the well logging curve mainly has the following disadvantages: (1) the lithology recognition precision of the existing method is not high, the dependence of all recognition methods on data is high, and the migration learning capability is poor; (2) the current automatic identification method for the lithology of the well logging cannot reflect that the sedimentary facies has certain continuity characteristics in depth. On the geological deposit of the oil reservoir, a sedimentary facies has certain continuity; (3) the cyclic neural network, especially the long-term memory network, is still in the beginning stage in the field of reservoir sedimentary facies identification, and the application in this respect is less. Therefore, there is a need to develop a carbonate reservoir sedimentary facies identification method, apparatus, electronic device and medium.
The information disclosed in this background section is only for enhancement of understanding of the general background of the invention and should not be taken as an acknowledgement or any form of suggestion that this information forms the prior art already known to a person skilled in the art.
Disclosure of Invention
The invention provides a carbonate reservoir sedimentary facies identification method, a carbonate reservoir sedimentary facies identification device, electronic equipment and a medium, which can realize a method for automatically dividing a reservoir sedimentary facies through a long-time memory network, have higher automatic identification precision, reduce manual workload and improve working efficiency.
In a first aspect, an embodiment of the present disclosure provides a carbonate reservoir sedimentary facies identification method, including: dividing a reservoir into a plurality of sedimentary facies, and carrying out pretreatment aiming at logging data to obtain the pretreated logging data; establishing a standard training library according to the preprocessed logging data of the multiple sedimentary facies, wherein the standard training library comprises a training set and a testing set; constructing a long-time memory neural network model, and setting initial parameters of the neural network; inputting the samples of the standard training library into a long-term and short-term memory neural network model for learning training, and adjusting neural network parameters to obtain a final long-term and short-term memory neural network model; inputting the preprocessed logging data of the target well into the final long-time and short-time memory neural network model, and predicting the reservoir sedimentary facies type of the target well.
Preferably, the pre-processing comprises filtering and de-noising and normalising the well log data.
Preferably, the well log data is normalized by equation (1):
wherein x is the logging data, mean (x) represents the average value of the logging data x, std (x) is the standard deviation of the logging data x, and y is the normalized data.
Preferably, the neural network initial parameters include the number of network layers, the number of hidden units, an error function, a learning rate, the number of learning times, and an initial value of the number of samples selected in a single training.
Preferably, inputting the sample of the standard training library into the long-term and short-term memory neural network model for learning training, and adjusting the neural network parameters to obtain the final long-term and short-term memory neural network model includes: inputting the samples of the training set into a long-time and short-time memory neural network model, carrying out single training, and outputting a simulated lithofacies label; calculating a label error according to the lithofacies label of the sample corresponding to the training set and the simulated lithofacies label; adjusting the neural network parameters by a gradient descent method until the label error is smaller than a set threshold value or reaches a set learning frequency, and obtaining an optimized long-time and short-time memory neural network model; and verifying the optimized long-time and short-time memory neural network model through a test set, and adjusting the number of layers of the network and the number of hidden units to obtain a final long-time and short-time memory neural network model.
Preferably, performing a single training session comprises: inputting the samples of the training set into the long-time memory neural network model for learning and training; after the calculation at the current time t is finished, interacting the sample calculation result through a forgetting gate, an input gate and an output gate, turning to the next time t +1, and performing learning training again; the above process is repeated until a single training is completed.
Preferably, the label error is calculated by equation (2):
wherein, L is the label error,simulating the kth sample of the facies label for time t,is composed ofAnd (4) corresponding to the kth sample of the lithofacies label at the time T, wherein n is the number of samples, and T is iteration time.
As a specific implementation of the embodiments of the present disclosure,
in a second aspect, embodiments of the present disclosure also provide a carbonate reservoir sedimentary facies identification apparatus, including: the preprocessing module is used for dividing the reservoir into a plurality of sedimentary facies, and preprocessing the logging data to obtain the preprocessed logging data; the database building module is used for building a standard training database according to the preprocessed logging data of the multiple sedimentary facies, wherein the standard training database comprises a training set and a testing set; the modeling module is used for constructing a long-term memory neural network model and setting initial parameters of the neural network; the adjusting module is used for inputting the samples of the standard training library into the long-time and short-time memory neural network model for learning training, adjusting neural network parameters and obtaining a final long-time and short-time memory neural network model; and the calculation module is used for inputting the preprocessed logging data of the target well into the final long-time and short-time memory neural network model and predicting the reservoir sedimentary facies type of the target well.
Preferably, the pre-processing comprises filtering and de-noising and normalising the well log data.
Preferably, the well log data is normalized by equation (1):
wherein x is the logging data, mean (x) represents the average value of the logging data x, std (x) is the standard deviation of the logging data x, and y is the normalized data.
Preferably, the neural network initial parameters include the number of network layers, the number of hidden units, an error function, a learning rate, the number of learning times, and an initial value of the number of samples selected in a single training.
Preferably, inputting the sample of the standard training library into the long-term and short-term memory neural network model for learning training, and adjusting the neural network parameters to obtain the final long-term and short-term memory neural network model includes: inputting the samples of the training set into a long-time and short-time memory neural network model, carrying out single training, and outputting a simulated lithofacies label; calculating a label error according to the lithofacies label of the sample corresponding to the training set and the simulated lithofacies label; adjusting the neural network parameters by a gradient descent method until the label error is smaller than a set threshold value or reaches a set learning frequency, and obtaining an optimized long-time and short-time memory neural network model; and verifying the optimized long-time and short-time memory neural network model through a test set, and adjusting the number of layers of the network and the number of hidden units to obtain a final long-time and short-time memory neural network model.
Preferably, performing a single training session comprises: inputting the samples of the training set into the long-time memory neural network model for learning and training; after the calculation at the current time t is finished, interacting the sample calculation result through a forgetting gate, an input gate and an output gate, turning to the next time t +1, and performing learning training again; the above process is repeated until a single training is completed.
Preferably, the label error is calculated by equation (2):
wherein, L is the label error,simulating the kth sample of the facies label for time t,is composed ofAnd (4) corresponding to the kth sample of the lithofacies label at the time T, wherein n is the number of samples, and T is iteration time.
In a third aspect, an embodiment of the present disclosure further provides an electronic device, where the electronic device includes:
a memory storing executable instructions;
a processor executing the executable instructions in the memory to implement the carbonate reservoir sedimentary facies identification method.
In a fourth aspect, the disclosed embodiments also provide a computer-readable storage medium storing a computer program, which when executed by a processor implements the carbonate reservoir depofacies identification method.
The method and apparatus of the present invention have other features and advantages which will be apparent from or are set forth in detail in the accompanying drawings and the following detailed description, which are incorporated herein, and which together serve to explain certain principles of the invention.
Drawings
The above and other objects, features and advantages of the present invention will become more apparent by describing in more detail exemplary embodiments thereof with reference to the attached drawings, in which like reference numerals generally represent like parts.
Fig. 1 shows a flow chart of the steps of a carbonate reservoir depofacies identification method according to one embodiment of the present invention.
FIGS. 2a and 2b are schematic diagrams illustrating results of well log data filtering and normalization, respectively, according to an embodiment of the invention.
FIG. 3 shows a schematic diagram of a depositional facies standards training library, according to one embodiment of the invention.
FIG. 4 shows a schematic diagram of a long term memory neural network training process according to an embodiment of the present invention.
FIG. 5 shows a schematic diagram of a test set confusion matrix, according to one embodiment of the invention.
FIG. 6 shows a schematic diagram of a target well prediction result according to an embodiment of the invention.
FIG. 7 illustrates a schematic diagram of a target well prediction confusion matrix, according to one embodiment of the invention.
Fig. 8 shows a block diagram of a carbonate reservoir sedimentary facies identification apparatus according to an embodiment of the present invention.
Description of reference numerals:
201. a preprocessing module; 202. building a library module; 203. a modeling module; 204. an adjustment module; 205. and a calculation module.
Detailed Description
Preferred embodiments of the present invention will be described in more detail below. While the following describes preferred embodiments of the present invention, it should be understood that the present invention may be embodied in various forms and should not be limited by the embodiments set forth herein.
The invention provides a carbonate reservoir sedimentary facies identification method, which comprises the following steps:
and (4) manually explaining the logging curve by combining geological sedimentary data, and dividing the reservoir into a plurality of sedimentary facies according to different sedimentary positions. Preprocessing logging data to obtain preprocessed logging data; in one example, the pre-processing includes filtering, de-noising, and normalizing the well log data.
Specifically, the logging data is required to be filtered and denoised in consideration of the fact that the logging data contains noise; the range of values for different types of logs varies, and therefore, the log data is normalized by equation (1).
Establishing a standard training library according to the preprocessed logging data of the multiple sedimentary facies; specifically, typical, representative and good-quality logging data are selected to construct a sedimentary facies standard training library, the sedimentary facies standard training library comprises depth information of a certain point of an oil reservoir stratum, various logging data and sedimentary facies types, a set of data of the depth position forms 1 sample, and the sample size of each sedimentary facies in the standard training library is kept consistent. Samples of the standard training library were divided into training and test sets on a 90% and 10% scale.
Constructing a long-time memory neural network model, and setting initial parameters of the neural network, including the number of network layers, the number of hidden units, an error function, a learning rate, the learning times and an initial value of the number of samples selected in single training;
and inputting the logging data of the standard training library into the long-term and short-term memory neural network model for learning training, and adjusting neural network parameters to obtain the final long-term and short-term memory neural network model.
In one example, inputting samples of the standard training library into the long-term and short-term memory neural network model for learning training, and adjusting neural network parameters to obtain a final long-term and short-term memory neural network model includes:
inputting samples of the training set into a long-time and short-time memory neural network model, carrying out single training, and outputting a simulated lithofacies label; in one example, performing a single training includes: inputting samples of the training set into a long-time memory neural network model for learning and training; after the calculation at the current time t is finished, interacting the sample calculation result through a forgetting gate, an input gate and an output gate, turning to the next time t + 1, and performing learning training again; the above process is repeated until a single training is completed.
Calculating a label error through a formula (2) according to the lithofacies label of the sample corresponding to the training set and the simulated lithofacies label; and a confusion matrix of lithofacies labels of the samples corresponding to the training set and the simulated lithofacies labels can be trained, and records in the data set can be summarized according to two standards of real category judgment and category judgment predicted by the classification model in a matrix form.
And adjusting the neural network parameters by a gradient descent method until the label error is smaller than a set threshold value or reaches a set learning frequency, and obtaining an optimized long-time and short-time memory neural network model.
Specifically, the learning training is divided into two phases: the first stage is forward propagation of data, and on the basis of a threshold value and a weight value of a network obtained by training in the previous step, the output of each neuron is calculated backwards from a first layer which is used for memorizing a neural network model at long and short time; the second stage is the process of error back propagation, label errors are reversely propagated to the input layer by layer through the hidden layer in a certain form, the errors are distributed to all units of each layer, the errors of the units of each layer are obtained, and then the neural unit parameters of the error are updated.
And continuously repeating the learning and training process, and adjusting the neural network parameters including the structure of the neural network model and the parameters of the neural units until the label error is reduced to an acceptable degree or the learning is carried out for a preset learning frequency, so as to obtain the optimized long-time memory neural network model.
And verifying the optimized long-time and short-time memory neural network model through the test set, and adjusting the number of layers of the network and the number of hidden units to obtain the final long-time and short-time memory neural network model.
And inputting the preprocessed logging data of the target well into a final long-term memory neural network model, and predicting the reservoir sedimentary facies type of the target well.
The invention also provides a carbonate reservoir sedimentary facies identification device, which comprises:
and the preprocessing module is used for manually explaining the logging curve by combining geological sedimentary data and dividing the reservoir into a plurality of sedimentary facies according to different sedimentary positions. Preprocessing logging data to obtain preprocessed logging data; in one example, the pre-processing includes filtering, de-noising, and normalizing the well log data.
Specifically, the logging data is required to be filtered and denoised in consideration of the fact that the logging data contains noise; the range of values for different types of logs varies, and therefore, the log data is normalized by equation (1).
The database building module builds a standard training database according to the preprocessed logging data of the multiple sedimentary facies; specifically, typical, representative and good-quality logging data are selected to construct a sedimentary facies standard training library, the sedimentary facies standard training library comprises depth information of a certain point of an oil reservoir stratum, various logging data and sedimentary facies types, a set of data of the depth position forms 1 sample, and the sample size of each sedimentary facies data in the standard training library is kept consistent. Samples of the standard training library were divided into training and test sets on a 90% and 10% scale.
And the modeling module is used for constructing a long-time memory neural network model and setting initial parameters of the neural network, including the number of network layers, the number of hidden units, an error function, a learning rate, the learning times and an initial value of the number of samples selected in single training.
And the adjusting module is used for inputting the logging data of the standard training library into the long-time and short-time memory neural network model for learning training, adjusting the neural network parameters and obtaining the final long-time and short-time memory neural network model.
In one example, inputting samples of the standard training library into the long-term and short-term memory neural network model for learning training, and adjusting neural network parameters to obtain a final long-term and short-term memory neural network model includes:
inputting samples of the training set into a long-time and short-time memory neural network model, carrying out single training, and outputting a simulated lithofacies label; in one example, performing a single training includes: inputting samples of the training set into a long-time memory neural network model for learning and training; after the calculation at the current time t is finished, interacting the sample calculation result through a forgetting gate, an input gate and an output gate, turning to the next time t + 1, and performing learning training again; the above process is repeated until a single training is completed.
Calculating a label error through a formula (2) according to the lithofacies label of the sample corresponding to the training set and the simulated lithofacies label; and a confusion matrix of lithofacies labels of the samples corresponding to the training set and the simulated lithofacies labels can be trained, and records in the data set can be summarized according to two standards of real category judgment and category judgment predicted by the classification model in a matrix form.
And adjusting the neural network parameters by a gradient descent method until the label error is smaller than a set threshold value or reaches a set learning frequency, and obtaining an optimized long-time and short-time memory neural network model.
Specifically, the learning training is divided into two phases: the first stage is forward propagation of data, and on the basis of a threshold value and a weight value of a network obtained by training in the previous step, the output of each neuron is calculated backwards from a first layer which is used for memorizing a neural network model at long and short time; the second stage is the process of error back propagation, label errors are reversely propagated to the input layer by layer through the hidden layer in a certain form, the errors are distributed to all units of each layer, the errors of the units of each layer are obtained, and then the neural unit parameters of the error are updated.
And continuously repeating the learning and training process, and adjusting the neural network parameters including the structure of the neural network model and the parameters of the neural units until the label error is reduced to an acceptable degree or the learning is carried out for a preset learning frequency, so as to obtain the optimized long-time memory neural network model.
And verifying the optimized long-time and short-time memory neural network model through the test set, and adjusting the number of layers of the network and the number of hidden units to obtain the final long-time and short-time memory neural network model.
And the calculation module is used for inputting the preprocessed logging data of the target well into the final long-time memory neural network model and predicting the reservoir sedimentary facies type of the target well.
The present invention also provides an electronic device, comprising: a memory storing executable instructions; and the processor runs executable instructions in the memory to realize the carbonate reservoir sedimentary facies identification method.
The present invention also provides a computer-readable storage medium having stored thereon a computer program which, when being executed by a processor, carries out the above-mentioned carbonate reservoir depofacies identification method.
To facilitate understanding of the scheme of the embodiments of the present invention and the effects thereof, four specific application examples are given below. It will be understood by those skilled in the art that this example is merely for the purpose of facilitating an understanding of the present invention and that any specific details thereof are not intended to limit the invention in any way.
Example 1
Fig. 1 shows a flow chart of the steps of a carbonate reservoir depofacies identification method according to the present invention.
As shown in fig. 1, the carbonate reservoir sedimentary facies identification method includes: step 101, dividing a reservoir into a plurality of sedimentary facies, and carrying out pretreatment on logging data to obtain pretreated logging data; 102, establishing a standard training library according to the preprocessed logging data of a plurality of sedimentary facies, wherein the standard training library comprises a training set and a test set; 103, constructing a long-term memory neural network model and setting initial parameters of the neural network; 104, inputting the samples of the standard training library into a long-and-short time memory neural network model for learning training, and adjusting neural network parameters to obtain a final long-and-short time memory neural network model; and 105, inputting the preprocessed logging data of the target well into the final long-time and short-time memory neural network model, and predicting the reservoir sedimentary facies type of the target well.
Taking a certain marine facies sedimentary carbonate reservoir as an example, according to the geological characteristics of the reservoir, the sedimentary facies is divided by logging data, and the main types of the sedimentary facies are as follows: shoal, before, between shoals, lagoon, shallow sea.
FIGS. 2a and 2b are schematic diagrams illustrating results of well log data filtering and normalization, respectively, according to an embodiment of the invention.
The logging data is subjected to denoising and filtering processing by adopting a mean filtering method, and then is normalized by a formula (1), and filtering and normalization results are respectively shown in fig. 2a and 2 b.
FIG. 3 shows a schematic diagram of a depositional facies standards training library, according to one embodiment of the invention.
The three logging curves of CGR, SGR and PE are used as input, and 5 sedimentary facies in lagoon, shallow sea, shoal, ahead of the shoal and between the shoals are used as output. And selecting sedimentary facies data with typical characteristics to establish a standard training library according to the reservoir sedimentary facies division results and corresponding logging data thereof, as shown in FIG. 3. 206 samples were selected for each dephasic type, for a total of 1030 samples. Samples of the standard training library were divided into training and test sets on a 90% and 10% scale.
Constructing a long-time memory neural network model, and setting initial parameters of the neural network;
and inputting samples of the standard training library into the long-term and short-term memory neural network model for learning training, and adjusting neural network parameters to obtain the final long-term and short-term memory neural network model.
FIG. 4 shows a schematic diagram of a long term memory neural network training process according to an embodiment of the present invention.
Inputting samples of the training set into a long-time and short-time memory neural network model, carrying out single training, and outputting a simulated lithofacies label; performing a single training session includes: inputting samples of the training set into a long-time memory neural network model for learning and training; after the calculation at the current time t is finished, interacting the sample calculation result through a forgetting gate, an input gate and an output gate, turning to the next time t + 1, and performing learning training again; the above process is repeated until a single training is completed.
Calculating a label error through a formula (2) according to the lithofacies label of the sample corresponding to the training set and the simulated lithofacies label; and obtaining a confusion matrix of the lithofacies labels and the simulated lithofacies labels of the samples corresponding to the training set.
And adjusting the neural network parameters by a gradient descent method until the label error is smaller than a set threshold value or reaches a set learning frequency, and obtaining an optimized long-time and short-time memory neural network model.
The learning and training is divided into two stages: the first stage is forward propagation of data, and on the basis of a threshold value and a weight value of a network obtained by training in the previous step, the output of each neuron is calculated backwards from a first layer which is used for memorizing a neural network model at long and short time; the second stage is the process of error back propagation, label errors are reversely propagated to the input layer by layer through the hidden layer in a certain form, the errors are distributed to all units of each layer, the errors of the units of each layer are obtained, and then the neural unit parameters of the error are updated.
The learning and training process is repeated continuously, the neural network parameters including the structure of the neural network model and the parameters of the neural units are adjusted until the label error is reduced to an acceptable degree or the learning is performed for a preset learning frequency, and an optimized long-time and short-time memory neural network model is obtained, as shown in fig. 4, it can be seen that the effect of memorizing the neural network model in the optimized long-time and short-time on the training set is better, and the training precision is 95.42%.
FIG. 5 shows a schematic diagram of a test set confusion matrix, according to one embodiment of the invention.
The learning training also comprises hiding the neural unit, and adjusting the structure of the long-time memory neural network model by using the test set data. The test set data is input into the optimized long-time and short-time memory neural network model, and the test effect is shown in fig. 5. And adjusting the number of the hidden neural units, improving the training precision and obtaining a final long-time and short-time memory neural network model. The relation between the number of the hidden neural units and the accuracy is shown in table 1, and it can be seen that when the number of the hidden neural units is 120, the training effect is the best, and the prediction precision is 82%.
TABLE 1
Number of hidden |
40 | 60 | 80 | 100 | 120 | 140 | 160 | 180 | 200 |
Rate of accuracy | 56% | 61% | 70% | 27% | 82% | 76% | 70% | 71% | 61% |
FIG. 6 shows a schematic diagram of a target well prediction result according to an embodiment of the invention.
FIG. 7 illustrates a schematic diagram of a target well prediction confusion matrix, according to one embodiment of the invention.
Inputting the preprocessed logging data of the target well into a final long-term memory neural network model, predicting the reservoir sedimentary facies type of the target well, wherein the total prediction precision reaches 79.8%, and the prediction result and the prediction confusion matrix are respectively shown in fig. 6 and 7.
Example 2
Fig. 8 shows a block diagram of a carbonate reservoir sedimentary facies identification apparatus according to an embodiment of the present invention.
As shown in fig. 8, the carbonate reservoir sedimentary facies identification apparatus includes:
the preprocessing module 201 is used for dividing the reservoir into a plurality of sedimentary facies, and preprocessing logging data to obtain preprocessed logging data;
the database building module 202 is used for building a standard training database according to the preprocessed logging data of the multiple sedimentary facies, wherein the standard training database comprises a training set and a testing set;
the modeling module 203 is used for constructing a long-term memory neural network model and setting initial parameters of the neural network;
the adjusting module 204 is used for inputting the samples of the standard training library into the long-term and short-term memory neural network model for learning training, adjusting the neural network parameters and obtaining a final long-term and short-term memory neural network model;
and the calculation module 205 is used for inputting the preprocessed logging data of the target well into the final long-term and short-term memory neural network model and predicting the reservoir sedimentary facies type of the target well.
Alternatively, the pre-processing includes filtering, de-noising and normalizing the well log data.
Alternatively, the well log data is normalized by equation (1):
wherein x is the logging data, mean (x) represents the average value of the logging data x, std (x) is the standard deviation of the logging data x, and y is the normalized data.
As an alternative, the initial parameters of the neural network include the number of network layers, the number of hidden units, an error function, a learning rate, the number of learning times, and an initial value of the number of samples selected in a single training.
As an alternative, inputting the samples of the standard training library into the long-term and short-term memory neural network model for learning training, adjusting the neural network parameters, and obtaining the final long-term and short-term memory neural network model comprises: inputting samples of the training set into a long-time and short-time memory neural network model, carrying out single training, and outputting a simulated lithofacies label; calculating a label error according to the lithofacies label and the simulated lithofacies label of the sample corresponding to the training set; adjusting the neural network parameters by a gradient descent method until the label error is smaller than a set threshold value or reaches a set learning frequency, and obtaining an optimized long-time and short-time memory neural network model; and verifying the optimized long-time and short-time memory neural network model through the test set, and adjusting the number of layers of the network and the number of hidden units to obtain the final long-time and short-time memory neural network model.
Alternatively, performing a single training comprises: inputting samples of the training set into a long-time memory neural network model for learning and training; after the calculation at the current time t is finished, interacting the sample calculation result through a forgetting gate, an input gate and an output gate, turning to the next time t + 1, and performing learning training again; the above process is repeated until a single training is completed.
Alternatively, the label error is calculated by equation (2):
wherein, L is the label error,simulating the kth sample of the facies label for time t,is composed ofAnd (4) corresponding to the kth sample of the lithofacies label at the time T, wherein n is the number of samples, and T is iteration time.
Example 3
The present disclosure provides an electronic device including: a memory storing executable instructions; and the processor runs executable instructions in the memory to realize the carbonate reservoir sedimentary facies identification method.
An electronic device according to an embodiment of the present disclosure includes a memory and a processor.
The memory is to store non-transitory computer readable instructions. In particular, the memory may include one or more computer program products that may include various forms of computer-readable storage media, such as volatile memory and/or non-volatile memory. The volatile memory may include, for example, Random Access Memory (RAM), cache memory (cache), and/or the like. The non-volatile memory may include, for example, Read Only Memory (ROM), hard disk, flash memory, etc.
The processor may be a Central Processing Unit (CPU) or other form of processing unit having data processing capabilities and/or instruction execution capabilities, and may control other components in the electronic device to perform desired functions. In one embodiment of the disclosure, the processor is configured to execute the computer readable instructions stored in the memory.
Those skilled in the art should understand that, in order to solve the technical problem of how to obtain a good user experience, the present embodiment may also include well-known structures such as a communication bus, an interface, and the like, and these well-known structures should also be included in the protection scope of the present disclosure.
For the detailed description of the present embodiment, reference may be made to the corresponding descriptions in the foregoing embodiments, which are not repeated herein.
Example 4
Embodiments of the present disclosure provide a computer-readable storage medium having stored thereon a computer program, which, when executed by a processor, implements the carbonate reservoir depofacies identification method.
A computer-readable storage medium according to an embodiment of the present disclosure has non-transitory computer-readable instructions stored thereon. The non-transitory computer readable instructions, when executed by a processor, perform all or a portion of the steps of the methods of the embodiments of the disclosure previously described.
The computer-readable storage media include, but are not limited to: optical storage media (e.g., CD-ROMs and DVDs), magneto-optical storage media (e.g., MOs), magnetic storage media (e.g., magnetic tapes or removable disks), media with built-in rewritable non-volatile memory (e.g., memory cards), and media with built-in ROMs (e.g., ROM cartridges).
It will be appreciated by persons skilled in the art that the above description of embodiments of the invention is intended only to illustrate the benefits of embodiments of the invention and is not intended to limit embodiments of the invention to any examples given.
Having described embodiments of the present invention, the foregoing description is intended to be exemplary, not exhaustive, and not limited to the embodiments disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments.
Claims (10)
1. A carbonate reservoir sedimentary facies identification method, comprising:
dividing a reservoir into a plurality of sedimentary facies, and carrying out pretreatment aiming at logging data to obtain the pretreated logging data;
establishing a standard training library according to the preprocessed logging data of the multiple sedimentary facies, wherein the standard training library comprises a training set and a testing set;
constructing a long-time memory neural network model, and setting initial parameters of the neural network;
inputting the samples of the standard training library into a long-term and short-term memory neural network model for learning training, and adjusting neural network parameters to obtain a final long-term and short-term memory neural network model;
inputting the preprocessed logging data of the target well into the final long-time and short-time memory neural network model, and predicting the reservoir sedimentary facies type of the target well.
2. The carbonate reservoir dephasing identification method of claim 1, wherein the pre-processing comprises filtering denoising and normalizing the log data.
3. The carbonate reservoir sedimentary facies identification method of claim 2, wherein the well log data is normalized by equation (1):
wherein x is the logging data, mean (x) represents the average value of the logging data x, std (x) is the standard deviation of the logging data x, and y is the normalized data.
4. The carbonate reservoir sedimentary facies identification method of claim 1, wherein the neural network initial parameters include a number of network layers, a number of hidden units, an error function, a learning rate, a learning number, and an initial value of a number of samples selected in a single training.
5. The carbonate reservoir sedimentary facies identification method of claim 4, wherein inputting samples of the standard training library into the long-term memory neural network model for learning training and adjusting neural network parameters to obtain a final long-term memory neural network model comprises:
inputting the samples of the training set into a long-time and short-time memory neural network model, carrying out single training, and outputting a simulated lithofacies label;
calculating a label error according to the lithofacies label of the sample corresponding to the training set and the simulated lithofacies label;
adjusting the neural network parameters by a gradient descent method until the label error is smaller than a set threshold value or reaches a set learning frequency, and obtaining an optimized long-time and short-time memory neural network model;
and verifying the optimized long-time and short-time memory neural network model through a test set, and adjusting the number of layers of the network and the number of hidden units to obtain a final long-time and short-time memory neural network model.
6. The carbonate reservoir sedimentary facies identification method of claim 5, wherein performing a single training session includes:
inputting the samples of the training set into the long-time memory neural network model for learning and training;
after the calculation at the current time t is finished, interacting the sample calculation result through a forgetting gate, an input gate and an output gate, turning to the next time t +1, and performing learning training again;
the above process is repeated until a single training is completed.
7. The carbonate reservoir sedimentary facies identification method of claim 6, wherein the tag error is calculated by equation (2):
8. A carbonate reservoir sedimentary facies identification apparatus, comprising:
the preprocessing module is used for dividing the reservoir into a plurality of sedimentary facies, and preprocessing the logging data to obtain the preprocessed logging data;
the database building module is used for building a standard training database according to the preprocessed logging data of the multiple sedimentary facies, wherein the standard training database comprises a training set and a testing set;
the modeling module is used for constructing a long-term memory neural network model and setting initial parameters of the neural network;
the adjusting module is used for inputting the samples of the standard training library into the long-time and short-time memory neural network model for learning training, adjusting neural network parameters and obtaining a final long-time and short-time memory neural network model;
and the calculation module is used for inputting the preprocessed logging data of the target well into the final long-time and short-time memory neural network model and predicting the reservoir sedimentary facies type of the target well.
9. An electronic device, characterized in that the electronic device comprises:
a memory storing executable instructions;
a processor executing the executable instructions in the memory to implement the carbonate reservoir depofacies identification method of any one of claims 1-7.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program which, when executed by a processor, implements the carbonate reservoir depofacies identification method according to any one of claims 1-7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010378218.6A CN113627607A (en) | 2020-05-07 | 2020-05-07 | Carbonate reservoir sedimentary facies identification method and device, electronic equipment and medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010378218.6A CN113627607A (en) | 2020-05-07 | 2020-05-07 | Carbonate reservoir sedimentary facies identification method and device, electronic equipment and medium |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113627607A true CN113627607A (en) | 2021-11-09 |
Family
ID=78376886
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010378218.6A Pending CN113627607A (en) | 2020-05-07 | 2020-05-07 | Carbonate reservoir sedimentary facies identification method and device, electronic equipment and medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113627607A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115421181A (en) * | 2022-07-27 | 2022-12-02 | 北京超维创想信息技术有限公司 | Three-dimensional geological model phase control attribute modeling method based on deep learning |
CN116108368A (en) * | 2023-02-21 | 2023-05-12 | 北京金阳普泰石油技术股份有限公司 | Deposition microphase identification method and device based on deep learning mixed model |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102650702A (en) * | 2012-05-03 | 2012-08-29 | 中国石油天然气股份有限公司 | Seismic waveform analysis and reservoir prediction method and device |
CN106529667A (en) * | 2016-09-23 | 2017-03-22 | 中国石油大学(华东) | Logging facies identification and analysis method based on fuzzy depth learning in big data environment |
CN107316341A (en) * | 2016-04-26 | 2017-11-03 | 中国石油化工股份有限公司 | A kind of Multiple-Point Geostatistics facies modelization method |
CN107545577A (en) * | 2017-08-23 | 2018-01-05 | 电子科技大学 | Sedimentary facies image partition method based on neutral net |
CN109212617A (en) * | 2018-08-24 | 2019-01-15 | 中国石油天然气股份有限公司 | Electric imaging logging phase automatic identifying method and device |
CN109271898A (en) * | 2018-08-31 | 2019-01-25 | 电子科技大学 | Solution cavity body recognizer based on optimization convolutional neural networks |
CN110045356A (en) * | 2019-03-14 | 2019-07-23 | 哈尔滨工程大学 | A kind of two-way length Memory Neural Networks waterborne target recognition methods in short-term |
CN110261467A (en) * | 2019-07-01 | 2019-09-20 | 西南石油大学 | A method of identification carbonate paleokarst reservoir vertical zonality |
-
2020
- 2020-05-07 CN CN202010378218.6A patent/CN113627607A/en active Pending
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102650702A (en) * | 2012-05-03 | 2012-08-29 | 中国石油天然气股份有限公司 | Seismic waveform analysis and reservoir prediction method and device |
CN107316341A (en) * | 2016-04-26 | 2017-11-03 | 中国石油化工股份有限公司 | A kind of Multiple-Point Geostatistics facies modelization method |
CN106529667A (en) * | 2016-09-23 | 2017-03-22 | 中国石油大学(华东) | Logging facies identification and analysis method based on fuzzy depth learning in big data environment |
CN107545577A (en) * | 2017-08-23 | 2018-01-05 | 电子科技大学 | Sedimentary facies image partition method based on neutral net |
CN109212617A (en) * | 2018-08-24 | 2019-01-15 | 中国石油天然气股份有限公司 | Electric imaging logging phase automatic identifying method and device |
CN109271898A (en) * | 2018-08-31 | 2019-01-25 | 电子科技大学 | Solution cavity body recognizer based on optimization convolutional neural networks |
CN110045356A (en) * | 2019-03-14 | 2019-07-23 | 哈尔滨工程大学 | A kind of two-way length Memory Neural Networks waterborne target recognition methods in short-term |
CN110261467A (en) * | 2019-07-01 | 2019-09-20 | 西南石油大学 | A method of identification carbonate paleokarst reservoir vertical zonality |
Non-Patent Citations (1)
Title |
---|
王俊等: "基于长短期记忆网络的钻前测井曲线预测方法", 成都理工大学学报(自然科学版), vol. 47, no. 2, 30 April 2020 (2020-04-30), pages 227 - 236 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115421181A (en) * | 2022-07-27 | 2022-12-02 | 北京超维创想信息技术有限公司 | Three-dimensional geological model phase control attribute modeling method based on deep learning |
CN115421181B (en) * | 2022-07-27 | 2023-10-20 | 北京超维创想信息技术有限公司 | Three-dimensional geological model phase control attribute modeling method based on deep learning |
CN116108368A (en) * | 2023-02-21 | 2023-05-12 | 北京金阳普泰石油技术股份有限公司 | Deposition microphase identification method and device based on deep learning mixed model |
CN116108368B (en) * | 2023-02-21 | 2023-09-01 | 北京金阳普泰石油技术股份有限公司 | Deposition microphase identification method and device based on deep learning mixed model |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN112989708B (en) | Well logging lithology identification method and system based on LSTM neural network | |
WO2019212750A1 (en) | A data fusion technique to compute reservoir quality and completion quality by combining various log measurements | |
US10087723B2 (en) | Methodology for building realistic numerical forward stratigraphic models in data sparse environment | |
CN113837501B (en) | Shale total organic carbon prediction method, system and equipment based on graph neural network | |
US20210286099A1 (en) | Enhanced-resolution rock formation body wave slowness determination from borehole guided waves | |
CN113627607A (en) | Carbonate reservoir sedimentary facies identification method and device, electronic equipment and medium | |
CN114723095A (en) | Missing well logging curve prediction method and device | |
CN110927791B (en) | Method and device for predicting fluid by utilizing seismic data based on deep learning | |
CN111767674A (en) | Well logging lithology identification method based on active domain adaptation | |
CN111027249B (en) | Machine learning-based inter-well connectivity evaluation method | |
CN115877464B (en) | Lithology recognition method and device, computer equipment and storage medium | |
CN115639605B (en) | Automatic identification method and device for high-resolution fault based on deep learning | |
Jervis et al. | Deep learning network optimization and hyperparameter tuning for seismic lithofacies classification | |
CN112444850A (en) | Seismic data velocity modeling method, storage medium, and computing device | |
CN114114421B (en) | Deep learning-based guided self-learning seismic data denoising method and device | |
CN115576011A (en) | Phased reservoir prediction method, device and system based on geological model | |
Denisenko et al. | Automated geosteering while drilling using machine learning. case studies | |
CN114021700A (en) | Permeability calculation method and device based on petrophysical constraint neural network | |
CN114280689A (en) | Method, device and equipment for determining reservoir porosity based on petrophysical knowledge | |
Yu et al. | Training image optimization method based on convolutional neural network and its application in discrete fracture network model selection | |
CN114594513A (en) | Rock physical template establishing method, and gas saturation quantitative prediction method and device | |
CN114528746A (en) | Complex lithology identification method, identification system, electronic device and storage medium | |
Athmer et al. | Integrating Seismic Interpretation, Classification and Geologic Process Modeling for Shale Reservoir Characterization | |
CN112990320A (en) | Lithology classification method and device, electronic equipment and storage medium | |
CN115128671A (en) | Three-dimensional seismic facies automatic identification method based on deep learning technology |
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 |