CN114429075B - Shale specific surface area parameter modeling method based on BP neural network - Google Patents
Shale specific surface area parameter modeling method based on BP neural network Download PDFInfo
- Publication number
- CN114429075B CN114429075B CN202111071183.2A CN202111071183A CN114429075B CN 114429075 B CN114429075 B CN 114429075B CN 202111071183 A CN202111071183 A CN 202111071183A CN 114429075 B CN114429075 B CN 114429075B
- Authority
- CN
- China
- Prior art keywords
- data
- specific surface
- surface area
- shale
- neural network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
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
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- 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
-
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/10—Numerical modelling
-
- 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
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/30—Assessment of water resources
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Geophysics And Detection Of Objects (AREA)
Abstract
The invention relates to the technical field of petroleum and natural gas geology, in particular to a shale specific surface area parameter modeling method based on a BP neural network. The method comprises the steps of obtaining geological classification evaluation parameters, wherein the geological classification evaluation parameters comprise rock characteristics from micro scale to macro scale, establishing a mapping relation corresponding to shale specific surface area parameters by taking the geological classification evaluation parameters as training data through a BP neural network model, and finally obtaining specific surface area parameter models of shale of different types which are continuous in the longitudinal direction. The technical scheme of the invention is adopted to obtain a single-well longitudinally continuous specific surface area data curve which has the advantages of low cost, time saving, low calculated amount and results approaching to true values; by utilizing the data curve, the longitudinal specific surface area condition of the single-well shale can be evaluated; the evaluation results are very beneficial for the optimization of the reservoir with high specific surface area in the longitudinal direction, and the reservoir with high possibility of absorbing gas is obtained.
Description
Technical Field
The invention relates to the technical field of petroleum and natural gas geology, in particular to a shale specific surface area parameter modeling method based on a BP neural network.
Background
The development of shale gas gradually becomes a new place for increasing the storage and production of natural gas in China. The economic recoverable reserves become an important evaluation index for shale gas development in the shale gas development process. Along with the continuous improvement of the shale gas fracturing improvement technology level, the development degree of shale gas will be continuously deepened in the future.
Shale gas is known to be enriched primarily in the free and adsorbed states, with the free state being more readily exploited than the adsorbed state. However, the gas content proportion of the shale in the adsorption state in, for example, the shale of the Chuannan Longmaxi group is higher than 30%, so that the shale adsorption gas is strived to be developed, the shale recovery rate can be improved to a greater extent, the single well yield is improved, and meanwhile, the stable production capacity of the gas well in the future development process can also be improved by combining the adsorption characteristic of the adsorption gas and efficient development of the adsorption gas. Therefore, it is very important to accurately characterize the shale adsorption capacity.
The specific surface area reflects the sum of the surface areas of the internal space and the external space of the shale, and part of natural gas can be adsorbed on the internal surface and the external surface of the shale, so that the specific surface area parameter is an important parameter for expressing the adsorption capacity of the shale. The conventional method for acquiring the specific surface area parameter of the shale at present is an experimental measurement method. The specific surface area for evaluating the underground shale is accurate, but the method has the following disadvantages: if the method is limited by the existence of rock samples and the sampling interval (namely, a few meters of collection unit) so that the specific surface area data cannot obtain continuous numerical values, meanwhile, the data obtained through experiments need to undergo the processes of coring, sampling, sending, experiments, waiting for experimental results and the like, the time cost of shale adsorption capacity evaluation is greatly increased, the experimental determination is also a small experimental cost, the experimental cost of a single sample is estimated to be 3000 yuan, if the gas well is used for determining the specific surface area of a stratum with the depth of 100 meters, the sample interval is one meter, the estimated cost is 30 ten thousand yuan, and by using the patented technology, the data can be obtained at the cost of 0, and the density of the continuous specific surface area data is higher.
Disclosure of Invention
The invention aims to overcome the defects of discontinuous data of specific surface area parameters, long acquisition time period and the like in the prior art, and provides a shale specific surface area parameter modeling method based on a BP neural network.
In order to achieve the above object, the present invention provides the following technical solutions:
a shale specific surface area parameter modeling method based on a BP neural network comprises the steps of obtaining geological classification evaluation parameters, wherein the geological classification evaluation parameters comprise rock features from a micro scale to a macro scale, and establishing a mapping relation corresponding to shale specific surface area parameters through the BP neural network model by taking the geological classification evaluation parameters as training data to obtain specific surface area parameter models of shale of different types which are continuous in the longitudinal direction.
According to the technical scheme, macroscopic geological data and microscopic geological data are combined, and a comprehensive classification result of the data is obtained based on a BP neural network; continuously running a modeling process by using continuous accurate basic data to obtain an accurate final result; recording the pore volume and specific surface area ratio under different rock classifications in the model result, namely a related database; calculating the longitudinal pore volume of a new well in the same area by using a logging curve of the new well in the same area and combining a model; meanwhile, determining the longitudinal lithology classification of the well by adopting a BP neural network, and determining the relation value between the pore volume and the specific surface area of the lithology classification in a database so as to calculate the longitudinal specific surface area of the well; and (4) longitudinally evaluating the reservoir by utilizing longitudinal continuous data of the specific surface area. The technical scheme of the invention is adopted to obtain a single-well longitudinally continuous specific surface area data curve which has the advantages of low cost, time saving, low calculated amount and results approaching to true values; by utilizing the data curve, the longitudinal specific surface area condition of the shale in a single well can be evaluated; the evaluation results are very beneficial for the optimization of the reservoir with high specific surface area in the longitudinal direction, and the reservoir with high possibility of absorbing gas is obtained.
As a preferred technical scheme of the invention, the geological classification evaluation parameters of the macro scale comprise lithology classification results obtained by core observation, XRD rock mineral components and proportion data results obtained by laboratory analysis, and logging comprehensive curves obtained by logging a gas well; and the geological classification evaluation parameters at the microscopic scale comprise microscopic reservoir structure characteristics and two-dimensional scanning electron microscope characteristics.
The microcosmic reservoir structural characteristics specifically refer to data such as pore volume, specific surface area, pore morphological characteristics and the like obtained through a carbon dioxide adsorption experiment, a nitrogen adsorption experiment, a high-pressure mercury injection experiment and the like.
The lithology classification comprises two lithology classifications, namely primary lithology classification and secondary lithology classification, wherein the lithology classification specifically refers to the primary lithology classification and is marked as first data, namely, an experimental researcher manually observes and classifies a sampled rock core sample; according to the rough classification result of the rock core, correcting the preliminary lithology classification by matching with an XRD experiment analysis characteristic result; so that the characteristics of the rock can be more accurately described, and the characteristics mainly comprise rock composition, pore size, component size and the like.
The XRD experimental analysis characteristics refer to that the shale core sample is sampled at equal intervals and marked as second data. And (4) obtaining a mineral component result of the rock core through XRD experimental analysis for further correcting the primary lithology classification.
The logging comprehensive curve comprises natural gamma rays, sound wave time difference, neutrons, density, double lateral resistivity and logging data information of energy spectrum logging; marked as third data.
The data mainly comprises two parts, one part is depth data, and the other part is rock characteristic data in depth, such as neutron data of well logging, which is expressed that the neutron data is 0.23 under the depth of 3000 m; the logging data (third data) can be regarded as continuous data in geological evaluation because the depth interval is only 0.1m, the first data is continuous data, and the second data is discrete data.
The microcosmic reservoir structural characteristics respectively comprise pore volume, specific surface area and pore form information under the scale of micropores, mesopores and macropores, and the microcosmic reservoir structural characteristics are marked as fourth data;
sampling is carried out on a rock core, the sampling depth is consistent with the sample depth used in the XRD experiment, N samples obtained in the longitudinal direction are subjected to a scanning electron microscope experiment, scanning electron microscope photos with the representativeness of the formation micro-pore structure (the photos can reflect the formation nano-level pore structure characteristics), namely N photos are selected, and each representative photo respectively reflects the micro-pore characteristics of the sample at the depth. And (3) setting a corresponding threshold value by using Image J Image recognition software for each picture, screening out the characteristic form of the pore, and acquiring data such as the pore diameter. The obtained data content comprises diameter data, area data, the number of pores in the photo and the like of all pores in the photo, the actual area of the photo is determined and recorded as S1 by using a photo scale, and the sum of the pore area data extracted from the photo is recorded as S2, so that the face porosity is equal to S2/S1.
Through the characteristics of the scanning electron microscope, selecting a plurality of scanning electron microscope photos capable of reflecting the characteristics of the nano-level pore structure of the stratum, obtaining face porosity data through aperture size and area data, wherein the face porosity data can reflect the development degree of organic matter pores in rocks, the higher the face porosity is, the better the development degree of the organic matter pores is, sequencing the development degrees of the organic matter pores in the shale from high to low, arranging the organic matter pores from large to small according to Arabic numerals, and marking the organic matter pores as fifth data.
As a preferred embodiment of the present invention, the acquired fifth data and the acquired fourth data correspond to a depth of the rock sample used by the second data. Therefore, enough data samples can be ensured during modeling, and the accuracy of the model is improved. The fourth data and the fifth data are both discrete data.
The evaluation of whether the formation representativeness exists is to observe the characteristic morphology of the shale under the whole mirror, and under full observation, an objective concept is formed on the microscopic characteristic morphology of the sample, for example, if the organic matter pore development in the shale of the sample is very good, the diameters of organic matter pores are all larger than 400nm and belong to good pores, then the organic matter pore development of the shale under microscopic observation can be selected to be good, and the pore diameter is larger than 400nm for photographing, so that the pore characteristics in the photo can basically reflect the most pore characteristics of the sample.
As a preferred technical scheme of the invention, the method specifically comprises the following steps:
step 1, obtaining geological classification evaluation parameters;
step 2, extracting 80% of the first data, the second data, the third data, the fourth data and the fifth data obtained in the step 1 as a whole to be used as training data to be input into a BP neural network model for operation; obtaining a first result, wherein the first result is secondary lithology classification; similar to but distinct from the first data; the first data is the classification of the rock, but from the preliminary classification performed by the laboratory staff, it is an inaccurate, macroscopic classification that does not accurately express the differences on a microscopic scale. The BP neural network is utilized to connect macroscopic information and microscopic information, and the macroscopic information and the microscopic information can be combined into a classification result which shows the macroscopic difference and reflects the microscopic difference, namely the first result), so the first result is closer to the first data, and the first result is further refined on the basis of the first data;
specifically, the classification result expressed by the BP neural network is a result under pure numerical clustering, and does not have any geological significance per se, but a worker can search the data characteristics under the classification according to the BP modeling result to find the corresponding geological significance. For example, in general, the classification result of BP shows that GR and DEN curves show positive correlation, i.e. when GR curve is increased and is at high position, the value of DEN curve tends to be increased or at high position, because GR can reflect the water environment when the formation is sedimented, and when GR is increased or at high value, which means that water is calmer, more fine-grained sediment can be sedimented, and the fine-grained sediment will cause the porosity of the formation to be greatly reduced, thereby causing the DEN curve to be increased.
However, not all BP neural network model curves can be well explained, because the slight change which can be reflected by the logging curve is probably caused by the longitudinal heterogeneity of the stratum, and the slight heterogeneity can be described only in a macroscopic way and cannot be described in a very specific way. Therefore, in the BP modeling process, for the large-section lithology classification result, data 1, data 2, data 3 and the like can be used for carrying out geological logic verification to ensure the accuracy of the model, but for the difference of individual micro lithology, the difference may be caused by the combined action of macro data and micro data, the objective fact of the data should be respected, and the data are not required to be changed freely.
Step 3, arranging the lithology of the secondary lithology classification, and extracting the ratio of the pore volume to the specific surface area of different depths under the same type of lithology; taking the average value and marking the average value as a database; the database comprises lithologic names and corresponding pore volume-to-specific surface ratio numerical values; the rock characteristics under the same lithology are considered to be close in geology, so that the pore volume and specific surface area ratio corresponding to the lithology classification result of the first result is extracted, and it can be determined that under the lithology, the ratio is relatively fixed, and the ratio results of different lithologies are different greatly;
in order to obtain continuous specific surface area data, the data related to the specific surface area needs to be obtained, and the continuous data can obtain the continuous specific surface area through a certain formula;
and (3) performing equation simultaneous by using pore volume data obtained by a microscopic experiment and comprehensive logging curve data, and solving the size of each coefficient in the equation to obtain a relational expression between the comprehensive logging curve and the pore volume.
By utilizing the relational expression, the comprehensive logging curve is re-introduced, and because the comprehensive logging curve is a continuous curve, a continuous curve of the pore volume can be obtained, continuous pore volume data under the longitudinal depth can be obtained, and the continuous pore volume data is marked as sixth data;
and 5, combining the database with the sixth data to obtain continuous specific surface area data under the longitudinal depth, namely, marking the continuous shale specific surface area parameter model under the longitudinal depth as seventh data.
As a preferred technical solution of the present invention, the step 2 further comprises the steps of: and correcting the first result by using the remaining 20% of data in the first data, the second data, the third data, the fourth data and the fifth data to obtain an optimized second result.
As a preferred embodiment of the present invention, when fourth data that does not relate to depth in the database appears in the second result, the fourth data that does not relate to depth is further supplemented.
As a preferred technical solution of the present invention, the method further includes step 6 of comparing the seventh data with the fourth data, if an error between the seventh data and the fourth data is greater than 30%, further refining the first data, the second data, the third data, the fourth data, and the fifth data to obtain new data, and repeating steps 2-5 until the error between the fourth data and the seventh data is less than 30%. The seventh data can be regarded as credible and reliable and can be used as data of geological research.
Compared with the prior art, the invention has the beneficial effects that:
according to the BP neural network training result obtained by the technical scheme, pore volume and specific surface area characteristics acquired by scattered points needing experiments are converted into a longitudinally continuous specific surface area curve, and the result can guide specific surface area data of other wells in the same deposition environment, so that the experiment cost is greatly saved, a large amount of sample waiting time is saved, and great convenience is provided for prediction of high-quality reservoir layers of other wells.
Description of the drawings:
FIG. 1 is a schematic flow diagram of a modeling method of the present invention;
FIG. 2 is a schematic diagram of a longitudinal and continuous specific surface area parametric model obtained by the technical scheme of the invention.
Detailed Description
The present invention will be described in further detail with reference to test examples and specific embodiments. It should be understood that the scope of the above-described subject matter is not limited to the following examples, and any techniques implemented based on the disclosure of the present invention are within the scope of the present invention.
Example 1
The specific surface area parameter information of the shale acquired by an experimental measurement method in the prior art is limited by sampling intervals, data is discontinuous, and the data acquisition time is long. The embodiment provides a shale specific surface area parameter modeling method based on a BP neural network.
The data used by the technical scheme of the invention is from a shale core sample of a well A in a certain area in south China, and the area belongs to the same deposition environment. A core sample with a depth of 3200m-3600m is described in detail, and the overall modeling process is shown in fig. 1, specifically, an experimental researcher classifies rock samples under the longitudinal depth of the sample according to observation. Obtaining lithology classification, and marking as first data; the lithology classification herein specifically refers to preliminary lithology classification, that is, rough classification of sampled core samples by experimental researchers; the first data is labeled by the arabic numerals 1, 2, 3, 4 … M, and so on.
Further carry out the XRD experiment to the rock thin slice of the different degree of depth in the coring sample of A well, XRD experimental analysis characteristic means that carry out equidistant sampling to the shale coring sample, obtains the mineral composition result of rock core through XRD experimental analysis, can carry out further correction to first data (lithology is categorised) through the XRD result for lithology is categorised more accurately, marks as the second data.
Acquiring a logging comprehensive curve, wherein the logging comprehensive curve comprises natural gamma, acoustic time difference, neutrons, density, bilateral resistivity and logging data information of energy spectrum logging; marked as third data;
the three types of data belong to description information of core samples under a macroscopic scale.
The following is a description of core samples performed from a microscopic level as follows:
and respectively carrying out experiments such as carbon dioxide adsorption, nitrogen adsorption, high-pressure mercury intrusion analysis and the like on core samples with different depths to obtain information such as pore volume, specific surface area, pore morphology and the like under the scale of micropores, mesopores and macropores. And the sample depth and the corresponding experimental data are counted as fourth data.
Sampling is carried out on the rock core, the sampling depth is consistent with the sample depth used in the XRD and fourth data experiment, a plurality of samples obtained in the longitudinal direction are subjected to scanning electron microscope experiment, scanning electron microscope photos with the formation micro pore structure representativeness (photos capable of reflecting formation nano-level pore structure characteristics) are selected, a plurality of photos are obtained, and each representative photo respectively reflects the micro pore characteristics of the sample under the corresponding depth. And (3) setting a corresponding threshold value for a certain picture by using Image J Image recognition software, screening out the characteristic form of the pore, and acquiring the data of the pore. Marked as fifth data;
the obtained data content comprises diameter data, area data, the number of pores in the picture and other data, the actual area of the picture is determined by using a picture scale and is recorded as S1, the sum of the pore area data extracted from the picture is recorded as S2, and the face porosity is equal to S2/S1. And (4) carrying out pore morphology feature description (the information of the picture is quantized according to the size and the number of the pores to obtain a pore size numerical value and a pore number numerical value, and finally obtaining an average value of the pore size and a total number numerical value), and calculating the surface porosity. The development degrees of organic pores in the shale are ranked from high to low, and the ranking is quantified, namely, the ranking is performed from large to small through Arabic numbers.
And randomly selecting 80% of the first data, the second data, the third data, the fourth data and the fifth data as sample training data to perform BP neural network simulation under the condition of the same depth, and establishing a characteristic classification result under the condition of synthesizing each characteristic characterization parameter of the shale. I.e., the first result, (lithology classification result) this step can distinguish subtle differences in the shale, such as subtle changes in lithology or differences in porosity in the shale, the simulation results can distinguish such differences,
and (3) correcting the first result for multiple times by using the remaining 20% of the data sample, wherein the obtained first result should contain the depth points in the remaining 20% of the data, and comparing the data obtained by the experiment with the data obtained by the patented method until the difference is small to obtain the optimal solution model of the first result, namely the second result. Exporting the lithology classification result of the second result; dividing 9 types of classifications in total as shown in FIG. 2, sorting the lithology of the secondary lithology classification, and extracting the ratio of the pore volume to the specific surface area at different depths under the same type of lithology; taking the average value and marking the average value as a database; establishing a multivariate linear regression equation by combining the third data and the fourth data to obtain continuous pore volume data under the longitudinal depth, and marking the continuous pore volume data as sixth data; and combining the database with the sixth data to obtain continuous specific surface area data under the longitudinal depth, namely, a continuous shale specific surface area parameter model under the longitudinal depth and marked as seventh data.
When the depth of the sample not involved in the second data appears in the second result, the experiment should be supplemented in time to ensure that the matching of the second data and the second result is enhanced;
under an ideal state, mineral components contained in different rock types and the development conditions of internal pores are different, so that the heterogeneity of the rock is formed, and the heterogeneity in the longitudinal direction can be distinguished through superposition comparison of logging parameters, so that the continuous result obtained by model calculation, namely sixth data, can well represent macroscopic data of the longitudinal heterogeneity of the rock;
comprehensively calculating the shale classification result under the second result calculation, correlation formulas of pore volumes and specific surface areas under different types and longitudinal continuous pore volume data of the shale, so as to obtain the specific surface area data of the shale of different types which are continuous on the longitudinal scale and are judged by using macroscopic and microscopic characteristics of the shale;
and when the longitudinal specific surface area of the single well under the same structure and the same deposition environment needs to be evaluated, the longitudinal and continuous comprehensive calculation result of the specific surface area of the single well can be obtained through the second result and the comprehensive logging data of the single well. As shown in fig. 2, the curves in the first four columns are the logging comprehensive curve characteristics, and the data in the fifth and sixth columns are scatter diagrams of the pore volume and the specific surface area respectively; the eighth, ninth and tenth columns are lithology classification results obtained by continuously optimizing data. And the last column is a longitudinally continuous comprehensive calculation model of the specific surface area.
The method is suitable for areas with relatively stable deposition environments, and provides support in the development process of the whole gas field. And when the structural characteristics of the micro pores of the core test in the longitudinal direction do not exceed the threshold value, all shale types are in the first result, and the continuous calculation of the longitudinal specific surface area of the new well can be simply and quickly carried out.
When the characteristic of the gas well needing to be evaluated exceeds the threshold range of the first result, systematic experimental analysis is recommended to be carried out on the new well, namely, the step of realizing the first result in the method is completed, and therefore all databases are enlarged. In order to cope with the evaluation of the specific surface area per well required for the boundary region of a field.
From the perspective of a single well, the shale reservoir with the high specific surface area preferably selected can be used as a potential horizon of an upper reservoir of a deep shale gas well which is developed at present, and when the current development well enters a low-yield stage or faces a water flooding shut-in well, after the lower development horizon is blocked by measures, a stratum is changed to carry out excavation and submergence work on the shale reservoir with the high specific surface area. Under the condition of saving the cost and the period of longitudinal drilling, the windowing sidetrack drilling can save the process cost to a certain extent, bring more industrial productivity for a single well, improve the recovery ratio of the single well, and reduce the decrement rate of the single well yield and even the integral decrement rate of a gas field;
from the perspective of three-dimensional exploration, under the condition that the mining right area is not changed, the shale gas field can create a reserve and capacity growth position for a company and develop a brand new situation for the development of shale gas of the company.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.
Claims (7)
1. A shale specific surface area parameter modeling method based on a BP neural network is characterized by comprising the steps of obtaining
Taking geological classification evaluation parameters which comprise rocks from micro scale to macro scale
The geological classification evaluation parameters are used as training data, and a mapping relation corresponding to shale specific surface area parameters is established through a BP neural network model to obtain shale specific surface area parameter models which are continuous in the longitudinal direction and different in type; the geological classification evaluation parameters of the macroscopic scale comprise lithology classification results obtained through core observation, XRD experimental analysis characteristics and logging comprehensive curves obtained through logging of the gas wells; the geological classification evaluation parameters at the microscale comprise microscopic reservoir structural characteristics and scanning electron microscope characteristics;
the modeling method specifically comprises the following steps:
step 1, obtaining geological classification evaluation parameters;
step 2, extracting 80% of the geological classification evaluation parameters obtained in the step 1 as a whole and inputting the extracted data as training data into a BP neural network model for training; obtaining a first result, wherein the first result is secondary lithology classification;
step 3, arranging the lithology of the secondary lithology classification, and extracting the ratio of the pore volume to the specific surface area of different depths under the same type of lithology; taking the average value and marking the average value as a database;
step 4, combining the XRD experimental analysis characteristics and the microscopic reservoir structure characteristics, establishing a multiple linear regression equation, obtaining continuous pore volume data under the longitudinal depth, and marking the continuous pore volume data as sixth data;
and 5, combining the database and the sixth data to obtain continuous specific surface area data under the longitudinal depth, namely a continuous shale specific surface area parameter model under the longitudinal depth and marked as seventh data.
2. The shale specific surface area parameter modeling method based on the BP neural network as claimed in claim 1, wherein the lithology classification refers to preliminary lithology classification of a rock core sample by a worker, and is marked as first data; the XRD experimental analysis characteristic refers to that the shale core sample is sampled at equal intervals, and the mineral component result of the rock core is obtained through laboratory analysis and is marked as second data; the XRD experimental analysis characteristics are used for correcting the preliminary lithology classification result; the logging comprehensive curve comprises logging data information of natural gamma rays, acoustic time difference, neutrons, density, double lateral resistivity and energy spectrum logging, and is marked as third data.
3. The shale specific surface area parameter modeling method based on the BP neural network according to claim 2, wherein the micro reservoir structural features respectively comprise pore volume, specific surface area and pore morphology information under micropore, mesopore and macropore scales, and are marked as fourth data; and selecting a plurality of scanning electron microscope photos capable of reflecting the nano-level pore structure characteristics of the stratum according to the scanning electron microscope characteristics, obtaining face porosity data according to the aperture size and area data, sequencing the face porosity data from high to low, arranging the face porosity data from large to small according to Arabic numbers, and marking the face porosity data as fifth data.
4. The modeling method of shale specific surface area parameters based on BP neural network as claimed in claim 3, wherein the fifth data, the fourth data obtained correspond to the depth of the rock sample used by the second data.
5. The shale specific surface area parameter modeling method based on the BP neural network according to claim 4, wherein in the step 2, the method further comprises the following steps: and correcting the first result by using the remaining 20% of data in the first data, the second data, the third data, the fourth data and the fifth data to obtain an optimized second result.
6. The shale specific surface area parameter modeling method based on the BP neural network as claimed in claim 5, further comprising step 6 of comparing the seventh data with the fourth data, if the error between the seventh data and the fourth data is greater than 30%, further refining the first data, the second data, the third data, the fourth data and the fifth data to obtain new data, and repeating the steps 2-5 until the error between the fourth data and the seventh data is less than 30%.
7. An application of the shale specific surface area parameter modeling method based on the BP neural network is characterized in that when a single well in the same structure and same deposition environment needs to be subjected to longitudinal specific surface area evaluation, the longitudinal and continuous comprehensive calculation result of the specific surface area of the single well can be obtained through the second result and the comprehensive logging data of the single well.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111071183.2A CN114429075B (en) | 2021-09-13 | 2021-09-13 | Shale specific surface area parameter modeling method based on BP neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111071183.2A CN114429075B (en) | 2021-09-13 | 2021-09-13 | Shale specific surface area parameter modeling method based on BP neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114429075A CN114429075A (en) | 2022-05-03 |
| CN114429075B true CN114429075B (en) | 2022-08-26 |
Family
ID=81310097
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111071183.2A Active CN114429075B (en) | 2021-09-13 | 2021-09-13 | Shale specific surface area parameter modeling method based on BP neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114429075B (en) |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105651966A (en) * | 2016-01-18 | 2016-06-08 | 山东科技大学 | Shale oil and gas high-quality reservoir stratum evaluation method and parameter determination method |
| CN112746835A (en) * | 2019-10-31 | 2021-05-04 | 中石化石油工程技术服务有限公司 | Optimized deep shale gas geology dessert logging comprehensive evaluation method |
| CN111048163B (en) * | 2019-12-18 | 2023-03-31 | 延安大学 | Shale oil hydrocarbon retention amount (S1) evaluation method based on high-order neural network |
| CN111461386B (en) * | 2019-12-27 | 2023-08-22 | 中国地质调查局成都地质调查中心 | Shale gas dessert prediction method based on BP neural network |
| US11624853B2 (en) * | 2020-01-31 | 2023-04-11 | ExxonMobil Technology and Engineering Company | Methods for performing formation evaluation and related systems |
| CN111425190B (en) * | 2020-03-19 | 2022-05-17 | 中国石油大学(华东) | Shale gas formation lithology identification method, system, storage medium and terminal |
-
2021
- 2021-09-13 CN CN202111071183.2A patent/CN114429075B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN114429075A (en) | 2022-05-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109061765B (en) | Trap evaluation method for heterogeneous thin sandstone interbed reservoir | |
| CN107703560B (en) | shale lithofacies fine identification method based on triple information | |
| CN110644980B (en) | Comprehensive classification evaluation method for ultra-low permeability oil reservoir | |
| CN104747185B (en) | Heterogeneous reservoir reservoir synthetical assortment evaluation method | |
| CN103590827B (en) | Based on the compact clastic rock natural gas well PRODUCTION FORECASTING METHODS of Reservoir Classification | |
| CN103670384B (en) | A kind of Reservoir Classification method and system | |
| CN102297828B (en) | Reservoir pore structure classification method based on nuclear magnetic three-component percentage | |
| RU2475646C1 (en) | Method of construction of geologic and hydrodynamic model of oil and gas deposits | |
| Zhu et al. | Rapid identification of high-quality marine shale gas reservoirs based on the oversampling method and random forest algorithm | |
| CN104047598A (en) | Heterogeneous paleo-karst carbonate reservoir productivity prediction method | |
| CN103616731A (en) | Method and device for determining altered volcanic rock effective reservoir in oil and gas exploration | |
| CN111429012A (en) | Shale brittle dessert evaluation method | |
| CN113033648A (en) | Method for realizing logging interpretation by using machine learning algorithm | |
| CN110208874B (en) | Method for identifying effective reservoir of tight sandstone reservoir | |
| CN110656922B (en) | Shale isochronous stratum logging dividing method and system based on pencils and stone belt characteristics | |
| CN117251802B (en) | Heterogeneous reservoir parameter prediction method and system based on transfer learning | |
| CN114114414A (en) | Artificial intelligence prediction method for 'dessert' information of shale reservoir | |
| CN116263809A (en) | Shale oil lithology recognition method based on artificial intelligence and logging layering modeling | |
| CN114429075B (en) | Shale specific surface area parameter modeling method based on BP neural network | |
| CN109738955B (en) | Metamorphic rock lithology comprehensive judgment method based on component-structure classification | |
| CN111580183A (en) | Method for carrying out deep quantitative reduction on ancient lake water | |
| CN103698494B (en) | Method and device for determining saturation degree of hydrocarbon in lithologic trap | |
| CN104834934A (en) | Nuclear body capturing method used for identifying reservoir fluid | |
| CN111155980A (en) | Water flow dominant channel identification method and device | |
| FANG et al. | Quantitative Method of Classification and Discrimination of a Porous Carbonate Reservoir Integrating K‐means Clustering and Bayesian Theory |
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 |