CN116047590A - Method and device for building earthquake phase prediction model and method and device for predicting earthquake phase - Google Patents
Method and device for building earthquake phase prediction model and method and device for predicting earthquake phase Download PDFInfo
- Publication number
- CN116047590A CN116047590A CN202111246409.8A CN202111246409A CN116047590A CN 116047590 A CN116047590 A CN 116047590A CN 202111246409 A CN202111246409 A CN 202111246409A CN 116047590 A CN116047590 A CN 116047590A
- Authority
- CN
- China
- Prior art keywords
- seismic
- seismic attribute
- attribute data
- sample set
- plane
- 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
- 238000000034 method Methods 0.000 title claims abstract description 88
- 238000012549 training Methods 0.000 claims abstract description 86
- 238000013145 classification model Methods 0.000 claims abstract description 77
- 238000012216 screening Methods 0.000 claims abstract description 18
- 208000035126 Facies Diseases 0.000 claims description 18
- 230000011218 segmentation Effects 0.000 claims description 16
- 230000008021 deposition Effects 0.000 claims description 14
- 238000009826 distribution Methods 0.000 claims description 10
- 238000003066 decision tree Methods 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 8
- 238000007477 logistic regression Methods 0.000 claims description 7
- 238000007621 cluster analysis Methods 0.000 claims 1
- 239000003550 marker Substances 0.000 abstract 1
- 238000004422 calculation algorithm Methods 0.000 description 40
- 238000004458 analytical method Methods 0.000 description 12
- 238000000151 deposition Methods 0.000 description 11
- 239000011148 porous material Substances 0.000 description 9
- 230000000694 effects Effects 0.000 description 8
- 230000008569 process Effects 0.000 description 6
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 6
- 230000008859 change Effects 0.000 description 4
- 238000007635 classification algorithm Methods 0.000 description 4
- 238000011161 development Methods 0.000 description 4
- 230000018109 developmental process Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 238000005553 drilling Methods 0.000 description 4
- 230000003628 erosive effect Effects 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 239000002245 particle Substances 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 235000019738 Limestone Nutrition 0.000 description 3
- 239000006028 limestone Substances 0.000 description 3
- 230000004044 response Effects 0.000 description 3
- 238000012795 verification Methods 0.000 description 3
- BVKZGUZCCUSVTD-UHFFFAOYSA-L Carbonate Chemical compound [O-]C([O-])=O BVKZGUZCCUSVTD-UHFFFAOYSA-L 0.000 description 2
- 238000013459 approach Methods 0.000 description 2
- 238000005266 casting Methods 0.000 description 2
- 238000011010 flushing procedure Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000003909 pattern recognition Methods 0.000 description 2
- 238000011158 quantitative evaluation Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 208000019888 Circadian rhythm sleep disease Diseases 0.000 description 1
- 238000000018 DNA microarray Methods 0.000 description 1
- 241001055367 Dario Species 0.000 description 1
- 208000001456 Jet Lag Syndrome Diseases 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 244000052616 bacterial pathogen Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000019771 cognition Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000004090 dissolution Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 239000010438 granite Substances 0.000 description 1
- 208000033915 jet lag type circadian rhythm sleep disease Diseases 0.000 description 1
- 238000003064 k means clustering Methods 0.000 description 1
- 230000007786 learning performance Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000000513 principal component analysis Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000011002 quantification Methods 0.000 description 1
- 239000013049 sediment Substances 0.000 description 1
- 230000035939 shock Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V1/00—Seismology; Seismic or acoustic prospecting or detecting
- G01V1/28—Processing seismic data, e.g. for interpretation or for event detection
- G01V1/30—Analysis
-
- 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
- Y02A10/00—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE at coastal zones; at river basins
- Y02A10/40—Controlling or monitoring, e.g. of flood or hurricane; Forecasting, e.g. risk assessment or mapping
Landscapes
- Engineering & Computer Science (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Acoustics & Sound (AREA)
- Environmental & Geological Engineering (AREA)
- Geology (AREA)
- General Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Geophysics (AREA)
- Geophysics And Detection Of Objects (AREA)
Abstract
The invention discloses a method and a device for building a seismic phase prediction model and predicting a seismic phase. The method for establishing the earthquake phase prediction model comprises the steps of obtaining screened multiple first-class plane earthquake attribute data and multiple second-class plane earthquake attribute data which are meshed according to the same rule; screening unlabeled grids from the plane seismic attribute data according to a set rule, and extracting first-class seismic attribute values and second-class seismic attribute values of the labeled grids and the screened unlabeled grids; obtaining a first marked sample set, a first unmarked sample set, a second marked sample set and a second unmarked sample set; and training the selected classification model by utilizing the obtained sample sets through a collaborative training method to obtain the classification model for earthquake phase prediction. The accuracy of prediction of the seismic phase can be improved and the dependence on the marker samples reduced.
Description
Technical Field
The invention relates to the technical field of oil and gas exploration and development, in particular to a method and a device for building a seismic phase prediction model and predicting a seismic phase.
Background
With the popularization of 3D earthquake and the progress of geophysical processing interpretation technology, the accuracy of subsurface geologic pattern cognition can be improved by utilizing earthquake attribute data to carry out geologic interpretation analysis, and the uncertainty of reservoir prediction is reduced. However, in geologic interpretation using seismic attribute data, quantification is difficult to achieve due to the multiple resolvability of seismic attributes and the artificial subjective factors. To address this problem, many experts propose supervised learning and unsupervised learning classification algorithms to address this problem.
In recent years, bacilli et al (2013) have adopted a classification algorithm based on differential analysis to analyze seismic phases; majid bageri & Mohammad Ali Riahi (2014) uses classification algorithms such as multi-layer perceptron, support vector machine, fischer, barsen and nearest neighbor algorithm to analyze the seismic phase; dario Grana (2017) uses Bayesian and prospect-mapping methods for seismic facies analysis; h.sabeti (2009) & Ivan S nchez Galvis (2017) performs seismic phase analysis by using a k-means clustering algorithm; fabio Radomille Santan & Arthur Ayres Neto (2014) uses principal component analysis, system tree and k-means algorithm to perform seismic phase analysis based on seismic attribute data, and further perform depositional phase classification; the Zhao et al (2015) comprehensively analyzes the k-means algorithm, self-organizing map, generating topology map algorithm, support vector machine, gaussian mixture model and artificial neural network to conduct seismic phase analysis, proposes to determine initial estimation of verification classification category by using unsupervised learning, and then classifies by using supervised learning.
However, the above method often has some unavoidable problems when the conventional supervised learning algorithm or the unsupervised learning algorithm is used for analyzing the seismic facies. The supervised learning algorithm uses only a small amount of marked data for training, but ignores a large amount of unmarked sample data, and needs to have enough marked samples to guarantee generalization of the training model. In the exploration and development of oil and gas fields, the number of well drilling is often small, so that the seismic phase analysis and prediction are carried out by using a supervised learning method, and the problem of poor generalization capability generally exists. The unsupervised learning algorithm uses full sample data for training, highlights the global characteristics of the seismic attributes, but the clustering decision boundary is not limited by well points, and the problem of low matching rate of clustering results and well points generally exists, so that the reservoir prediction based on the seismic attributes by using the clustering algorithm is difficult to obtain a geological interpretation result with high matching rate with the well point data.
Therefore, the conventional supervised learning algorithm or the unsupervised learning algorithm is used for seismic facies analysis in the prior art, and the problems of limited applicable conditions or low accuracy of prediction results exist.
Disclosure of Invention
Miller and Uyar (1997) have shown, through theoretical derivation, that training a classifier with unlabeled samples can achieve the effect of improving classification performance if a correlation can be established between target and unlabeled sample distribution, thus yielding a semi-supervised learning approach. Semi-supervised learning methods, i.e. co-training algorithms, were originally proposed by Blum and Mitchell (1998). Generally, co-training algorithms assume that the feature set of the data can be divided into two subsets, and training on each subset results in a good classifier. The classifiers on the two feature subsets complete classification work in a mutual learning mode: and retraining by using the classification results of the other side until the two classifiers have the same classification results for most of data.
In view of the foregoing, the present invention has been made to provide a seismic phase prediction model creation, seismic phase prediction method and apparatus that overcomes or at least partially solves the foregoing problems, and that can improve the accuracy of prediction of seismic phases and reduce the reliance on labeled samples.
In a first aspect, an embodiment of the present invention provides a method for building a seismic phase prediction model, including:
a data acquisition step: acquiring screened multiple first-class plane seismic attribute data and multiple second-class plane seismic attribute data which are gridded according to the same rule, wherein grids matched with the position information of the well in each plane seismic attribute data carry out seismic phase marking according to the destination layer deposition microphase type of the well;
a sample set establishing step: screening unlabeled grids from the plane seismic attribute data according to a set rule, and extracting first-class seismic attribute values and second-class seismic attribute values of the labeled grids and the screened unlabeled grids; the seismic phase marks of the marking grid and the seismic attribute values of the first class are used as a marking sample to obtain a first marking sample set; taking each seismic attribute value of the first type of unlabeled grid as an unlabeled sample to obtain a first unlabeled sample set; the seismic phase marks of the marking grid and the second type of seismic attribute values are used as a marking sample, and a second marking sample set is obtained; taking each seismic attribute value of the second type of unlabeled grid as an unlabeled sample to obtain a second unlabeled sample set;
Model training: and training the selected classification model by using the first marked sample set, the first unmarked sample set, the second marked sample set and the second unmarked sample set through a co-training method to obtain the classification model for seismic phase prediction.
In a second aspect, an embodiment of the present invention provides a method for predicting a seismic phase, including:
and inputting various types of seismic attribute data into a classification model, and carrying out seismic phase prediction according to an output result, wherein the classification model is obtained by using the seismic phase prediction model building method, and the types are consistent with the types of plane seismic attribute data screened during the classification model building.
In a third aspect, an embodiment of the present invention provides a device for establishing a seismic phase prediction model, including:
the data acquisition module is used for acquiring the screened multiple first-class plane seismic attribute data and multiple second-class plane seismic attribute data which are meshed according to the same rule, and grids matched with the position information of the well in the plane seismic attribute data carry out seismic phase marking according to the destination layer deposition microphase type of the well;
the sample set establishing module is used for screening unlabeled grids from the plane seismic attribute data according to a set rule, and extracting first-class seismic attribute values and second-class seismic attribute values of the labeled grids and the screened unlabeled grids; the seismic phase marks of the marking grid and the seismic attribute values of the first class are used as a marking sample to obtain a first marking sample set; taking each seismic attribute value of the first type of unlabeled grid as an unlabeled sample to obtain a first unlabeled sample set; the seismic phase marks of the marking grid and the second type of seismic attribute values are used as a marking sample, and a second marking sample set is obtained; taking each seismic attribute value of the second type of unlabeled grid as an unlabeled sample to obtain a second unlabeled sample set;
And the model training module is used for training the selected classification model by utilizing the first marked sample set, the first unmarked sample set, the second marked sample set and the second unmarked sample set through a collaborative training method so as to obtain the classification model for the seismic phase prediction.
In a fourth aspect, an embodiment of the present invention provides a computer program product with a function of predicting a seismic phase, including a computer program/instruction, where the computer program/instruction when executed by a processor implements the foregoing seismic phase prediction model establishment, or implements the foregoing seismic phase prediction method.
The technical scheme provided by the embodiment of the invention has the beneficial effects that at least:
(1) According to the seismic phase prediction model establishing method provided by the embodiment of the invention, the screened multiple first-class plane seismic attribute data and multiple second-class plane seismic attribute data which are meshed according to the same rule are obtained; screening unlabeled grids from the plane seismic attribute data according to a set rule, and extracting first-class seismic attribute values and second-class seismic attribute values of the labeled grids and the screened unlabeled grids; obtaining a first marked sample set, a first unmarked sample set, a second marked sample set and a second unmarked sample set; and training the selected classification model by utilizing the obtained sample sets through a collaborative training method to obtain the classification model for earthquake phase prediction. The earthquake and well data are utilized to carry out training of the classification model by adopting a cooperative training semi-supervised learning algorithm, so that the positive judgment rate of the classification model is improved; meanwhile, the dependence on a marked sample is reduced, and the application range of the sample is widened.
(2) According to the earthquake phase prediction method provided by the embodiment of the invention, the earthquake phase identification noise interference is reduced through the collaborative training semi-supervised learning algorithm, the continuity of plane distribution of earthquake phase prediction results is enhanced, and the effect of carrying out quantitative evaluation on sedimentary microphase by utilizing earthquake attributes is obviously improved.
The technical scheme of the invention is further described in detail through the drawings and the embodiments.
Drawings
The accompanying drawings are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate the invention and together with the embodiments of the invention, serve to explain the invention. In the drawings:
FIG. 1 is a flowchart of a method for establishing a seismic phase prediction model according to a first embodiment of the present invention;
FIG. 2 is a flowchart showing the implementation of step S11 in FIG. 1;
FIG. 3 is a flowchart showing a specific implementation of step S13 in FIG. 1;
FIG. 4 is a flowchart of a method for predicting a seismic phase according to a second embodiment of the invention;
FIG. 5-a is a schematic view of a tidal water course microphase cast body sheet of developing granitic limestone;
FIG. 5-b is a schematic illustration of a thin slab of a weak eroding lagoon microphase cast;
FIG. 5-c is a schematic illustration of a lagoon microphase cast sheet developing mudstone, cementing;
FIG. 5-d is another schematic illustration of a lagoon microphase cast sheet developing mudstone, cementing;
FIG. 6-a is a core schematic of a developmental interlacing layer;
FIG. 6-b is a core schematic of developing biological debris particles;
FIG. 6-c is a core schematic of developing a flushing face;
FIG. 6-d is a core schematic of a developing biological wormhole construction;
FIG. 7-a is a schematic illustration of a tidal waterway microphase log;
FIG. 7-b is a schematic diagram of a lagoon microphase log;
FIG. 8-a is a graph of total positive rate change for seismic phase identification;
FIG. 8-b is a graph of the variation of the positive rate of judgment of tidal waterways and lagoon earthquakes;
fig. 9 is a schematic structural diagram of a seismic phase prediction model building device according to an embodiment of the present invention.
Detailed Description
Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
In order to solve the problems of strong dependence of earthquake phase prediction on data or low accuracy of prediction results in the prior art, the embodiment of the invention provides a method and a device for establishing an earthquake phase prediction model and predicting earthquake phases, which can improve the prediction accuracy of earthquake phases and reduce the dependence on marked samples.
Example 1
The first embodiment of the invention provides a method for establishing a seismic phase prediction model, the flow of which is shown in fig. 1, comprising the following steps:
step S11: and (3) data acquisition: and acquiring the screened multiple first-class plane seismic attribute data and multiple second-class plane seismic attribute data which are meshed according to the same rule.
The grids matched with the position information of the wells in the plane seismic attribute data are marked with the seismic phases according to the type of the sedimentary microphase of the destination layer of the wells.
The specific method for acquiring the seismic attribute data of each plane, as shown in fig. 2, may include the following steps:
step S111: and acquiring multiple types of plane seismic attribute data at a target layer of the seismic data volume, and gridding the plane seismic attribute data according to the same rule.
In some embodiments, the extraction of the planar seismic attribute data may include determining a destination layer derived from a target seismic event zero-phase wrap constraint tracking interpretation; and converting the seismic data volume into a seismic attribute data volume of a set type, and extracting the seismic attribute at the target layer from the seismic attribute data volume to obtain the plane seismic attribute data of the type.
According to the method for extracting the layer-controlled seismic attribute, the target seismic horizon is isolated from other seismic in-phase axes through zero-phase wrap constraint tracking interpretation of the target seismic in-phase axes, so that the seismic attribute is extracted. Compared with the traditional method for extracting slice seismic attributes by tracking peaks and troughs, the method for extracting the layer-control seismic attributes weakens the interference of information between stratum interfaces of different periods, ensures the accuracy of the seismic attributes, and can acquire global seismic attribute characteristics of the same phase axis of the earthquake instead of local seismic slice attribute characteristics.
Step S112: and determining grids matched with the well positions in the plane seismic attribute data according to the well position information, and marking the grids by seismic facies according to the well target layer deposition micro-facies type.
For example, if the deposited microphase type of a well includes both a waterway and a non-waterway, the grid at the waterway may be marked 1 and the grid at the non-waterway may be marked 0.
Step S113: and screening plane seismic attribute data with the seismic attribute distribution and the seismic coincidence degree of the marks meeting the set requirements.
In some embodiments, the method may include, for each type of acquired planar seismic attribute data, determining a plurality of water channel segmentation values according to the seismic attribute values thereof, predicting a seismic phase type of each grid contained in each water channel segmentation value, determining a total positive judgment rate of the predicted seismic phase type according to the marked seismic phases, and obtaining a plurality of total positive judgment rates; and screening the plane seismic attribute data with the maximum total positive judgment rate larger than the set total positive judgment rate threshold value.
In some embodiments, the method may further include, for each type of obtained planar seismic attribute data, determining a plurality of water channel segmentation values according to the seismic attribute values thereof, predicting the seismic phase type of each grid contained in each water channel segmentation value, and determining a total positive judgment rate and a water channel positive judgment rate of the predicted seismic phase type according to the marked seismic phases to obtain a plurality of total positive judgment rates and water channel positive judgment rates; and screening the plane seismic attribute data with the maximum total positive judgment rate larger than the set total positive judgment rate threshold value and the maximum water channel positive judgment rate larger than the set water channel positive judgment rate threshold value.
The screening of the plane seismic attribute data not only considers the total positive judgment rate, but also considers the positive judgment rate of the water channel, so that the screened plane seismic attribute data can better reflect the distribution of the water channel, and the water channel prediction capability of the finally obtained classification model is better.
Step S114: and clustering the screened multiple plane seismic attribute data, and dividing the data into a first class and a second class.
In some embodiments, determining correlation coefficients between the screened multiple planar seismic attribute data by hierarchical clustering operation; and the screened multiple plane seismic attribute data are gathered into two types according to the correlation, so that multiple first plane seismic attribute data and multiple second plane seismic attribute data are obtained.
Step S12: sample set establishment: screening unlabeled grids from the plane seismic attribute data according to a set rule, and extracting first-class seismic attribute values and second-class seismic attribute values of the labeled grids and the screened unlabeled grids; the seismic phase marks of the marking grid and the seismic attribute values of the first class are used as a marking sample to obtain a first marking sample set; taking each seismic attribute value of the first type of unlabeled grid as an unlabeled sample to obtain a first unlabeled sample set; the seismic phase marks of the marking grid and the second type of seismic attribute values are used as a marking sample, and a second marking sample set is obtained; and taking each seismic attribute value of the second type of unlabeled grid as an unlabeled sample to obtain a second unlabeled sample set.
Step S13: model training: training the selected classification model by a co-training method using the first labeled sample set, the first unlabeled sample set, the second labeled sample set, and the second unlabeled sample set to obtain a classification model for seismic facies prediction.
In the screening process of the plane seismic attribute data, the water channel segmentation value corresponding to the maximum total positive judgment rate of the screened plane seismic attribute data can be recorded simultaneously and used as the optimal water channel segmentation value.
The implementation of step S13 may further include training the selected classification model by a co-training method using the screened optimal water channel segmentation value, the first labeled sample set, the first unlabeled sample set, the second labeled sample set, and the second unlabeled sample set of the planar seismic attribute data.
The selected classification model may be one of the following:
a logistic regression classification model, a decision tree classification model, and a nearest neighbor classification model.
The training process of the specific model, as shown in fig. 3, may include the following steps:
step S131: training the selected classification model by using the first marked sample set to obtain a first classification model, and training the selected classification model by using the second marked sample set to obtain a second classification model.
Specifically, the classification model trained using the first labeled sample set and the classification model trained using the second labeled sample set may be the same classification model or different classification models.
Step S132: a set number of first unlabeled samples are selected from the current first unlabeled sample set, and a set number of grid-consistent second unlabeled samples are selected from the current second unlabeled sample set.
The selected identification may be set for the selected first unlabeled exemplar in a randomly selected manner, e.g., when a set number of first unlabeled exemplars are selected from the current first unlabeled exemplar set; each time a selection is made from only unlabeled samples that do not have a selected identification.
Step S133: and inputting the selected first unlabeled sample into the current first classification model, marking the selected second unlabeled sample according to the output result, and adding the second unlabeled sample into the current second marked sample set.
Since the grids of the selected first unlabeled exemplars and the second unlabeled exemplars are identical, the samples with identical grids in the selected second unlabeled exemplars can be labeled according to the output result.
Step S134: and inputting the selected second unlabeled sample into the current second classification model, marking the selected first unlabeled sample according to the output result, and adding the first unlabeled sample into the current first marked sample set.
Step S133 and step S134 may be performed either or both of them.
Step S135: the current first classification model is trained using the current first set of labeled samples, and the current second classification model is trained using the current second set of labeled samples.
Step S136: and judging whether the predictive fitness of the current first classification model and the current second classification model meets the set requirement or not.
If yes, step S136 is executed to finish training the model; otherwise, the process returns to step S132.
Step S137: and (5) finishing training of the model.
According to the seismic phase prediction model establishing method provided by the embodiment of the invention, the screened multiple first-class plane seismic attribute data and multiple second-class plane seismic attribute data which are meshed according to the same rule are obtained; screening unlabeled grids from the plane seismic attribute data according to a set rule, and extracting first-class seismic attribute values and second-class seismic attribute values of the labeled grids and the screened unlabeled grids; obtaining a first marked sample set, a first unmarked sample set, a second marked sample set and a second unmarked sample set; and training the selected classification model by utilizing the obtained sample sets through a collaborative training method to obtain the classification model for earthquake phase prediction. The earthquake and well data are utilized to carry out training of the classification model by adopting a cooperative training semi-supervised learning algorithm, so that the positive judgment rate of the classification model is improved; meanwhile, the dependence on a marked sample is reduced, and the application range of the sample is widened.
Based on the inventive concept, the embodiment of the invention also provides a seismic phase prediction method, which comprises the following steps:
and inputting the seismic attribute data of various types into the classification model, and carrying out seismic phase prediction according to the output result. The classification model is obtained by using the method for establishing the earthquake phase prediction model, and the type is consistent with the type of the plane earthquake attribute data screened during the establishment of the classification model.
According to the earthquake phase prediction method, the earthquake phase identification noise interference is reduced through the collaborative training semi-supervised learning algorithm, the continuity of plane distribution of earthquake phase prediction results is enhanced, and the effect of carrying out quantitative evaluation on sedimentary micro-phases by utilizing earthquake attributes is obviously improved.
In some embodiments, the method may include inputting seismic attribute data consistent with a type of planar seismic attribute data of a first type into a first classification model to obtain a first output result; inputting the seismic attribute data consistent with the type of the seismic attribute data of the second class plane into a second classification model to obtain a second output result; and carrying out earthquake phase prediction according to the first output result and the second output result.
The seismic phase prediction may be performed only based on the first output result, may be performed only based on the second output result, or may be performed by referring to both the first output result and the second output result.
Example two
The second embodiment of the present invention provides a specific implementation flow of a seismic facies prediction method, taking a target layer of a research area as a carbonate reservoir layer as an example, where the target layer mainly develops two sedimentary microphases of a tidal water channel and a lagoon, and the specific seismic facies prediction is a process for identifying the sedimentary microphases, as shown in fig. 4, and includes the following steps:
the data preprocessing step includes the following steps S41 to S43:
step S41: and calibrating a logging curve by using a small amount of sheet and core data to identify lithology and sedimentary microphases, and identifying the logging phases as tag data.
The tidal water channel is micro-phase with the lagoon, the tidal water channel is mainly composed of granular limestone, the sorting rounding is good, the biological debris crushing degree is high, and the strong hydrodynamic deposition is reflected. Erosion is dominant in tidal waterway reservoirs, and the pore types are mainly inter-granular pores, and secondary pores such as intra-granular pores, casting pores and the like are developed (see figure 5-a). The core is seen to have a sedimentary configuration of staggered layering (fig. 6-a), bio-chip particles (fig. 6-b), tidal channel flushing face (fig. 6-c), etc. Tidal watercourse reservoirs exhibit low GR, high ILD, high AC and low density log response characteristics on log (fig. 7-a). The lagoon reservoirs were dominated by argillite particles (FIG. 5-b), granulite mudstones (FIG. 5-c) and mudstones (FIG. 5-d), poor sort rounding, high stucco content, reflecting a weak hydrodynamic deposition environment. Lagoon reservoirs are dominated by weak erosion and cementation and pore types are dominated by intergranular dissolution pores, cast mold pores and intergranular pores. Obvious biological wormholes are visible in the core and the biological perturbation effect is obvious (FIG. 6-d). Lagoon reservoirs exhibited high GR, low resistance and low sonic jet lag high density log response characteristics (fig. 7-b).
FIGS. 5-a-5-d are tidal water course and lagoon casting sheet diagrams: in FIG. 5-a, it is seen that the granular limestone is strongly eroded, reflecting the tidal channel microphase; in FIG. 5-b, the argillite granites are seen, weakly eroding, reacting the lagoon microphase; in FIG. 5-c, it can be seen that the granular mudstone, cementation, reacts to the lagoon microphase; in FIG. 5-d, mudstone, cementation, reaction to lagoon microphases is seen. FIGS. 6 a-6-d are diagrams of tidal waterway and lagoon microphase core phase markers: the staggered layer is visible in FIG. 6-a, reflecting the tidal channel microphase; biological debris particles can be seen in FIG. 6-b, reflecting tidal waterway microphases; the flush face, tidal waterway, is visible in fig. 6-c; the biological worm hole configuration can be seen in FIG. 6-d.
And (3) completing identification of the logging phases according to the logging response characteristics of the tidal water channel and the lagoon reservoir.
Step S42: and extracting the plane seismic attribute data by using a zero-phase wrapping method.
In order to acquire the seismic attribute information of the target seismic event, the target seismic horizon is isolated from other seismic event by tracking and interpretation of zero-phase wrapping constraint of the target seismic event, so that the seismic attribute is extracted. Compared with the traditional method for extracting slice seismic attributes by tracking peaks and troughs, the method for extracting the layer-control seismic attributes weakens the interference of information between stratum interfaces of different periods, ensures the accuracy of the seismic attributes, and can acquire global seismic attribute characteristics of the same phase axis of the earthquake instead of local seismic slice attribute characteristics.
The 19 planar seismic attribute data of the destination layer are co-extracted, including Maximum amplitude, sum of amplitudes, mean amplitude, trace AGC, RMS amplitude, average energy, sum of magnitudes, most of, median, average magnitude, maximum magnitude, sum of positive amplitudes, structure sample, sweeteen, average positive amplitude, arcength, minimum amplitude, interval Average Arithmetic, and Extract value.
Step S43: and calibrating the well logging data to the plane seismic attribute data according to the geodetic coordinates, wherein the result of well logging phase identification of the well points is used as tag data, and the data without the well points is used as unlabeled data.
The research area shares 257 drilling data, links according to longitude and latitude coordinates of well point data and seismic attributes, and annotates labels and unlabeled data, wherein seismic attribute grid data of well point development tidal water channel deposition units are marked as 1, seismic attribute grid data of non-development tidal water channel deposition units are marked as 0, and no well point matching seismic attribute grid data are marked as-1. The study divides 257 drilling data in a study area into a training set and a verification set, randomly selects 70% of wells as the training set for model training, and 30% of wells as the verification set for model checking. The seismic attribute grid point data is used as data to be predicted.
The plane seismic attribute data optimization and grouping step includes the following steps S44 and S45:
because the collaborative training requires two groups of views X1 and X2, and the optimization and grouping of the seismic attributes are key factors for determining the analysis effect of the seismic facies, the result adopts a sliding window method to count the coincidence rate between well points and the seismic attributes, and adopts a hierarchical clustering algorithm based on a correlation coefficient to perform clustering analysis of the seismic attributes aiming at the seismic attributes with higher coincidence rate, so that the two views X1 and X2 are constructed.
Step S44: planar seismic attribute data is preferred.
The different earthquake attributes have different distinguishing capability on earthquake phases, and the embodiment adopts a sliding window method to carry out statistical analysis on the well earthquake coincidence rate aiming at the two classification problems of the tidal water channel and the lagoon. Firstly, uniformly dividing single-plane seismic attribute data into N equal parts (N can be 50), dividing tidal water micro-phases and lagoon micro-phases point by point, calculating classification positive judgment rates, and determining the coincidence rate of seismic attributes and well point data according to the positive judgment rates and the change curve of seismic attribute dividing points. The maximum value of the total positive judgment rate is the matching rate of the seismic attribute and well point data, and the seismic attribute value corresponding to the maximum value of the total positive judgment rate is the optimal segmentation value (figure 8 a) used as the seismic attribute seismic phase analysis and operation. FIG. 8 is a graph of the Maximamphite seismic attribute sliding window method well shock compliance statistics: FIG. 8-a is a graph of total positive rate change for seismic phase identification; FIG. 8-b contains a plot of the positive rate of change of tidal water course versus lagoon seismic facies. Curve a in fig. 8B is the variation curve of the positive determination rate of the tidal water course and the seismic attribute division point, and curve B is the variation curve of the positive determination rate of the lagoon and the seismic attribute division point. The coincidence rate of the seismic attribute and the well point deposition is shown in table 1, and the seismic attribute with the total positive judgment rate being more than 80% and the tidal water channel identification positive judgment rate being more than 30% is preferably used as the characteristic parameter data of the seismic phase pattern identification in the embodiment.
TABLE 1 statistics of the seismic Properties and well Point depositions
Step S45: and carrying out plane seismic attribute data grouping by a hierarchical clustering algorithm based on the correlation coefficient.
The hierarchical clustering method sequentially forms a clustering tree by taking a certain similarity or dissimilarity coefficient as an index. Hierarchical clustering is an important method for establishing a data structure among data and mining the correlation relationship among different data points or data clusters. The conventional hierarchical clustering method takes the distance between data points or data clusters as a parameter index of clustering, and two points or two data clusters with the smallest distance are clustered together in a priority mode, and clustering is carried out sequentially until the condition is ended. In the embodiment, the correlation coefficient between the plane seismic attribute data is used as a parameter index of clustering instead of the distance to perform hierarchical clustering operation, and the plane seismic attribute data with large correlation coefficient are clustered together preferentially.
According to a hierarchical clustering algorithm based on correlation coefficients, the extracted planar seismic attribute data can be divided into two large groups according to the correlation coefficients, and the first group of planar seismic attribute data comprises: maximum amplitude, sum of amplitudes, mean amplite, sum of positiveamplitudes, average positiveamplitude, minimum amplite, most of and Median; the second set of planar seismic attribute data includes: trace AGC, RMS amplitude, average energy, sum of magnitudes, average magnitude, maximummagnitude, sweeteen, arclength, interval Average Arithmetic, and Extract value. The two sets of planar seismic attribute data can be used as two sets of characteristic parameters for analyzing and identifying tidal water channel and lagoon seismic facies for collaborative training.
The method comprises the steps of training a classification model by a cooperative training semi-supervised learning algorithm and intelligently identifying a tidal water channel, wherein the specific steps are as follows:
step S46: and training a semi-supervised learning algorithm classification model through collaborative training.
Co-training (Co-training) is an important paradigm in semi-supervised learning, which uses dual view training of two classifiers to label samples against each other to expand the training set, thereby improving learning performance with unlabeled samples.
Basic principle of cooperative training algorithm:
assuming that there are two sufficiently redundant views 1 and 2 of the dataset properties, set to X1 and X2, one example can be represented as (X1, X2), where X1 is the eigenvector of X in the X1 view and X2 is the eigenvector of X in the X2 view. Assuming that f is an objective function in the example space X, f (X) =f1 (X1) =f2 (X2) =l should be given by the label of X as l. Blum and t.mitchell define a so-called "compatibility", i.e. for a certain distribution D on X, C1 and C2 are conceptual classes defined on X1 and X2, respectively, an objective function f= (f 1, f 2) ec1×c2 is said to be "compatible" with D if D assigns zero probabilities to an instance (X1, X2) that satisfies f1 (X) noteqf 2 (X2).
The specific calculation flow of the cooperative training semi-supervised learning algorithm is as follows:
1) Training a classifier h1 by using the X1 part of L;
2) Training a classifier h2 by using the X2 part of L;
3) Marking all elements in U 'by h1, marking U' by using the marking result of U ', selecting p positive marks and n negative marks from U', and putting the positive marks and the n negative marks into X2;
4) Marking all elements in the U ', marking U' by using the marking result of the U ', selecting p positive marks and n negative marks from the U', and putting the p positive marks and the n negative marks into X1;
5) Training a classifier h1 with X1; training a classifier h2 with X2;
6) N data are selected from U1 randomly to be supplemented into U ', and N data are selected from U2 randomly to be supplemented into U'.
And (3) circulating the iteration 3) -6) until the training is finished.
Step S47: the classification model is verified using the verification set.
In the second embodiment, three classical classification algorithms, namely logistic regression, decision tree and nearest neighbor algorithm, are adopted as classifiers for collaborative training semi-supervised learning to perform pattern recognition on tidal water channels and lagoon microphases. Referring to tables 2 and 3, pattern recognition is performed on two groups of seismic attributes by using logistic regression, decision tree and nearest neighbor algorithm to find X1 and X2 groups of seismic attribute supervision learning results, the prediction results have obvious difference, and have higher positive judgment rate, so that the conditions that each view learning of collaborative training is mutually referred and the obtained classifier has certain accuracy are satisfied.
TABLE 2 statistics of positive judgment rate of seismic facies identification for X1 and X2 seismic attribute supervision learning
TABLE 3 statistics of positive judgment rate of seismic facies identification for all seismic attribute supervised learning and packet seismic attribute collaborative training semi-supervised learning
Step S48: and carrying out deposition microphase prediction of the mesh corresponding to the non-well points of the target layer by using the classification model.
The second method of the above example is summarized as follows:
(1) The highest positive judgment rate of the seismic facies identification by utilizing the single seismic attribute of the sliding window method can reach 86.4%, but the positive judgment rate of the tidal water course seismic facies identification is generally lower.
(2) The tidal water channel reservoir identification precision is obviously improved by utilizing the supervised learning and collaborative training semi-supervised learning algorithm of the classifier based on the decision tree and the nearest neighbor algorithm by utilizing the multi-seismic attribute.
(3) The total positive judgment rate of multi-seismic attribute supervision learning can reach 95.3%, and the total positive judgment rate after collaborative training can reach 96.6%.
(4) Through comparison of a logistic regression classifier, a decision tree classifier and a nearest neighbor classifier, different classifier algorithms have obvious differences on the identification precision and effect of the earthquake phases, and the decision tree nearest neighbor algorithm classifier is superior to the logistic regression classifier.
(5) According to the comparison analysis of the tidal water channel identification results of the X1 group of seismic attributes, the X2 group of seismic attributes and the all seismic attribute supervised learning and collaborative training semi-supervised learning algorithm, the collaborative training semi-supervised learning algorithm is utilized to improve the positive judgment rate of the seismic phase identification of the tidal water channel, the collaborative training semi-supervised learning algorithm is utilized to identify the characteristics of the X1 group of seismic attributes and the X2 group of seismic phase identification on the plane distribution of the prediction result of the seismic phase identification, the background noise is obviously reduced, and the continuity of the tidal water channel seismic phase identification is obviously enhanced.
(6) After the cooperative training semi-supervised learning, the test set effect of the nearest neighbor classifier is optimal, the highest positive judgment rate can reach 88.5%, and the highest total positive judgment rate of the decision tree classifier can reach 96.5%.
The second embodiment of the invention provides a collaborative training semi-supervised learning algorithm for carrying out the identification of a sediment microphase mode of a tidal water channel in the inside of a carbonate reservoir by utilizing earthquake and drilling data. Co-training is a semi-supervised machine learning algorithm. The algorithm divides the feature set of the data into two subsets, respectively establishes a classifier on each feature subset, and retrains the feature set by using the classification result of the other party until the two classifiers have the same classification result for most of the data. Taking a target layer of a research area as an example, extracting 18 seismic attributes of the target layer by using a phase wrapping method, optimizing 12 seismic attributes by using a sliding window method, and grouping seismic attribute characteristics by using a hierarchical clustering algorithm based on a correlation coefficient. And taking the two groups of seismic attributes as characteristic data, and carrying out cooperative training semi-supervised learning by utilizing logistic regression, decision trees and nearest neighbor classifiers to classify and identify tidal channels and lagoons. The research result shows that the cooperative training semi-supervised learning algorithm can not only improve the positive judgment rate of each classifier, but also reduce the background noise clutter signals, and enhance the continuity of the tidal water channel prediction result plane distribution. The collaborative training semi-supervised learning algorithm obviously improves the effect of identifying the earthquake facies modes of the tidal water channel.
Based on the inventive concept of the present invention, an embodiment of the present invention further provides a device for building a seismic phase prediction model, where the structure of the device is shown in fig. 9, and the device includes:
the data acquisition module 91 is configured to acquire multiple types of screened first-class plane seismic attribute data and multiple types of screened second-class plane seismic attribute data that are gridded according to the same rule, where grids matched with the position information of the well in each type of plane seismic attribute data perform seismic phase marking according to the destination layer deposition microphase type of the well;
the sample set establishing module 92 is configured to screen unlabeled grids from the plane seismic attribute data according to a set rule, and extract first-class seismic attribute values and second-class seismic attribute values of the labeled grids and the screened unlabeled grids; the seismic phase marks of the marking grid and the seismic attribute values of the first class are used as a marking sample to obtain a first marking sample set; taking each seismic attribute value of the first type of unlabeled grid as an unlabeled sample to obtain a first unlabeled sample set; the seismic phase marks of the marking grid and the second type of seismic attribute values are used as a marking sample, and a second marking sample set is obtained; taking each seismic attribute value of the second type of unlabeled grid as an unlabeled sample to obtain a second unlabeled sample set;
The model training module 93 is configured to train the selected classification model by using the first labeled sample set, the first unlabeled sample set, the second labeled sample set, and the second unlabeled sample set through a co-training method, so as to obtain a classification model for seismic phase prediction.
The specific manner in which the various modules perform the operations in the apparatus of the above embodiments have been described in detail in connection with the embodiments of the method, and will not be described in detail herein.
Based on the inventive concept, the embodiment of the invention further provides a computer program product with the function of earthquake phase prediction, which comprises a computer program/instruction, wherein the computer program/instruction realizes the method for establishing the earthquake phase prediction model or the method for predicting the earthquake phase when being executed by a processor.
It should be understood that the specific order or hierarchy of steps in the processes disclosed are examples of exemplary approaches. Based on design preferences, it is understood that the specific order or hierarchy of steps in the processes may be rearranged without departing from the scope of the present disclosure. The accompanying method claims present elements of the various steps in a sample order, and are not meant to be limited to the specific order or hierarchy presented.
In the foregoing detailed description, various features are grouped together in a single embodiment for the purpose of streamlining the disclosure. This method of disclosure is not to be interpreted as reflecting an intention that the claimed embodiments of the subject matter require more features than are expressly recited in each claim. Rather, as the following claims reflect, invention lies in less than all features of a single disclosed embodiment. Thus the following claims are hereby expressly incorporated into this detailed description, with each claim standing on its own as a separate preferred embodiment of this invention.
The foregoing description includes examples of one or more embodiments. It is, of course, not possible to describe every conceivable combination of components or methodologies for purposes of describing the aforementioned embodiments, but one of ordinary skill in the art may recognize that many further combinations and permutations of various embodiments are possible. Accordingly, the embodiments described herein are intended to embrace all such alterations, modifications and variations that fall within the scope of the appended claims. Furthermore, as used in the specification or claims, the term "comprising" is intended to be inclusive in a manner similar to the term "comprising," as interpreted when employed as a transitional word in a claim. Furthermore, any use of the term "or" in the specification of the claims is intended to mean "non-exclusive or".
Claims (14)
1. The method for establishing the earthquake phase prediction model is characterized by comprising the following steps of:
a data acquisition step: acquiring screened multiple first-class plane seismic attribute data and multiple second-class plane seismic attribute data which are gridded according to the same rule, wherein grids matched with the position information of the well in each plane seismic attribute data carry out seismic phase marking according to the destination layer deposition microphase type of the well;
a sample set establishing step: screening unlabeled grids from the plane seismic attribute data according to a set rule, and extracting first-class seismic attribute values and second-class seismic attribute values of the labeled grids and the screened unlabeled grids; the seismic phase marks of the marking grid and the seismic attribute values of the first class are used as a marking sample to obtain a first marking sample set; taking each seismic attribute value of the first type of unlabeled grid as an unlabeled sample to obtain a first unlabeled sample set; the seismic phase marks of the marking grid and the second type of seismic attribute values are used as a marking sample, and a second marking sample set is obtained; taking each seismic attribute value of the second type of unlabeled grid as an unlabeled sample to obtain a second unlabeled sample set;
Model training: and training the selected classification model by using the first marked sample set, the first unmarked sample set, the second marked sample set and the second unmarked sample set through a co-training method to obtain the classification model for seismic phase prediction.
2. The method of claim 1, wherein the obtaining the screened plurality of first-type plane seismic attribute data and the plurality of second-type plane seismic attribute data, which are meshed according to the same rule, specifically comprises:
acquiring multiple types of plane seismic attribute data at a destination layer of a seismic data volume, and gridding the plane seismic attribute data according to the same rule;
determining grids matched with well positions in the plane seismic attribute data according to well position information, and marking the grids by seismic facies according to the well target layer deposition micro-facies type;
screening plane seismic attribute data with the seismic attribute distribution and the seismic coincidence degree of the marks meeting the set requirements;
and clustering the screened multiple plane seismic attribute data, and dividing the data into a first class and a second class.
3. The method of claim 2, wherein if the sedimentary microphase type of the well comprises both a waterway and a non-waterway, the screening of the planar seismic attribute data for which the seismic attribute distribution meets the set requirements with respect to the labeled seismic attribute profile, specifically comprises:
Determining a plurality of water channel segmentation values according to the obtained plane seismic attribute data of each type and according to the seismic attribute values of the plane seismic attribute data, predicting the seismic phase types of each grid contained in the plane seismic attribute data according to each water channel segmentation value, and determining the total positive judgment rate of the predicted seismic phase types according to the marked seismic phases to obtain a plurality of total positive judgment rates;
and screening the plane seismic attribute data with the maximum total positive judgment rate larger than the set total positive judgment rate threshold value.
4. A method as recited in claim 3, further comprising:
determining a plurality of water channel segmentation values according to the obtained plane seismic attribute data of each type and according to the seismic attribute values of the plane seismic attribute data, predicting the seismic phase type of each grid contained in the plane seismic attribute data according to each water channel segmentation value, and determining the total positive judgment rate and the water channel positive judgment rate of the predicted seismic phase type according to the marked seismic phases to obtain a plurality of total positive judgment rates and water channel positive judgment rates;
and screening the plane seismic attribute data with the maximum total positive judgment rate larger than the set total positive judgment rate threshold value and the maximum water channel positive judgment rate larger than the set water channel positive judgment rate threshold value.
5. The method of claim 3 or 4, further comprising, after screening the planar seismic attribute data:
and acquiring a water channel segmentation value corresponding to the maximum total positive judgment rate of the screened plane seismic attribute data, and taking the water channel segmentation value as an optimal water channel segmentation value.
6. The method of claim 5, wherein training the selected classification model by a co-training method using the first labeled sample set, the first unlabeled sample set, the second labeled sample set, and the second unlabeled sample set, further comprises:
and training the selected classification model by a co-training method by utilizing the screened optimal water channel segmentation value of the plane seismic attribute data, the first marked sample set, the first unmarked sample set, the second marked sample set and the second unmarked sample set.
7. The method of claim 1, wherein training the selected classification model by a co-training method using the first labeled sample set, the first unlabeled sample set, the second labeled sample set, and the second unlabeled sample set, specifically comprises:
training the selected classification model by using the first marked sample set to obtain a first classification model, and training the selected classification model by using the second marked sample set to obtain a second classification model;
and the following collaborative training steps are circularly executed until the predictive fitness of the current first classification model and the current second classification model reaches the set requirement:
Selecting a set number of first unlabeled samples from the current first unlabeled sample set, and selecting a set number of second unlabeled samples consistent in grid from the current second unlabeled sample set; inputting the selected first unlabeled sample into a current first classification model, marking the selected second unlabeled sample according to an output result, and adding the second unlabeled sample into a current second marked sample set; inputting the selected second unlabeled sample into a current second classification model, marking the selected first unlabeled sample according to an output result, and adding the first unlabeled sample into a current first marked sample set; the current first classification model is trained using the current first set of labeled samples, and the current second classification model is trained using the current second set of labeled samples.
8. The method of claim 2, wherein the acquiring of the plurality of types of planar seismic attribute data at the destination layer of the volume of seismic data comprises:
determining a target layer obtained through target earthquake phase axis zero-phase package constraint tracking interpretation;
and converting the seismic data volume into a seismic attribute data volume of a set type, and extracting the seismic attribute at the target layer from the seismic attribute data volume to obtain the plane seismic attribute data of the type.
9. The method of claim 2, wherein the classifying the screened plurality of planar seismic attribute data into the first class and the second class after the cluster analysis comprises:
determining correlation coefficients between the screened multiple plane seismic attribute data by hierarchical clustering operation;
and gathering the screened multiple plane seismic attribute data into two types according to the correlation to obtain multiple first plane seismic attribute data and multiple second plane seismic attribute data.
10. The method of any one of claims 1-4 and 7-9, wherein the selected classification model is one of the following models:
a logistic regression classification model, a decision tree classification model, and a nearest neighbor classification model.
11. A method of seismic phase prediction, comprising:
inputting various types of seismic attribute data into a classification model, and carrying out seismic phase prediction according to an output result, wherein the classification model is obtained by using the seismic phase prediction model building method according to any one of claims 1-10, and the types are consistent with the types of plane seismic attribute data screened during building the classification model.
12. The method of claim 11, wherein inputting the plurality of types of seismic attribute data into the classification model and performing the seismic phase prediction based on the output result comprises:
inputting the seismic attribute data with the same type as the first type of plane seismic attribute data into a first classification model to obtain a first output result;
inputting the seismic attribute data consistent with the type of the seismic attribute data of the second class plane into a second classification model to obtain a second output result;
and carrying out earthquake phase prediction according to the first output result and the second output result.
13. A seismic phase prediction model building apparatus, comprising:
the data acquisition module is used for acquiring the screened multiple first-class plane seismic attribute data and multiple second-class plane seismic attribute data which are meshed according to the same rule, and grids matched with the position information of the well in the plane seismic attribute data carry out seismic phase marking according to the destination layer deposition microphase type of the well;
the sample set establishing module is used for screening unlabeled grids from the plane seismic attribute data according to a set rule, and extracting first-class seismic attribute values and second-class seismic attribute values of the labeled grids and the screened unlabeled grids; the seismic phase marks of the marking grid and the seismic attribute values of the first class are used as a marking sample to obtain a first marking sample set; taking each seismic attribute value of the first type of unlabeled grid as an unlabeled sample to obtain a first unlabeled sample set; the seismic phase marks of the marking grid and the second type of seismic attribute values are used as a marking sample, and a second marking sample set is obtained; taking each seismic attribute value of the second type of unlabeled grid as an unlabeled sample to obtain a second unlabeled sample set;
And the model training module is used for training the selected classification model by utilizing the first marked sample set, the first unmarked sample set, the second marked sample set and the second unmarked sample set through a collaborative training method so as to obtain the classification model for the seismic phase prediction.
14. A computer program product having a seismic phase prediction function, comprising computer programs/instructions which when executed by a processor implement the method of seismic phase prediction model creation of any of claims 1 to 10, or the method of seismic phase prediction of claim 11 or 12.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111246409.8A CN116047590A (en) | 2021-10-26 | 2021-10-26 | Method and device for building earthquake phase prediction model and method and device for predicting earthquake phase |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111246409.8A CN116047590A (en) | 2021-10-26 | 2021-10-26 | Method and device for building earthquake phase prediction model and method and device for predicting earthquake phase |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116047590A true CN116047590A (en) | 2023-05-02 |
Family
ID=86124131
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111246409.8A Pending CN116047590A (en) | 2021-10-26 | 2021-10-26 | Method and device for building earthquake phase prediction model and method and device for predicting earthquake phase |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116047590A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116626753A (en) * | 2023-07-25 | 2023-08-22 | 东北石油大学三亚海洋油气研究院 | Microseism event identification method and system based on multi-modal neural network |
| CN117607956A (en) * | 2023-12-20 | 2024-02-27 | 东北石油大学 | Earthquake phase identification method and device, electronic equipment and readable storage medium |
-
2021
- 2021-10-26 CN CN202111246409.8A patent/CN116047590A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116626753A (en) * | 2023-07-25 | 2023-08-22 | 东北石油大学三亚海洋油气研究院 | Microseism event identification method and system based on multi-modal neural network |
| CN116626753B (en) * | 2023-07-25 | 2023-10-13 | 东北石油大学三亚海洋油气研究院 | Microseism event identification method and system based on multi-modal neural network |
| CN117607956A (en) * | 2023-12-20 | 2024-02-27 | 东北石油大学 | Earthquake phase identification method and device, electronic equipment and readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104570083B (en) | Geologic body automatic identifying method based on multi-dimensional earthquake attribute | |
| CN107356958B (en) | A kind of fluvial depositional reservoir substep seismic facies prediction technique based on geological information constraint | |
| US6295504B1 (en) | Multi-resolution graph-based clustering | |
| CN110609320B (en) | Pre-stack seismic reflection pattern recognition method based on multi-scale feature fusion | |
| US11226424B2 (en) | Method for detecting geological objects in a seismic image | |
| CN116047590A (en) | Method and device for building earthquake phase prediction model and method and device for predicting earthquake phase | |
| US7519476B1 (en) | Method of seismic interpretation | |
| CN104020509B (en) | Dam, chiltern beach based on Bayes discriminant analysis sedimentary micro Logging Identification Method | |
| CN104318109A (en) | Shale gas reservoir identification method based on support vector machine | |
| CN108952699B (en) | Intelligent stratum lithology identification method in complex geological drilling process | |
| EP2271954A2 (en) | System and method for interpretation of well data | |
| Eskandari et al. | Reservoir Modelling of Complex Geological Systems--A Multiple-Point Perspective | |
| CN110412662A (en) | Method for prediction of reservoirs of thin interbeded based on seismic multi-attribute deep learning | |
| CN113343574A (en) | Mishrif group lithology logging identification method based on neural network | |
| CN113033648A (en) | Method for realizing logging interpretation by using machine learning algorithm | |
| CN105700016A (en) | Organic bank reservoir earthquake prediction method | |
| Kang et al. | Managing Uncertainty in Geological Scenarios Using Machine Learning‐Based Classification Model on Production Data | |
| CN113378998B (en) | Stratum lithology while-drilling identification method based on machine learning | |
| CN114548299A (en) | Earthquake phase identification method and device, computer equipment and storage medium | |
| CN115330007A (en) | Reservoir multi-classification prediction method based on deep fusion model | |
| CN116201535B (en) | Automatic dividing method, device and equipment for oil and gas reservoir target well sign stratum | |
| Lu et al. | Identifying flow units by FA-assisted SSOM—An example from the Eocene basin-floor-fan turbidite reservoirs in the Daluhu Oilfield, Dongying Depression, Bohai Bay Basin, China | |
| CN117093922A (en) | Improved SVM-based complex fluid identification method for unbalanced sample oil reservoir | |
| CN109143355B (en) | Semi-supervised global optimization seismic facies quantitative analysis method based on SOM | |
| Yuan et al. | Application of Active Learning in Carbonate Lithologic Identification |
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 |