CN107102379A - A kind of method that seat earth watery prediction is carried out based on many attribution inversions - Google Patents
A kind of method that seat earth watery prediction is carried out based on many attribution inversions Download PDFInfo
- Publication number
- CN107102379A CN107102379A CN201610094397.4A CN201610094397A CN107102379A CN 107102379 A CN107102379 A CN 107102379A CN 201610094397 A CN201610094397 A CN 201610094397A CN 107102379 A CN107102379 A CN 107102379A
- Authority
- CN
- China
- Prior art keywords
- attribute
- carried out
- prediction
- data
- attributes
- 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
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V11/00—Prospecting or detecting by methods combining techniques covered by two or more of main groups G01V1/00 - G01V9/00
Abstract
The present invention provides a kind of method that seat earth watery prediction is carried out based on many attribution inversions, belongs to coal exploration field, including：The step of extracting some seismic properties；Mathematic(al) manipulation to the apparent resistivity and seismic properties simultaneously carries out the step of related intersection obtains sensitive earthquake attribute；The sensitive earthquake attribute is carried out to screen the step of obtaining optimal models；The step of apparent resistivity body is obtained according to the 3D seismic data in the neural network training model and area to be predicted；Choose reference plane and the step of section is extracted is carried out to the apparent resistivity body.The present invention is adaptable to various geological datas, it is time-consuming short to calculate, and has a wide range of application.
Description
Technical field
The invention belongs to field of geophysical exploration, and in particular to a kind of method that seat earth watery prediction is carried out based on many attribution inversions.
Background technology
With the continuous intensification of coal mining depth, requirement to coalfield 3-d seismic exploration technology also more and more higher.In terms of lithology and fluid prediction, the wave impedance information that the inverting of poststack P-wave impedance is provided can not meet the requirement of Seam Roof And Floor fluid prediction, although and there is huge advantage in terms of fluid prediction in prestack inversion method, but it is due to that it requires the reasons such as height, processing method complexity to initial data, fails to have given play to due effect in terms of the lithology and fluid prediction of coal seam always.
The content of the invention
The present invention in view of the shortcomings of the prior art, proposes a kind of method applied many attribution inversion technologies, be predicted on the basis of preferred Sensitive Attributes to seat earth watery.
To reach above-mentioned purpose, technical scheme includes：
The step of using the apparent resistivity curve in area to be predicted as aim curve；
The step of some seismic properties being extracted from the aim curve；
The step of mathematic(al) manipulation is carried out to the seismic properties and apparent resistivity；
The step of related intersection is carried out to the seismic properties, apparent resistivity and the result of mathematic(al) manipulation；
The step of sensitive earthquake attribute is obtained according to the related intersection；
The sensitive earthquake attribute is carried out to screen the step of obtaining optimal models；
The step of apparent resistivity body is obtained according to the 3D seismic data in the neural network training model and area to be predicted；
Choose reference plane and the step of section is extracted is carried out to the apparent resistivity body；
The step of exporting the section.
The beneficial effects of the present invention are there is provided a kind of method that seat earth watery prediction is carried out based on many attribution inversions, the non-linear relation that the present invention is set up between physical properties of rock and sensitive earthquake combinations of attributes using probabilistic neural network algorithm, the direct inversion of petrophysical parameter is realized, and its, calculating adaptable to various geological datas is time-consuming short.
Brief description of the drawings
Fig. 1 show a kind of flow chart of the embodiment for the method that seat earth watery prediction is carried out based on many attribution inversions of the present invention.
Embodiment
The specific embodiment of the invention is described in detail below in conjunction with specific accompanying drawing.It should be noted that the combination of the technical characteristic or technical characteristic described in following embodiments is not construed as isolated, they can be mutually combined so as to reaching superior technique effect.
Fig. 1 show a kind of flow chart of the embodiment for the method that seat earth watery prediction is carried out based on many attribution inversions of the present invention.
Below in conjunction with a kind of embodiment for the method that seat earth watery prediction is carried out based on many attribution inversions of the present invention shown in Fig. 1, is discussed in detail, it includes：
Step 101：The step of some seismic properties being extracted from the aim curve：Local area is extracted 26 kinds of seismic properties altogether；
Step 102：The step of mathematic(al) manipulation is carried out to the seismic properties and apparent resistivity：Made 4 kinds of mathematic(al) manipulations to apparent resistivity and seismic properties, square, evolution, logarithm and inverse, then apparent resistivity and seismic properties made 400 kinds to mathematic(al) manipulation in itself related intersected；
Step 103：The mathematic(al) manipulation related intersection of progress to the apparent resistivity and seismic properties and the step of obtain sensitive earthquake attribute：The fundamentum of seismic-sensitive attribute selection is：The best attribute of correlation is chosen first from parameter list, the criterion of judge is that root-mean-square error is minimum.Then allow remaining attribute and this attribute is matched one by one, find best that a pair of correlation, the best attribute of two attributes match found and above selected in remaining attribute afterwards is found, until whole attributes of offer successively（26 kinds of seismic properties）Screening is finished；
Step 104：The step of neural metwork training obtains neural network training model is carried out to the sensitive earthquake attribute and the apparent resistivity curve；
Step 105：The step of apparent resistivity body is obtained according to the 3D seismic data in the neural network training model and area to be predicted；
Step 106：Choose reference plane and the step of section is extracted is carried out to the apparent resistivity body.
Claims (6)
1. a kind of method that seat earth watery prediction is carried out based on many attribution inversions, it is characterised in that including：
The step of some seismic properties being extracted from the aim curve；
The step of mathematic(al) manipulation is carried out to the seismic properties and apparent resistivity；
It is related to the mathematic(al) manipulation progress of the apparent resistivity and seismic properties the step of intersect and obtain sensitive earthquake attribute；
The sensitive earthquake attribute is carried out to screen the step of obtaining optimal models；
The step of apparent resistivity body is obtained according to the 3D seismic data in the area to be predicted；
Choose reference plane and the step of section is extracted is carried out to the apparent resistivity body.
2. the method according to claim 1 that seat earth watery prediction is carried out based on many attribution inversions, it is characterised in that the seismic properties include wave impedance, amplitude, frequency, phase, curvature, energy, variance, relevant, dip angle attribute etc..
3. the method according to claim 1 that seat earth watery prediction is carried out based on many attribution inversions, it is characterised in that the mathematic(al) manipulation includes the mathematical operation to the seismic properties and apparent resistivity progress square, evolution, logarithm and inverse.
4. the method according to claim 1 that seat earth watery prediction is carried out based on many attribution inversions, it is characterised in that related intersection is carried out to the seismic properties, apparent resistivity and the result of mathematic(al) manipulation and is included the step of obtaining Sensitive Attributes：（1）Single attribute forecast：Refer to using a kind of seismic properties and target logging trace opening relationships, this relation is applied into whole region to predict this target logging trace characteristic, the shortcoming of this method is that the precision predicted is not high；（2）Many attribute forecasts：In order to improve the precision of prediction, it is necessary to predict rock parameter using one group of attribute simultaneously, in particular, it may be desired to by following several steps：A. investigate, analyze well logging and geological data at well point, determine a suitable seismic properties collection；B. well logging and the relation of seismic properties are asked for using multicomponent linear regressioning technology or nerual network technique；C. related operation is carried out to geological data, obtains desired rock parameter data volume；It is the size from which attribute and every kind of attribute weights using the key of a variety of attribute forecast rock parameters, multiattribute method can be divided into arithmetic of linearity regression and neural network algorithm etc. again, the result and the correlation of real data that multicomponent linear regressioning technology is predicted are not fine sometimes, and neural network can improve its correlation, probabilistic neural network（Probabilistic neural network, abbreviation PNN）It is a kind of method that utilization neural network structure carries out mathematical interpolation, used packet contains a series of training samples, training process is to find optimal weights in node, training data includes a series of training sample, compared with traditional inversion method, carrying out lithology prediction using probabilistic neural network has advantages below：In addition to prestack and poststack data, can also using other attributes carry out objective attribute target attribute prediction, and independent of specific forward model, do not need seismic wavelet relevant knowledge, it is possible to the confidence level that it predicts the outcome is examined with cross-validation method.
5. the method according to claim 1 that seat earth watery prediction is carried out based on many attribution inversions, it is characterised in that screening the step of obtaining optimal models to sensitive earthquake attribute progress includes：（1）Attribute is in optimized selection, least square method can for selection best attributes, but how specifically chosen one group of optimal attributeGenerally there are following two methods：A kind of method is by exhaustive search, it is assumed that it is desirable that selecting M attribute from N attribute, convolution factor length is L, test the combination of all attribute, what predicated error was minimum in all combinations is exactly required solution, and problem is that exhaustive search operand is very big, is taken very long；Method that is another quick but not being optimization is the algorithm of single step optimizing, this algorithm is assumed to have known best M kinds combination, so best M+1 kinds combination necessarily includes this combinations of attributes, certainly, the coefficient being previously calculated must be recalculated, assuming that there is many attribute, the first step finds out a best attribute by exhaustive search, and we are called attribute 1；Second step finds out a pair of best attributes, and all properties and attribute 1 are combined into attribute pair, best one attribute pair is obtained by asking for minimum predicated error, determines therefrom that attribute 2；3rd step finds three best combinations of attributes, in all three combinations of attributes constituted with attribute 1, attribute 2, the combination of three best attributes is found out by asking for minimum predicated error, obtain accordingly attribute 3 then the like, profit can make the time that run time is considerably less than needed for exhaustive search in this way；（2）Cross validation：For using how many attributes, come the prediction effect being optimal, cross validation can be passed through（Cross-Validation）To solve, mathematically say, increasing more attributes will make to predict the outcome more preferably, it is not intended that increasing more attributes can obtain closer to really predicting the outcome, cross validation is that all training datas are divided into two groups first, training data group and checking data group：Training data group is used for carrying out attribute conversion, and checking data group is used for the predicated error of the result；By cross validation, the overfitting to attribute data or over training can be prevented effectively from, during analysis, typically data are grouped according to well location, i.e., training data group includes the training sample of all wells（Remove some and hide well）；Checking data group is included in the data of all hiding wells, cross-validation process, how many mouthful well, it is necessary to which how many times verify analysis process, and different wells must all be left out each time by verifying；（3）The selection of convolution operator：The frequency of log is higher than the frequency of geological data, relation and non-best choice that directly " point-to-point " is set up between target component and geological data between log parameter and seismologic parameter, therefore agreement " one " log parameter can be corresponding with " one group " geological data, generally, for any log parameter, equal expected wavelet is capable of all geological data sampling points of scanning neighboring, convolution operator is equivalent to a series of new attributes are introduced, and these new attributes are the deformation or conversion that primitive attribute is changed over time.
6. the method according to claim 1 that seat earth watery prediction is carried out based on many attribution inversions, it is characterised in that the step of obtaining apparent resistivity body according to the 3D seismic data in the neural network training model and area to be predicted includes：（1）Seismic properties and the relation of apparent resistivity at well point are established based on multiattribute neural network training model, this relation can be generalized to whole region；（2）The computing of this relation and geological data is carried out at non-well point, the apparent resistivity attribute of whole data volume has been obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610094397.4A CN107102379A (en) | 2016-02-19 | 2016-02-19 | A kind of method that seat earth watery prediction is carried out based on many attribution inversions |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610094397.4A CN107102379A (en) | 2016-02-19 | 2016-02-19 | A kind of method that seat earth watery prediction is carried out based on many attribution inversions |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107102379A true CN107102379A (en) | 2017-08-29 |
Family
ID=59658675
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610094397.4A Pending CN107102379A (en) | 2016-02-19 | 2016-02-19 | A kind of method that seat earth watery prediction is carried out based on many attribution inversions |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107102379A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107894615A (en) * | 2017-11-13 | 2018-04-10 | 中国石油化工股份有限公司华北油气分公司勘探开发研究院 | A kind of method of quantitative evaluation 3-D seismics attribute forecast reservoir parameter validity |
| CN108645994A (en) * | 2018-04-25 | 2018-10-12 | 中国石油大学(北京) | A kind of geology stochastic inversion methods and device based on Multiple-Point Geostatistics |
| CN111580181A (en) * | 2020-04-22 | 2020-08-25 | 中国矿业大学(北京) | Water guide collapse column identification method based on multi-field multi-feature information fusion |
| CN112068221A (en) * | 2020-09-07 | 2020-12-11 | 中国煤炭地质总局地球物理勘探研究院 | Coal bed water-rich analysis method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102478668A (en) * | 2010-11-30 | 2012-05-30 | 中国石油天然气集团公司 | Method for applying seismic multiattribute parameters to predicting coal seam thickness |
| CN103954991A (en) * | 2014-03-13 | 2014-07-30 | 中国石油集团川庆钻探工程有限公司地球物理勘探公司 | Multi-attribute earthquake data inversion method and device |
| CN105223612A (en) * | 2015-06-10 | 2016-01-06 | 中国矿业大学 | A kind of coal mine flood prediction and evaluation method based on earthquake information |
-
2016
- 2016-02-19 CN CN201610094397.4A patent/CN107102379A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102478668A (en) * | 2010-11-30 | 2012-05-30 | 中国石油天然气集团公司 | Method for applying seismic multiattribute parameters to predicting coal seam thickness |
| CN103954991A (en) * | 2014-03-13 | 2014-07-30 | 中国石油集团川庆钻探工程有限公司地球物理勘探公司 | Multi-attribute earthquake data inversion method and device |
| CN105223612A (en) * | 2015-06-10 | 2016-01-06 | 中国矿业大学 | A kind of coal mine flood prediction and evaluation method based on earthquake information |
Non-Patent Citations (4)
| Title |
|---|
| 张丽红 等: "多属性技术在底板砂岩富水性预测中的应用", 《北京工业职业技术学院学报》 * |
| 张丽红: "桃园矿区陷落柱及其富水性地球物理识别方法", 《中国博士学位论文全文数据库 基础科学辑》 * |
| 李雪梅 等: "利用地震属性预测煤层顶底板含水性研究", 《煤炭工程》 * |
| 陆大华 等: "地震多属性反演预测煤层顶底板富水性", 《中国煤炭地质》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107894615A (en) * | 2017-11-13 | 2018-04-10 | 中国石油化工股份有限公司华北油气分公司勘探开发研究院 | A kind of method of quantitative evaluation 3-D seismics attribute forecast reservoir parameter validity |
| CN107894615B (en) * | 2017-11-13 | 2019-06-18 | 中国石油化工股份有限公司华北油气分公司勘探开发研究院 | A kind of method of quantitative evaluation 3-D seismics attribute forecast reservoir parameter validity |
| CN108645994A (en) * | 2018-04-25 | 2018-10-12 | 中国石油大学(北京) | A kind of geology stochastic inversion methods and device based on Multiple-Point Geostatistics |
| CN111580181A (en) * | 2020-04-22 | 2020-08-25 | 中国矿业大学(北京) | Water guide collapse column identification method based on multi-field multi-feature information fusion |
| CN111580181B (en) * | 2020-04-22 | 2021-07-20 | 中国矿业大学(北京) | Water guide collapse column identification method based on multi-field multi-feature information fusion |
| CN112068221A (en) * | 2020-09-07 | 2020-12-11 | 中国煤炭地质总局地球物理勘探研究院 | Coal bed water-rich analysis method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111126591B (en) | Magnetotelluric deep neural network inversion method based on space constraint technology | |
| US8255165B2 (en) | Method for predicting differences in subsurface conditions | |
| US10890688B2 (en) | Method for generating secondary data in geostatistics using observed data | |
| CN107102379A (en) | A kind of method that seat earth watery prediction is carried out based on many attribution inversions | |
| US8706462B2 (en) | System and method for providing a physical property model | |
| CN110954950B (en) | Underground transverse wave velocity inversion method, device, computing equipment and storage medium | |
| AU2013272212A1 (en) | System and method for facies classification | |
| CN111596354A (en) | Seismic reservoir prediction method considering space constraint under machine learning framework | |
| CN111722283B (en) | Stratum velocity model building method | |
| CN114529110A (en) | Lithofacies inversion method and system based on deep neural network model | |
| CN110927793B (en) | Reservoir prediction method and system based on sequential random fuzzy simulation | |
| CN112100930B (en) | Formation pore pressure calculation method based on convolutional neural network and Eaton formula | |
| CN110118994B (en) | Continental facies hydrocarbon source rock quantitative prediction method based on seismic inversion and machine learning | |
| US20150063650A1 (en) | Identifying Potential Fracture Treatment Locations In a Formation Based on Production Potential | |
| CN111894563B (en) | Classification determination method and system for crack type reservoir in submarine mountain section | |
| CN106842316A (en) | Crack determines method and apparatus | |
| Sarkheil et al. | The fracture network modeling in naturally fractured reservoirs using artificial neural network based on image loges and core measurements | |
| CN109188513A (en) | Generate the method, apparatus and storage medium of Depth Domain data volume | |
| An et al. | Reservoir characterization using seismic waveform and feedforword neural networks | |
| CN110568149B (en) | Fine and rapid quantitative simulation method for hydrocarbon generation and discharge history of sedimentary basin hydrocarbon source rock | |
| CN115880455A (en) | Three-dimensional intelligent interpolation method based on deep learning | |
| Bashore et al. | Multivariate statistical applications for quantifying contribution of seismic data to production prediction in the Wolfcamp, Midland Basin | |
| CN112147677A (en) | Oil and gas reservoir parameter tag data generation method and device | |
| CN106291748A (en) | A kind of method obtaining content of organic carbon of hydrocarbon source rock | |
| CN111815769A (en) | Modeling method, computing device and storage medium for thrust-driven tectonic belt structure |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170829 |
|
| WD01 | Invention patent application deemed withdrawn after publication |