CN106226816A - A kind of pre-stack seismic signal waveform sorting technique - Google Patents
A kind of pre-stack seismic signal waveform sorting technique Download PDFInfo
- Publication number
- CN106226816A CN106226816A CN201610815668.0A CN201610815668A CN106226816A CN 106226816 A CN106226816 A CN 106226816A CN 201610815668 A CN201610815668 A CN 201610815668A CN 106226816 A CN106226816 A CN 106226816A
- Authority
- CN
- China
- Prior art keywords
- degree
- membership
- training
- data
- neuron
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 26
- 210000002569 neuron Anatomy 0.000 claims abstract description 38
- 239000000284 extract Substances 0.000 claims abstract description 11
- 230000009467 reduction Effects 0.000 claims abstract description 8
- 238000013528 artificial neural network Methods 0.000 claims abstract description 6
- 230000008859 change Effects 0.000 claims description 8
- 238000011946 reduction process Methods 0.000 claims description 7
- 238000001914 filtration Methods 0.000 claims description 3
- 239000000203 mixture Substances 0.000 claims description 3
- 238000013075 data extraction Methods 0.000 claims description 2
- 241001269238 Data Species 0.000 abstract description 3
- 230000010354 integration Effects 0.000 abstract description 3
- 239000010410 layer Substances 0.000 description 23
- 230000000694 effects Effects 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 6
- 238000000605 extraction Methods 0.000 description 6
- 238000000926 separation method Methods 0.000 description 6
- 230000000007 visual effect Effects 0.000 description 5
- 230000006870 function Effects 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 208000035126 Facies Diseases 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000001537 neural effect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000007664 blowing Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000011229 interlayer Substances 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 239000003208 petroleum Substances 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
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
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 open a kind of pre-stack seismic signal waveform sorting technique of the present invention, first extracts data, and noise reduction, obtains the integration of all bearing datas;Then the pre-training using DBN extracts characteristic and dimensionality reduction;SOM training is carried out according to the characteristic obtained;Using the neuron weights of SOM as FCM algorithm initial cluster center.The present invention uses the DBN model that the degree of depth learns, and had both been effectively maintained the abstract characteristics of initial data, and had presented again and reduce time complexity for traditional method;The method that make use of self organizing neural network and fuzzy C-mean algorithm, has obtained good classifying quality, and the output data of fuzzy clustering has been more than simple key words sorting, and each sample also leaves the possibility continuing heuristic data to the degree of membership of each bunch.
Description
Technical field
The present invention relates to seismic data process field, particularly relate to a kind of seismic signal waveform sorting technique.
Background technology
Waveform separation is that oil reservoir prediction stores and exploration also plays as an important technology of geophysical exploration technology
A lot of effects, becomes an important technology in petroleum exploration field.Waveform separation also plays in ground Quality Research
In critically important effect, identifying sedimentary facies structure especially.The effect that the result of waveform separation algorithm is final depends on carrying out
Actual geologic signals and the seismic phase number exactly that contains of this geologic signals of locality has been had the when of classification.?
Before three-dimensional geological waveform separation is not introduced into oil exploration, the prediction that oil is distributed by people is mainly by experienced
Matter worker is analyzed according to usual experience and knowledge.The substantial amounts of manpower and materials of such needs, and due to artificially
Intervention too much, the result drawn is the most convincing, and actual application also tends to go wrong.
During seismology technology is incorporated into oil exploration, it is accurate that people slowly find that geological data is carried out science
Early stage prediction can play the biggest effect during oil exploitation, reduce manpower and materials cost put into while add
The fast exploitation rate of exploration.And 3-D seismics signal waveform is classified as a seismographic important branch, also function to lift foot
The effect of weight.By analytically descending signal, (these signals are then to pass through sensor acquisition by blowing out in three-dimensional waveform classification
The signal amplitude returned), then according to by these signals are carried out feature extraction, characterize originally by the feature extracted
Signal, then sorting algorithm is according to the similarity of these features, signal is divided into a certain class setting in classification, according to it
Positional information in the plane generate seismic facies map.People can dope the information feelings of underground with the situation of base area seismic phase
Condition.
In unsupervised waveform separation algorithm, self organizing neural network is obtained in that good result at present, makes it
Become a kind of main flow algorithm solved without supervision waveform separation problem.The method good classification effect and fast convergence rate;It is main
Being the size being regulated neuron by constantly training, the direction of regulation is so that what triumph neuron was won in upper once training
Probability is bigger, and the neuron around triumph neuron also advantage can be improved in training next time.
Traditional Modulation recognition method is primarily directed to poststack signal, and poststack signal is the rough letter to prestack signal
Change, though data volume reduces, but also with large losses anisotropy information as cost.Poststack signal is obtained by conventional needle simultaneously
Classification results the most increasingly cannot meet requirement the highest to seismic data interpretation precision in current oil gas exploration, and traditional
Waveform classification cannot process high-dimensional pre stack data.
Summary of the invention
The present invention solves above-mentioned technical problem, it is proposed that a kind of pre-stack seismic signal waveform sorting technique, first extract
Data, and noise reduction, obtain the integration of all bearing datas;Then the pre-training using DBN extracts feature and dimensionality reduction;In utilization
The characteristic that one step obtains carries out SOM training;Using the neuron weights of SOM as FCM algorithm initial cluster center;Relative to biography
System method reduces time complexity.
The technical solution used in the present invention is: a kind of pre-stack seismic signal waveform sorting technique, including:
S1, to input wide-azimuth geological data noise reduction process;
S2, to the data extraction layer bit field data obtained through step S1, obtain original training sample;Particularly as follows: each layer
Bit data sample points t, m bearing data composition training sample X={x1,...xn, xiDimension is t × m, i=1,2 ...,
N, n represent the number of training sample;
S3, the original training sample obtaining step S2, according to the restriction Boltzmann machine built in degree of depth confidence network
Network extraction characteristic;Specifically include following step by step:
S31, input original training sample, obtain the dimension of original training sample;
S32, initialization K layer limit Boltzmann machine number of network node;
S33, last layer is limited Boltzmann machine network output as next layer limit Boltzmann machine network defeated
Enter, train each layer to limit Boltzmann machine network successively;
The characteristic that S34, output are extracted;
S4, the characteristic input self organizing neural network training that step S3 is extracted, by the neuron weights that obtain to
Measure the initial cluster center as fuzzy C-mean algorithm, when FCM training terminates, obtain classification results.
Further, the noise reduction process described in step S1 is for using structure directing filtering to realize noise reduction.
Further, step S4 specifically include following step by step:
The characteristic that S41, input step S3 extract, and initialize SOM parameter;
S42, carry out SOM training, when iterations is more than pre-set threshold value, obtain neuron;
S43, neuron step S42 obtained are as FCM initial cluster center;
S44, calculate each sample degree of membership to each bunch, then update bunch center;
S45, judge that degree of membership rate of change, whether less than preset value ε, if then output category result, otherwise returns step
S44。
Further, described degree of membership rate of change refers to that the degree of membership currently calculated deducts and adjacent once calculates
The degree of membership come, the value then obtained divided by the adjacent degree of membership once calculated.
Beneficial effects of the present invention: first the present invention extracts data, and noise reduction, obtains the integration of all bearing datas;
Then the pre-training using DBN extracts characteristic and dimensionality reduction;SOM training is carried out according to the characteristic obtained;By the god of SOM
Through unit's weights as FCM algorithm initial cluster center.The present invention uses the DBN model that the degree of depth learns, and has both been effectively maintained original
The abstract characteristics of data, presents again and reduces time complexity for traditional method;Make use of self organizing neural network and Fuzzy C equal
The method of value, has obtained good classifying quality, and the output data of fuzzy clustering has been more than simple key words sorting, each
Sample also leaves the possibility continuing heuristic data to the degree of membership of each bunch.
Accompanying drawing explanation
The protocol procedures figure that Fig. 1 provides for the present invention.
Fig. 2 is typical DBN network.
The flow chart of the RBM network extraction characteristic that Fig. 3 provides for the present invention.
Fig. 4 is that neuron adjusts exemplary plot.
SOM and FCM that Fig. 5 provides for the present invention clusters flow chart.
Detailed description of the invention
For ease of skilled artisan understands that the technology contents of the present invention, below in conjunction with the accompanying drawings present invention is entered one
Step explaination.
Being illustrated in figure 1 the protocol procedures figure of the application, the technical scheme of the application is: a kind of pre-stack seismic signal waveform
Sorting technique, including:
S1, to input wide-azimuth geological data noise reduction process;Here noise reduction process uses structure directing filtering real
Existing.
S2, the wide-azimuth geological data extract layer bit field data after noise reduction process obtaining step S1, obtain former
Beginning training sample;Particularly as follows: each layer of bit data sample points t, m bearing data composition training sample X={x1,
...xn, xiDimension is t × m, i=1,2 ..., n, n represent the number of training sample.
S3, the original training sample obtaining step S2, according to the restriction Boltzmann machine built in degree of depth confidence network
Network extraction characteristic;The present invention, after obtaining original training sample, builds DBN degree of depth confidence network.Because the application is only
It is to use DBN to extract feature, the most reversely adjusting, limiting Boltzmann machine (RBM) network extraction original training sample so building
This characteristic.
The deep belief network (DBN) that the application uses is a generative probabilistic model, with the nerve of traditional discrimination model
Network is relative, and generating model is to set up a Joint Distribution between observed data and label.
DBNs is made up of multiple restriction Boltzmann machines (Restricted Boltzmann Machines) layer, an allusion quotation
The network of type is as shown in Figure 2.It is a visual layers and a hidden layer that these networks " are limited ", and interlayer exists and connects, but in layer
Unit between there is not connection.Hidden unit is trained to catch the dependency of the high level data showed in visual layers.
The connection of one DBN is instructed by top-down generation weights and is determined.First, non-supervisory by one
Greedy successively method goes pre-training to obtain the weights generating model, and non-supervisory greediness successively method is proved effective by Hinton
, and be referred to as contrasting difference (contrastive divergence) by it.
In this training stage, visual layers can produce a vector v, by it, value is delivered to hidden layer.In turn, visually
The input of layer can be by random selection, to attempt going to reconstruct original input signal.Finally, these new visual neurons swash
Forward direction transmission reconstruct hidden layer is activated unit by unit of living, it is thus achieved that h.The step of these backward and forwards is referred to as Gibbs sampling, and hidden
Layer activates the correlation difference between unit and visual layers input just as the Main Basis of right value update.
Be illustrated in figure 3 RBM network extraction feature flow process, specifically include following step by step:
S31, input original training sample, obtain the dimension of original training sample;
S32, initialization K layer limit Boltzmann machine number of network node;
S33, last layer is limited Boltzmann machine network output as next layer limit Boltzmann machine network defeated
Enter, train each layer to limit Boltzmann machine network successively;
The characteristic that S34, output are extracted;
S4, characteristic input self organizing neural network (Self-Organizing Map, SOM) that step S3 is extracted
Training, using the neuron weight vector that obtains as the initial cluster center of fuzzy C-mean algorithm, when FCM training reaches end condition,
Obtain classification results.
As it is shown in figure 5, step S4 specifically include following step by step:
The characteristic that S41, input step S3 extract, and initialize SOM parameter;
S42, carry out SOM training, it is not necessary to Complete Convergence, when iterations is more than pre-set threshold value, obtain neuron;Here
Pre-set threshold value take about 10 times, with specific reference to actual sample data training, naturally it is also possible to neuronal center change
Rate is less than stopping iteration during certain value.
S43, neuron step S42 obtained are as FCM initial cluster center;
S44, calculate each sample degree of membership to each bunch, then update bunch center;
S45, judge that degree of membership rate of change, whether less than preset value ε, if then output category result, otherwise returns step
S44.Degree of membership rate of change refers to that the degree of membership currently calculated deducts the adjacent degree of membership once calculated, then divided by
The value that the adjacent degree of membership once calculated obtains.
Self-organizing Maps (SOM) algorithm used in the application, is that Kohonen proposes a kind of unsupervised clustering algorithm.
The purpose of the method is to find out the structural model being hidden under data, and this structural model is to find out the most accurately
's.SOM is a network structure being made up of neuron, has a variety of neuron layout type, and the network structure of selection is general
It is two-dimentional.All neurons, M is represented with MiRepresent i-th neuron, the dimension of each neuron and the dimension of input data
Degree formed objects, is all for n, so Mi=[Mi1,Mi2,…,Min].The most random initialization each neuron MiSize, so
After further according to input data go adjust neuron size, Self-organizing Maps is the algorithm that a kind of winner wins entirely, because of according to
Rule, the neuron of triumph can adjust towards the direction of input data, because input data are all have to a certain extent similar
Property, after so adjusting neuron, next input enters network, and the probability that the neuron that last time wins is won again can increase
Greatly, here it is the meaning entirely won of winner.Fig. 4 is that neuron adjusts example.
For study every time, from data volume to be processed, choose input data randomly, calculate this input data
With the Euclidean distance of all neurons, the nearest neuron of distance input data is triumph neuron.Computing formula is as follows:
||X-mb| |=mini{||x-mi} (1-1)
The weight of triumph neuron and the neuron adjacent with it can be updated, from formula according to equation below (1-2)
It is known that the weight of triumph neuron and its neuron adjacent with it changes according to the direction to input data.More new formula
As follows:
mi(t+1)=mi(t)+λ(t)hbi(t)[x-mi(t)] (1-2)
Wherein, t is the number of times of iteration, and λ (t) is learning rate, hbiT () is the distance of the neuron i neural b of distance triumph, should
Distance diminishes greatly along with the change of iterations, shown in specific rules such as formula (1-3):
Wherein, rb-ri represents triumph neuron b and some neuron i distance on the topology.
SOM algorithm can above-mentioned two steps of continuous iteration, until iteration stopping condition reaches.Iterated conditional is all artificial
Set, usually hbiT the σ (t) in () is little to a certain extent, σ (t) about a decreasing function of iterations t, such as, takes
σ (t)=1/t.
Fuzzy C-Means Clustering (FCM) fuzzy c-means algorithm (FCMA) algorithm of the application employing or title
FCM.In numerous fuzzy clustering algorithms, FCM (FCM) algorithm is most widely used general and relatively successful, and it passes through optimization aim
Function obtains each sample point degree of membership to all class centers, thus determines that the generic of sample point is to reach automatically sample number
According to carrying out the purpose classified.
Specifically comprise the following steps that
1. input object X={x1,...,xm, determine cluster class number N, determine FUZZY WEIGHTED index p (typically taking 2),
[0,1] initial degree of membership is randomly updated as shown in (1-4)
2. for a bunch CjCalculate barycenter cjFormula is as shown in (1-4)
3. update each object degree of membership w to each cluster centreij, as shown in (1-5)
Repeatedly performing the 2nd, 3 step until algorithmic statement, i.e. front and back cluster centre difference reaches less than threshold value or iterations
Predetermined number of times.
Those of ordinary skill in the art it will be appreciated that embodiment described here be to aid in reader understanding this
Bright principle, it should be understood that protection scope of the present invention is not limited to such special statement and embodiment.For ability
For the technical staff in territory, the present invention can have various modifications and variations.All within the spirit and principles in the present invention, made
Any modification, equivalent substitution and improvement etc., within should be included in scope of the presently claimed invention.
Claims (4)
1. a pre-stack seismic signal waveform sorting technique, it is characterised in that including:
S1, to input wide-azimuth geological data noise reduction process;
S2, to the data extraction layer bit field data obtained through step S1, obtain original training sample;Particularly as follows: each layer of figure place
According to sample points t, m bearing data composition training sample X={x1,...xn, xiDimension is t × m, i=1,2 ..., n, n
Represent the number of training sample;
S3, the original training sample obtaining step S2, according to the restriction Boltzmann machine network built in degree of depth confidence network
Extract characteristic;Specifically include following step by step:
S31, input original training sample, obtain the dimension of original training sample;
S32, initialization K layer limit Boltzmann machine number of network node;
S33, last layer is limited Boltzmann machine network output as next layer limit Boltzmann machine network input, depend on
Each layer of secondary training limits Boltzmann machine network;
The characteristic that S34, output are extracted;
S4, characteristic input self organizing neural network training step S3 extracted, make the neuron weight vector obtained
For the initial cluster center of fuzzy C-mean algorithm, when FCM training terminates, obtain classification results.
A kind of pre-stack seismic signal waveform sorting technique the most according to claim 1, it is characterised in that described in step S1
Noise reduction process realizes noise reduction for using structure directing filtering.
A kind of pre-stack seismic signal waveform sorting technique the most according to claim 1, it is characterised in that step S4 is specifically wrapped
Include following step by step:
The characteristic that S41, input step S3 extract, and initialize SOM parameter;
S42, carry out SOM training, when iterations is more than pre-set threshold value, obtain neuron;
S43, neuron step S42 obtained are as FCM initial cluster center;
S44, calculate each sample degree of membership to each bunch, then update bunch center;
S45, judge that degree of membership rate of change, whether less than preset value ε, if then output category result, otherwise returns step S44.
A kind of pre-stack seismic signal waveform sorting technique the most according to claim 3, it is characterised in that described degree of membership becomes
Rate refers to that the degree of membership currently calculated deducts the adjacent degree of membership once calculated, and then once calculates divided by adjacent
The value that degree of membership out obtains.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610815668.0A CN106226816B (en) | 2016-09-12 | 2016-09-12 | A kind of pre-stack seismic signal waveform sorting technique |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610815668.0A CN106226816B (en) | 2016-09-12 | 2016-09-12 | A kind of pre-stack seismic signal waveform sorting technique |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106226816A true CN106226816A (en) | 2016-12-14 |
CN106226816B CN106226816B (en) | 2018-03-09 |
Family
ID=58073782
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610815668.0A Active CN106226816B (en) | 2016-09-12 | 2016-09-12 | A kind of pre-stack seismic signal waveform sorting technique |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106226816B (en) |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106886043A (en) * | 2017-03-01 | 2017-06-23 | 成都理工大学 | Reservoir detecting method based on geological data deep learning |
CN107688201A (en) * | 2017-08-23 | 2018-02-13 | 电子科技大学 | Based on RBM earthquake prestack signal clustering methods |
CN108226997A (en) * | 2017-11-16 | 2018-06-29 | 中国石油天然气集团公司 | A kind of seismic facies analysis method based on earthquake data before superposition |
CN108680954A (en) * | 2018-08-01 | 2018-10-19 | 中国石油天然气股份有限公司 | Window wave shape clustering method and its device when a kind of more data volumes of frequency domain become |
CN109143355A (en) * | 2018-08-23 | 2019-01-04 | 电子科技大学 | Semi-supervised global optimization seismic facies quantitative analysis method based on SOM |
CN110135236A (en) * | 2019-03-21 | 2019-08-16 | 云南路普斯数据科技有限公司 | A kind of video face identification method based on wavelet transformation and neural network algorithm |
CN110462445A (en) * | 2017-02-09 | 2019-11-15 | 地质探索系统公司 | Geophysics deep learning |
CN111898650A (en) * | 2020-07-08 | 2020-11-06 | 国网浙江省电力有限公司杭州供电公司 | Marketing and distribution data automatic clustering analysis equipment and method based on deep learning |
CN112905717A (en) * | 2021-02-25 | 2021-06-04 | 北方工业大学 | Public safety data distribution method and device |
CN113109869A (en) * | 2021-03-30 | 2021-07-13 | 成都理工大学 | Automatic picking method for first arrival of shale ultrasonic test waveform |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20020161526A1 (en) * | 2001-04-27 | 2002-10-31 | Frederique Fournier | Method for facilitating monitoring, in the course of time, of the evolution of an underground zone by compared analysis of various seismic record sets |
CN1382262A (en) * | 1999-10-20 | 2002-11-27 | 菲利浦石油公司 | Multi-attibute seismic waveform classification |
US6993440B2 (en) * | 2002-04-22 | 2006-01-31 | Harris Corporation | System and method for waveform classification and characterization using multidimensional higher-order statistics |
CN102650702A (en) * | 2012-05-03 | 2012-08-29 | 中国石油天然气股份有限公司 | Seismic waveform analysis and reservoir prediction method and device |
CN103487832A (en) * | 2013-09-12 | 2014-01-01 | 电子科技大学 | Method for classifying supervised waveforms in three-dimensional seismic signal |
CN104181597A (en) * | 2014-08-31 | 2014-12-03 | 电子科技大学 | Seismic facies analysis method based on prestack seismic data |
CN104914467A (en) * | 2015-05-22 | 2015-09-16 | 中国石油天然气股份有限公司 | Classification model channel extracting seismic facies clustering analysis method |
CN105008963A (en) * | 2012-11-03 | 2015-10-28 | 钻井信息公司 | Seismic waveform classification system and method |
-
2016
- 2016-09-12 CN CN201610815668.0A patent/CN106226816B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1382262A (en) * | 1999-10-20 | 2002-11-27 | 菲利浦石油公司 | Multi-attibute seismic waveform classification |
US20020161526A1 (en) * | 2001-04-27 | 2002-10-31 | Frederique Fournier | Method for facilitating monitoring, in the course of time, of the evolution of an underground zone by compared analysis of various seismic record sets |
US6993440B2 (en) * | 2002-04-22 | 2006-01-31 | Harris Corporation | System and method for waveform classification and characterization using multidimensional higher-order statistics |
CN102650702A (en) * | 2012-05-03 | 2012-08-29 | 中国石油天然气股份有限公司 | Seismic waveform analysis and reservoir prediction method and device |
CN105008963A (en) * | 2012-11-03 | 2015-10-28 | 钻井信息公司 | Seismic waveform classification system and method |
CN103487832A (en) * | 2013-09-12 | 2014-01-01 | 电子科技大学 | Method for classifying supervised waveforms in three-dimensional seismic signal |
CN104181597A (en) * | 2014-08-31 | 2014-12-03 | 电子科技大学 | Seismic facies analysis method based on prestack seismic data |
CN104914467A (en) * | 2015-05-22 | 2015-09-16 | 中国石油天然气股份有限公司 | Classification model channel extracting seismic facies clustering analysis method |
Non-Patent Citations (2)
Title |
---|
SONG CHENGYUN ET AL.: "Pre-stack-texture-based reservior characteristics and seismic facies analysis", 《APPLIED GEOPHYSICS》 * |
邓传伟 等: "波形分类技术在储层沉积微相预测中的应用", 《石油物探》 * |
Cited By (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110462445A (en) * | 2017-02-09 | 2019-11-15 | 地质探索系统公司 | Geophysics deep learning |
CN106886043A (en) * | 2017-03-01 | 2017-06-23 | 成都理工大学 | Reservoir detecting method based on geological data deep learning |
CN107688201A (en) * | 2017-08-23 | 2018-02-13 | 电子科技大学 | Based on RBM earthquake prestack signal clustering methods |
CN107688201B (en) * | 2017-08-23 | 2019-12-31 | 电子科技大学 | RBM-based seismic prestack signal clustering method |
CN108226997A (en) * | 2017-11-16 | 2018-06-29 | 中国石油天然气集团公司 | A kind of seismic facies analysis method based on earthquake data before superposition |
CN108680954A (en) * | 2018-08-01 | 2018-10-19 | 中国石油天然气股份有限公司 | Window wave shape clustering method and its device when a kind of more data volumes of frequency domain become |
CN109143355A (en) * | 2018-08-23 | 2019-01-04 | 电子科技大学 | Semi-supervised global optimization seismic facies quantitative analysis method based on SOM |
CN109143355B (en) * | 2018-08-23 | 2019-11-05 | 电子科技大学 | Semi-supervised global optimization seismic facies quantitative analysis method based on SOM |
CN110135236A (en) * | 2019-03-21 | 2019-08-16 | 云南路普斯数据科技有限公司 | A kind of video face identification method based on wavelet transformation and neural network algorithm |
CN111898650A (en) * | 2020-07-08 | 2020-11-06 | 国网浙江省电力有限公司杭州供电公司 | Marketing and distribution data automatic clustering analysis equipment and method based on deep learning |
CN112905717A (en) * | 2021-02-25 | 2021-06-04 | 北方工业大学 | Public safety data distribution method and device |
CN113109869A (en) * | 2021-03-30 | 2021-07-13 | 成都理工大学 | Automatic picking method for first arrival of shale ultrasonic test waveform |
Also Published As
Publication number | Publication date |
---|---|
CN106226816B (en) | 2018-03-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106226816B (en) | A kind of pre-stack seismic signal waveform sorting technique | |
CN104570083B (en) | Geologic body automatic identifying method based on multi-dimensional earthquake attribute | |
CN105510970B (en) | Obtain seismic facies optimal classes purpose method | |
CN110609320B (en) | Pre-stack seismic reflection pattern recognition method based on multi-scale feature fusion | |
CN114463333B (en) | While-drilling geosteering real-time stratum lattice intelligent updating method and system | |
CN106886043B (en) | Reservoir detection method based on seismic data deep learning | |
CN104866810A (en) | Face recognition method of deep convolutional neural network | |
CN106778921A (en) | Personnel based on deep learning encoding model recognition methods again | |
CN102279929B (en) | Remote-sensing artificial ground object identifying method based on semantic tree model of object | |
CN106707335B (en) | A kind of poststack seismic signal waveform sorting technique | |
CN104181597B (en) | Seismic facies analysis method based on prestack seismic data | |
CN107688201A (en) | Based on RBM earthquake prestack signal clustering methods | |
CN102840860B (en) | A kind of method for recognising star map based on Hybrid Particle Swarm | |
CN109670539A (en) | A kind of silt particle layer detection method based on log deep learning | |
CN106443822A (en) | Geological integrated identification method and device based on gravity-magnetic-electric-seismic three-dimensional joint inversion | |
Chopra et al. | Unsupervised machine learning facies classification in the Delaware Basin and its comparison with supervised Bayesian facies classification | |
Zhu et al. | An automatic identification method of imbalanced lithology based on Deep Forest and K-means SMOTE | |
CN117236330B (en) | Mutual information and antagonistic neural network based method for enhancing theme diversity | |
Ali et al. | Data-driven lithofacies prediction in complex tight sandstone reservoirs: a supervised workflow integrating clustering and classification models | |
CN109143355B (en) | Semi-supervised global optimization seismic facies quantitative analysis method based on SOM | |
Jervis et al. | Deep learning network optimization and hyperparameter tuning for seismic lithofacies classification | |
Smith et al. | Self-organizing artificial neural nets for automatic anomaly identification | |
CN104714250A (en) | Practical internal substratum automatic interpretation method | |
DELL'AVERSANA | An integrated multi-physics Machine Learning approach for exploration risk mitigation. | |
Graciolli | A novel classification method applied to well log data calibrated by ontology based core descriptions |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |