CN112434878A - Cascade sample equalization-based seismic fluid prediction method - Google Patents
Cascade sample equalization-based seismic fluid prediction method Download PDFInfo
- Publication number
- CN112434878A CN112434878A CN202011429199.1A CN202011429199A CN112434878A CN 112434878 A CN112434878 A CN 112434878A CN 202011429199 A CN202011429199 A CN 202011429199A CN 112434878 A CN112434878 A CN 112434878A
- Authority
- CN
- China
- Prior art keywords
- samples
- sample
- prediction
- equalization
- fluid
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Geophysics And Detection Of Objects (AREA)
Abstract
The invention relates to a seismic fluid prediction method based on Cascade sample equalization, which comprises the following steps of: 1) acquiring an original learning sample comprising a plurality of types of samples and a plurality of types of samples; 2) and carrying out Cascade sample equalization according to the original learning sample to obtain a new class equalization sample, and constructing a base classifier by adopting the new class equalization sample to carry out integrated prediction on the class of the seismic fluid. Compared with the prior art, the method has the advantages of improving the prediction accuracy and the like.
Description
Technical Field
The invention relates to the field of geophysical exploration, in particular to a seismic fluid prediction method based on Cascade sample equalization.
Background
Seismic prediction of fluids under an artificial intelligence framework has wide application in many fields of geoscience, such as oil and gas exploration and development, geothermal energy development, CO2 storage and the like.
The training examples of lithology and fluid prediction in the field of geoscience always have the problem of unbalanced category, for example, the proportion of gas layers in the training examples of lithology and fluid three-classification tasks (water layers, gas layers and mudstone) is often the least, which causes the prediction result of a machine learning model to be more inclined to most types of samples (water layers and mudstone), so that the problems of unbalanced lithology and fluid category and less target fluid training samples are often faced when the samples are trained under an artificial intelligence framework, and the difficulty of fluid earthquake prediction based on artificial intelligence is greatly increased.
The category imbalance refers to the condition that the number of training samples of different categories in a classification task is greatly different in the field of machine learning, very important reservoir prediction in geophysical exploration often faces the problem of rock property and fluid category imbalance, and the category imbalance can cause that machine learning model prediction is more biased to most samples, so that the classification capability is reduced; particularly, when a few types of samples (such as oil and gas reservoirs) are concerned by geophysical exploration and development, the problem of unbalanced types needs to be solved urgently.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provide a seismic fluid prediction method based on Cascade sample equalization.
The purpose of the invention can be realized by the following technical scheme:
a seismic fluid prediction method based on Cascade sample equalization comprises the following steps:
1) acquiring an original learning sample comprising a plurality of types of samples and a plurality of types of samples;
2) and carrying out Cascade sample equalization according to the original learning sample to obtain a new class equalization sample, and constructing a base classifier by adopting the new class equalization sample to carry out integrated prediction on the class of the seismic fluid.
In the step 1), the data of the original learning sample comprises a lithologic fluid type label, and seismic record data, inversion result data and seismic attribute data corresponding to the lithologic fluid type.
The lithologic fluid categories include water, gas and mudstone.
The seismic record data comprise pre-stack seismic gather data, and the inversion result data comprise longitudinal wave impedance and longitudinal and transverse wave velocity ratio.
The seismic attribute data includes AVO intercept, AVO gradient, instantaneous amplitude, instantaneous phase and instantaneous frequency.
In the step 1), in the original learning sample, the samples with the real lithologic fluid labels as the water layer and the mudstone are majority samples, and the samples with the real lithologic fluid labels as the gas layer are minority samples.
The step 2) specifically comprises the following steps:
21) randomly down-sampling a plurality of types of samples in the original learning sample to ensure that the number of the down-sampled plurality of types of samples is equal to the number of the few types of samples, thereby merging and constructing a class-balanced sub-sample;
22) constructing a base classifier, training by adopting class-balanced sub-samples, and predicting all most types of samples in the original learning samples by adopting the trained base classifier to obtain a prediction result;
23) according to the prediction result, eliminating the majority samples with correct prediction results from all the majority samples in the original learning samples, combining the rest majority samples and the minority samples in the original learning samples into a new class unbalanced sample, taking the new class unbalanced sample as a sample for random down-sampling in the next round, and returning to the step 21);
24) repeating steps 21) -23) to obtain N trained base classifiers;
25) and respectively inputting sample data to be predicted into N base classifiers for prediction to carry out integrated prediction.
The step 25) is specifically as follows:
and respectively inputting sample data to be predicted into N base classifiers for prediction, taking the average value of the obtained N prediction probabilities as a final prediction probability, and selecting the maximum final prediction probability as a final prediction result.
In the step 22), the base classifier adopts a convolutional neural network model.
The number N of the base classifiers is automatically adjusted or manually set according to the class unbalance degree.
Compared with the prior art, the invention has the following advantages:
compared with the existing up-sampling strategy (such as SMOTE and variants thereof), the method not only avoids introducing more noises attached by artificial synthesized samples into a data set, but also improves the calculation efficiency, overcomes the defect of discarding a large amount of effective information of most samples caused by down-sampling compared with the existing down-sampling strategy (such as NearMiss) and can further improve the performance and stability of the model by adopting an integrated prediction method.
The method can fully excavate the multidimensional information of the seismic records, the utilized seismic information comprises a seismic inversion result, post-stack extraction attributes and pre-stack seismic channel sets, the longitudinal and transverse wave velocity ratio in the inversion result is closely related to lithology, and the attribute extraction can obtain more lithology and fluid sensitive characteristics by utilizing the post-stack seismic records; compared with the pre-stack seismic channel set and the post-stack seismic record, the pre-stack seismic channel set can completely reflect AVO changes of different lithologies and fluids, and finally accurate depiction of the gas layer can be realized by comprehensively utilizing multidimensional information.
Drawings
FIG. 1 is a flow chart of the method of the present invention.
Fig. 2 is a diagram of the predicted results of ten well unified training and validation (training set: validation set: 8: 2) without using the category balancing strategy.
Fig. 3 is a diagram of the prediction results of ten wells uniformly trained and validated (training set: validation set: 8: 2) by using the BalanceCascade category equalization strategy.
Fig. 4 is a graph comparing the average gas layer F1 value (unified training, validation) with the class equalization strategy and the average gas layer F1 value (unified training, validation) without the class equalization strategy.
Fig. 5 is a graph comparing the average blind gas zone F1 value with the category equalization strategy to the average blind gas zone F1 value without the category equalization strategy.
Detailed Description
The invention is described in detail below with reference to the figures and specific embodiments.
Examples
As shown in figure 1, the invention provides a Cascade sample equalization-based seismic fluid prediction method, which is used for solving the problem of class imbalance in lithology and fluid prediction multi-classification tasks and effectively improving the prediction accuracy of a minority sample (such as a gas reservoir) under a machine learning frameworkiTraining according to the subdata set to obtain a base classifier CiBy using CiPredicting all the majority samples, discarding all the majority samples with correct prediction, merging the minority samples as a new class unbalanced data set, continuing to perform down-sampling and training the base classifier Ci+1And finally obtaining N base classifiers to realize integrated prediction.
The prediction process of the invention is specifically divided into the following steps:
1) establishing a learning sample: and acquiring real lithologic fluid (water layer, gas layer and mudstone) labels and corresponding seismic records, inversion results (longitudinal wave impedance and longitudinal-transverse wave velocity ratio), seismic attributes and the like by using the drilling, logging and seismic data.
2) The main process of BalanceCascade is as follows:
21) training 1 base classifier: randomly down-sampling the majority of samples to make the number equal to the number of the minority of samples, namely constructing a sub-data set with balanced classes, and training by utilizing the data set to obtain a base classifier
22) Discard most class samples that predict correctly: predicting all the majority samples by using the base classifier, based on the assumption that the predicted correct samples have small contribution to the model, removing the predicted correct majority samples from all the majority samples, and combining the predicted correct majority samples with the minority samples to be used as new class unbalanced samples
23) Training N base classifiers: repeating the step 21) and the step 22) until N base classifiers are obtained, wherein N can be automatically adjusted or set manually according to the class imbalance degree, and N is generally 5/10 in consideration of the calculation efficiency.
N base classifier ensemble prediction: after the lithology probability predicted by each base classifier is averaged, the lithology with the maximum probability is the prediction result
3) Selecting a convolutional neural network model as a base classifier: and (3) the model training sample is a sub data set with balanced categories obtained by down sampling, the pre-stack seismic records and the multiple attributes extracted by the seismic records are used as model input, the lithologic fluid type is used as model output, N base classifiers are obtained according to the flow training in the step two, and finally the prediction is integrated.
Fig. 2 is a diagram of the prediction results of ten wells uniformly trained and validated (training set: validation set: 8: 2) without the category equalization strategy, and the gas layer recall rate without the category equalization strategy is lower than that of the real lithology, for example, W1, W3, W6, W9 and W10, and the partial gas layers of the five wells cannot be predicted.
Fig. 3 is a diagram of the prediction results of ten wells uniformly trained and validated (training set: validation set: 8: 2) by using the BalanceCascade category equalization strategy. The specific implementation process is that the quantity of each type of sample of ten wells is counted, the quantity of all samples is 2815, wherein 828 samples of water layers, 344 samples of gas layers and 1643 samples of mudstone are counted, and the ratio of the quantity of the gas layers to the quantity of the other types of samples is 1: 7.18. and then, performing random downsampling on the water layer samples and the mudstone samples to obtain two types of samples with the same sample size of 344, combining the samples with the 344 gas layer samples to form a sub-data set with balanced classes (the sample size is 1032), and training by using the sub-data set to obtain a base classifier (selecting the convolutional neural network as the base classifier). And then predicting all water layers and gas layers (828 water layer samples and 1643 mudstone samples) by using the base classifier, updating the majority samples according to the prediction result (removing the majority samples with correct prediction), merging the updated majority samples with 344 gas layer samples to form new training samples with unbalanced classes, and further repeating the processes of down-sampling and training the base classifier. And finally, obtaining N base classifiers to realize integrated prediction. N can be set manually or determined according to the unbalanced sample amount of the current category. Compared with the prediction results of the Balancecade type equilibrium strategy without the type equilibrium strategy, the prediction of the gas layers of W1, W3, W6, W9 and W10 is obviously improved, particularly the gas layers of W3, W6 and W10 can not be predicted when the equilibrium strategy is not adopted, and the gas layers can be accurately predicted after the equilibrium strategy is adopted.
Fig. 4 is a comparison graph of the average gas formation F1 value (unified training, verification) using the category equalization strategy and the average gas formation F1 value (unified training, verification) without the category equalization strategy, and the ten well average gas formation F1 value is increased by 3% after the balancescade category equalization strategy is used.
FIG. 5 is a comparison graph of average gas layer F1 values obtained by blind measurements of ten wells using different types of equalization strategies, and the comparison results show that the up-sampling method is poor in effect, possibly because the synthesized samples provide no useful information, and even more samples overlap in the feature space (noise is introduced), so that a few types of samples do not perform well; the down-sampling method (NearMiss) and the integration method (easy Ensemble, Balancecade) have better performance, and the up-and-down sampling combined method (SMOTE + Tomek) has certain effect. Compared with the balance strategy without the balance strategy, the balance gas layer F1 value is improved by 6 percent by adopting the balance Cascade class balance strategy.
FIGS. 2-5 evaluate the effectiveness of the Balancecade class-balancing strategy in terms of the generalization and expression capabilities of the model, respectively. The result shows that the BalanceCascade category equalization strategy can effectively improve the prediction accuracy of the gas layer.
Claims (10)
1. A seismic fluid prediction method based on Cascade sample equalization is characterized by comprising the following steps:
1) acquiring an original learning sample comprising a plurality of types of samples and a plurality of types of samples;
2) and carrying out Cascade sample equalization according to the original learning sample to obtain a new class equalization sample, and constructing a base classifier by adopting the new class equalization sample to carry out integrated prediction on the class of the seismic fluid.
2. The Cascade sample equalization-based seismic fluid prediction method according to claim 1, wherein in the step 1), the data of the original learning sample comprises lithologic fluid class labels and seismic record data, inversion result data and seismic attribute data corresponding to the lithologic fluid class.
3. The Cascade sample equalization-based seismic fluid prediction method of claim 2, wherein the lithological fluid categories comprise water layers, gas layers and mudstones.
4. The method of claim 2, wherein the seismic recording data comprises prestack seismic gather data, and the inversion result data comprises compressional wave impedance and compressional-compressional wave velocity ratio.
5. The Cascade sample equalization-based seismic fluid prediction method of claim 2, wherein the seismic attribute data comprises AVO intercept, AVO gradient, instantaneous amplitude, instantaneous phase and instantaneous frequency.
6. The Cascade sample equalization-based seismic fluid prediction method according to claim 2, wherein in the original learning sample in step 1), the samples with real lithologic fluid labels of water layers and mudstones are majority samples, and the samples with real lithologic fluid labels of gas layers are minority samples.
7. The method for seismic fluid prediction based on Cascade sample equalization as claimed in claim 2, wherein said step 2) comprises the following steps:
21) randomly down-sampling a plurality of types of samples in the original learning sample to ensure that the number of the down-sampled plurality of types of samples is equal to the number of the few types of samples, thereby merging and constructing a class-balanced sub-sample;
22) constructing a base classifier, training by adopting class-balanced sub-samples, and predicting all most types of samples in the original learning samples by adopting the trained base classifier to obtain a prediction result;
23) according to the prediction result, eliminating the majority samples with correct prediction results from all the majority samples in the original learning samples, combining the rest majority samples and the minority samples in the original learning samples into a new class unbalanced sample, taking the new class unbalanced sample as a sample for random down-sampling in the next round, and returning to the step 21);
24) repeating steps 21) -23) to obtain N trained base classifiers;
25) and respectively inputting sample data to be predicted into N base classifiers for prediction to carry out integrated prediction.
8. The method for seismic fluid prediction based on Cascade sample equalization as claimed in claim 7, wherein said step 25) is specifically:
and respectively inputting sample data to be predicted into N base classifiers for prediction, taking the average value of the obtained N prediction probabilities as a final prediction probability, and selecting the maximum final prediction probability as a final prediction result.
9. The Cascade sample equalization-based seismic fluid prediction method as claimed in claim 7, wherein in step 22), the basis classifier employs a convolutional neural network model.
10. The method as claimed in claim 7, wherein the number N of the base classifiers is automatically adjusted or manually set according to the class imbalance degree.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011429199.1A CN112434878B (en) | 2020-12-09 | 2020-12-09 | Cascade sample equalization-based seismic fluid prediction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011429199.1A CN112434878B (en) | 2020-12-09 | 2020-12-09 | Cascade sample equalization-based seismic fluid prediction method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112434878A true CN112434878A (en) | 2021-03-02 |
| CN112434878B CN112434878B (en) | 2022-09-20 |
Family
ID=74692385
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011429199.1A Active CN112434878B (en) | 2020-12-09 | 2020-12-09 | Cascade sample equalization-based seismic fluid prediction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112434878B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113052244A (en) * | 2021-03-30 | 2021-06-29 | 歌尔股份有限公司 | Classification model training method and classification model training device |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107341497A (en) * | 2016-11-11 | 2017-11-10 | 东北大学 | The unbalanced weighting data streams Ensemble classifier Forecasting Methodology of sampling is risen with reference to selectivity |
| US20180210944A1 (en) * | 2017-01-26 | 2018-07-26 | Agt International Gmbh | Data fusion and classification with imbalanced datasets |
| CN108416369A (en) * | 2018-02-08 | 2018-08-17 | 同济大学 | Based on Stacking and the random down-sampled sorting technique of overturning, system, medium and equipment |
| CN109559358A (en) * | 2018-10-22 | 2019-04-02 | 天津大学 | It is a kind of that the method for sampling is risen from the image pattern of coding based on convolution |
| CN110346831A (en) * | 2019-07-19 | 2019-10-18 | 同济大学 | A kind of intelligent earthquake Fluid Identification Method based on random forests algorithm |
| CN110554427A (en) * | 2019-07-23 | 2019-12-10 | 中国石油化工股份有限公司 | Lithology combination prediction method based on forward modeling of seismic waveform |
| CN111596354A (en) * | 2020-05-11 | 2020-08-28 | 同济大学 | Seismic reservoir prediction method considering space constraint under machine learning framework |
-
2020
- 2020-12-09 CN CN202011429199.1A patent/CN112434878B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107341497A (en) * | 2016-11-11 | 2017-11-10 | 东北大学 | The unbalanced weighting data streams Ensemble classifier Forecasting Methodology of sampling is risen with reference to selectivity |
| US20180210944A1 (en) * | 2017-01-26 | 2018-07-26 | Agt International Gmbh | Data fusion and classification with imbalanced datasets |
| CN108416369A (en) * | 2018-02-08 | 2018-08-17 | 同济大学 | Based on Stacking and the random down-sampled sorting technique of overturning, system, medium and equipment |
| CN109559358A (en) * | 2018-10-22 | 2019-04-02 | 天津大学 | It is a kind of that the method for sampling is risen from the image pattern of coding based on convolution |
| CN110346831A (en) * | 2019-07-19 | 2019-10-18 | 同济大学 | A kind of intelligent earthquake Fluid Identification Method based on random forests algorithm |
| CN110554427A (en) * | 2019-07-23 | 2019-12-10 | 中国石油化工股份有限公司 | Lithology combination prediction method based on forward modeling of seismic waveform |
| CN111596354A (en) * | 2020-05-11 | 2020-08-28 | 同济大学 | Seismic reservoir prediction method considering space constraint under machine learning framework |
Non-Patent Citations (1)
| Title |
|---|
| 迷雾总会解: "不平衡数据的介绍和处理(ImbalancedDatasets)", 《HTTPS://BLOG.CSDN.NET/QQ_44766883/ARTICLE/DETAILS/109688785#:~:TEXT=BALANCEC,OOSTING%EF%BC》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113052244A (en) * | 2021-03-30 | 2021-06-29 | 歌尔股份有限公司 | Classification model training method and classification model training device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112434878B (en) | 2022-09-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Marano et al. | Generative adversarial networks review in earthquake-related engineering fields | |
| Dorrington et al. | Genetic-algorithm/neural-network approach to seismic attribute selection for well-log prediction | |
| Lin et al. | Automatic lithology identification by applying LSTM to logging data: A case study in X tight rock reservoirs | |
| Akkurt et al. | Accelerating and enhancing petrophysical analysis with machine learning: a case study of an automated system for well log outlier detection and reconstruction | |
| CN108732620B (en) | Multi-wave earthquake oil and gas reservoir prediction method under unsupervised and supervised learning | |
| Ismail et al. | Gas channels and chimneys prediction using artificial neural networks and multi-seismic attributes, offshore West Nile Delta, Egypt | |
| Lubo-Robles et al. | Machine learning model interpretability using SHAP values: Application to a seismic facies classification task | |
| CN111596978A (en) | Web page display method, module and system for lithofacies classification by artificial intelligence | |
| CN110412662A (en) | Method for prediction of reservoirs of thin interbeded based on seismic multi-attribute deep learning | |
| Song et al. | k NN-based gas-bearing prediction using local waveform similarity gas-indication attribute—An application to a tight sandstone reservoir | |
| CN113608265A (en) | Reservoir prediction method based on deep hybrid neural network | |
| CN112434878B (en) | Cascade sample equalization-based seismic fluid prediction method | |
| CN117452518B (en) | Reservoir lithology prediction method based on multidisciplinary data fusion clustering algorithm | |
| Alameedy et al. | Evaluating machine learning techniques for carbonate formation permeability prediction using well log data | |
| Ren et al. | Data driven oil production prediction and uncertainty quantification for unconventional asset development planning through machine learning | |
| CN110118994B (en) | Continental facies hydrocarbon source rock quantitative prediction method based on seismic inversion and machine learning | |
| Dai et al. | Stratigraphic automatic correlation using SegNet semantic segmentation model | |
| CN115880455A (en) | Three-dimensional intelligent interpolation method based on deep learning | |
| CN114137610A (en) | Low exploration area stratum and lithology earthquake evaluation method based on joint supervision and unsupervised learning | |
| Akkurt et al. | An Unsupervised Machine-Learning Workflow for Outlier Detection and Log Editing With Prediction Uncertainty | |
| CN106405638A (en) | Method and apparatus for determining gas saturation | |
| de Oliveira et al. | Ensemble of heterogeneous classifiers applied to lithofacies classification using logs from different wells | |
| CN117407841B (en) | Shale layer seam prediction method based on optimization integration algorithm | |
| Berry et al. | Integrated Assisted History Matching and Forecast Optimisation Under Uncertainty for More Robust Mature Field Redevelopment Project | |
| Wang et al. | Application of multiple attributes fusion technology in the Su-14 Well Block |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |