CN112052977A - Oil reservoir reserve prediction method based on deep space-time attention network - Google Patents
Oil reservoir reserve prediction method based on deep space-time attention network Download PDFInfo
- Publication number
- CN112052977A CN112052977A CN201911343248.7A CN201911343248A CN112052977A CN 112052977 A CN112052977 A CN 112052977A CN 201911343248 A CN201911343248 A CN 201911343248A CN 112052977 A CN112052977 A CN 112052977A
- Authority
- CN
- China
- Prior art keywords
- oil reservoir
- network
- reserves
- attention network
- deep
- 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 36
- 238000013135 deep learning Methods 0.000 claims abstract description 5
- 238000012549 training Methods 0.000 claims description 13
- 230000006870 function Effects 0.000 claims description 9
- 238000010606 normalization Methods 0.000 claims description 2
- 238000012360 testing method Methods 0.000 claims description 2
- 238000005457 optimization Methods 0.000 claims 2
- 238000007500 overflow downdraw method Methods 0.000 claims 1
- 230000000694 effects Effects 0.000 abstract description 3
- 238000013528 artificial neural network Methods 0.000 abstract description 2
- 125000004122 cyclic group Chemical group 0.000 abstract description 2
- 230000002411 adverse Effects 0.000 abstract 1
- 238000004519 manufacturing process Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000000523 sample Substances 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
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
- G06N3/044—Recurrent networks, e.g. Hopfield 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/02—Agriculture; Fishing; Forestry; Mining
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- General Engineering & Computer Science (AREA)
- Molecular Biology (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Data Mining & Analysis (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Biophysics (AREA)
- Tourism & Hospitality (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Agronomy & Crop Science (AREA)
- Animal Husbandry (AREA)
- Marine Sciences & Fisheries (AREA)
- Mining & Mineral Resources (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
In oil field exploration planning decision schemes, reservoir reserves are always difficult to predict. The precision of the existing prediction method can not meet the requirement of practical application. The invention is inspired by a cyclic neural network and an attention mechanism, provides a deep space-time attention model focusing on reservoir reserve prediction, and can relieve the adverse effect of data fluctuation on a prediction result, thereby greatly reducing a prediction error. The experimental result on the real data of a large oil field shows that compared with the traditional method and the existing deep learning method, the model prediction precision is obviously improved, and a better choice is provided for the future oil reservoir reserves prediction.
Description
Technical Field
The invention belongs to the field of time sequence prediction, is an important application in the field of deep learning, and particularly relates to a method for predicting oil reservoir reserves based on a deep space-time attention network.
Technical Field
The oil reservoir resource is a precondition for oil field enterprises to live, and determines the development prospect of the enterprises. Therefore, analysis and prediction of oil and gas reserves are an extremely important basic work in the strategic development research of oil and gas enterprises, and the quality of the basic work is related to the success or failure of exploration work and the production benefit of the enterprises. Only through scientific analysis and calculation can the greatest benefit be created as far as possible with fixed investment. The method has great significance for improving the oil reservoir discovery efficiency and optimizing the exploration planning structure by accurately predicting the oil reservoir reserves. In the process of oil and gas exploitation, related researchers are always seeking an accurate and efficient oil reservoir reserve prediction method. However, in the process of predicting the oil and gas reserves and the production trend, because certain uncertain factors are difficult to accurately quantify, a small error exists between the predicted result and the actual production.
On the basis of exploration planning data analysis, a method for predicting oil reservoir reserves based on a deep space-time attention network is provided. By establishing prediction models of different levels of geological units and utilizing historical data of recent decades, the oil and gas exploration reserves of different geological units can be predicted, and the oil and gas exploration reserves of different exploration stages of a specific geological unit can be predicted in the next year.
Compared with the traditional method and the existing deep learning method, the prediction accuracy of the oil reservoir reserves prediction method based on the deep space-time attention network is obviously improved, and a better choice is provided for the future oil reservoir reserves prediction.
Disclosure of Invention
The method provides an oil reservoir reserve prediction method based on a deep space-time attention network. The method combines a cyclic neural network and an attention mechanism, greatly improves the precision of reservoir reserves prediction, and verifies the effectiveness of the method on real data of a certain large oil field.
The technical solution is as follows:
step 1), extracting the number of well openings, the number of oil reservoirs and the oil reservoir reserves of the historical exploration from a database and carrying out normalization;
step 2), designing a deep space-time attention network, including a time attention network, a space attention network and an LSTM main network;
step 3), establishing a loss function and a proportional coefficient between the oil reservoir reserves and the oil reservoir number loss function;
step 4), training the network by using the data prepared in the step 1);
and 5) obtaining a trained model, and predicting the year to be predicted by using the model.
In the step 1), the data are normalized to enable the distribution interval to be [0,1], so that the variation degree of parameters in the network fitting process is reduced, and the network fitting capability is improved.
And 2) designing an attention network aiming at the characteristics of input data and intermediate output hidden states, and capturing key values in the data.
And 3) determining a weight factor of the loss function in the step 3), and balancing the prediction precision between the number of the oil reservoirs and the oil reservoir reserves.
A fixed weight algorithm is used in the step 4); the training mode is a classic deep learning training method.
In the whole method, the network is realized end to end in the training and testing processes. After large-scale data training, the effectiveness of the effect is verified on the real data of the oil field.
The invention combines an attention mechanism in a general LSTM network, greatly improves the precision of a prediction result under the condition of not increasing the training cost basically, and has high use value and strong expandability
Drawings
FIG. 1 is a deep neural network model schematic diagram for reservoir reserves prediction based on a deep spatiotemporal attention network, which is constructed by the invention.
Detailed Description
A reservoir reserve prediction method based on a deep space-time attention network comprises the following steps:
1) extracting the number of well detection ports, the number of oil reservoirs and the oil reservoir reserve from the database in the past year (1964-;
2) and constructing a universal LSTM network which is a universal two-layer LSTM structure, wherein each layer has 3 nodes, the output result is connected with a full-connection layer, and finally the number of the output oil reservoirs and the oil reservoir reserves are obtained.
3) And 2) adding a space-time attention network on the basis of the network in the step 2), wherein the space-time attention network is similar in structure and consists of two fully-connected layers, the number of input nodes is increased by three times, and then the number of the input nodes is reduced by three times, so that the number is kept unchanged. The temporal attention network is added to the number of probe wells at the input and the spatial attention network is added to the hidden state at the output.
4) Defining a loss function of the network, and simultaneously optimizing two values of the number of oil reservoirs and the oil reservoir reserves, so that the loss functions are jointly calculated, wherein a hyper-parameter alpha is defined, and the loss function formula is as follows:
and respectively using first-order distance loss for the two, wherein g is an actual value of the number of the oil reservoir, g 'is a predicted value of the number of the oil reservoir, c is an actual value of the reserve of the oil reservoir, c' is a predicted value of the reserve of the oil reservoir, alpha is a hyper-parameter, manual predefining is required, and the given value is 0.4.
5) In the training process of the specific example of the oil reservoir reserve prediction method based on the deep space-time attention network, the learning rate is defined to be 0.001, the size of a training batch is 1, and the training is carried out until the loss function is basically not reduced.
6) And (3) forming a batch by the number of the well detecting ports of the year to be tested (2013) and the first nine years (2004-.
The invention provides a deep space-time attention model focusing on reservoir reserves prediction, and the attention mechanism is introduced into a reservoir reserves task for the first time, so that the prediction error is greatly reduced. The experimental result of the real data verifies the effectiveness of the method. The attention model achieves the best results known to date on this data set.
The technical content which is not described in the above mode can be realized by adopting or referring to the prior art. It is noted that those skilled in the art, having the benefit of the teachings of this specification, may effect these and other changes in a manner similar to the equivalent or obvious variations thereof. All such variations are intended to be within the scope of the present invention.
Claims (7)
1. A reservoir reserve prediction method based on a deep spatiotemporal attention network, the method comprising:
step 1), extracting the number of well openings, the number of oil reservoirs and the oil reservoir reserves of the historical exploration from a database and carrying out normalization;
step 2), designing a deep space-time attention network, including a time attention network, a space attention network and an LSTM main network;
step 3), establishing a loss function and a proportionality coefficient between the oil reservoir reserves and the oil reservoir number loss function
Step 4), training the network;
and 5) obtaining a trained model, and predicting the year to be predicted by using the model.
2. The method for predicting the reserves of the oil reservoir based on the deep spatiotemporal attention network as claimed in claim 1, which is characterized in that: and normalizing the data in the step 1) to enhance the fitting capability of the network.
3. The method for predicting the reserves of the oil reservoir based on the deep spatiotemporal attention network as claimed in claim 1, which is characterized in that: and 3) carrying out weighted optimization on the input well probing number by the depth time attention network designed in the step 2).
4. The method for predicting the reserves of the oil reservoir based on the deep spatiotemporal attention network as claimed in claim 1, which is characterized in that: and 3) carrying out weighted optimization on the intermediate hidden state generated by the LSTM by the depth space attention network designed in the step 2).
5. The method for predicting the reserves of the oil reservoir based on the deep spatiotemporal attention network as claimed in claim 1, which is characterized in that: and 2) the fusion method between the deep space-time network and the LSTM main network reduces the influence of fluctuation data on the result.
6. The method for predicting the reserves of the oil reservoir based on the deep spatiotemporal attention network as claimed in claim 1, which is characterized in that: the training mode is a classic deep learning training method.
7. The method for predicting the reserves of the oil reservoir based on the deep spatiotemporal attention network as claimed in claim 1, which is characterized in that: the whole network is realized end to end in the training and testing process.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911343248.7A CN112052977A (en) | 2019-12-24 | 2019-12-24 | Oil reservoir reserve prediction method based on deep space-time attention network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911343248.7A CN112052977A (en) | 2019-12-24 | 2019-12-24 | Oil reservoir reserve prediction method based on deep space-time attention network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112052977A true CN112052977A (en) | 2020-12-08 |
Family
ID=73609585
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911343248.7A Pending CN112052977A (en) | 2019-12-24 | 2019-12-24 | Oil reservoir reserve prediction method based on deep space-time attention network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112052977A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113379164A (en) * | 2021-07-16 | 2021-09-10 | 国网江苏省电力有限公司苏州供电分公司 | Load prediction method and system based on deep self-attention network |
| CN113435662A (en) * | 2021-07-14 | 2021-09-24 | 中国石油大学(华东) | Water-drive reservoir yield prediction method and device and storage medium |
-
2019
- 2019-12-24 CN CN201911343248.7A patent/CN112052977A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113435662A (en) * | 2021-07-14 | 2021-09-24 | 中国石油大学(华东) | Water-drive reservoir yield prediction method and device and storage medium |
| CN113379164A (en) * | 2021-07-16 | 2021-09-10 | 国网江苏省电力有限公司苏州供电分公司 | Load prediction method and system based on deep self-attention network |
| CN113379164B (en) * | 2021-07-16 | 2024-03-26 | 国网江苏省电力有限公司苏州供电分公司 | Load prediction method and system based on deep self-attention network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109611087B (en) | Volcanic oil reservoir parameter intelligent prediction method and system | |
| CN103198354B (en) | One utilizes BP neural network and non-dominated sorted genetic algorithm NSGA2 to realize oil-field oil pumper oil recovery energy-saving and production-increase optimization method | |
| CN105760673A (en) | Fluvial facies reservoir earthquake sensitive parameter template analysis method | |
| CN111048163B (en) | Shale oil hydrocarbon retention amount (S1) evaluation method based on high-order neural network | |
| CN115375031A (en) | Oil production prediction model establishing method, capacity prediction method and storage medium | |
| CN112052977A (en) | Oil reservoir reserve prediction method based on deep space-time attention network | |
| CN113627639A (en) | Well testing productivity prediction method and system for carbonate fracture-cave reservoir | |
| KR101474874B1 (en) | computing system for well placement optimization developed by SA/ANN and well placement optimization method using Thereof | |
| CN116384554A (en) | Method and device for predicting mechanical drilling speed, electronic equipment and computer storage medium | |
| CN117541082B (en) | Comprehensive evaluation method based on oil reservoir-shaft-equipment evaluation index integration | |
| CN116911216B (en) | Reservoir oil well productivity factor assessment and prediction method | |
| CN110956388B (en) | Method for generating yield increasing scheme of offshore oil and gas reservoir | |
| CN118133104A (en) | Rapid identification method for lithofacies of deep sea-phase shale gas well | |
| Gudmundsdottir et al. | Inferring interwell connectivity in fractured geothermal reservoirs using neural networks | |
| Yan et al. | Research on transformer fault diagnosis based on an IWHO optimized MS1DCNN algorithm and LIF spectrum | |
| CN116415625A (en) | Logging curve characteristic learning method and system based on deep neural network | |
| Yang et al. | Research on intelligent identification technology of lithology wireless remote transmission based on element measurement | |
| CN117390973B (en) | Mine blasting hole utilization rate prediction method based on multilayer perceptron model | |
| CN118114812B (en) | Shale gas yield prediction method, computer equipment and storage medium | |
| CN113344729B (en) | Residual oil submergence digging method based on small sample learning | |
| CN116720662B (en) | Distributed energy system applicability evaluation method based on set pair analysis | |
| Ye et al. | A hybrid data-driven approach for shale gas production performance analysis via machine learning, game theory, and Pareto optimization: Weirong shale gas field, Sichuan Basin, China | |
| Na et al. | Optimization of Logging Interpretation Parameters Based on Committee Machine | |
| Zhao et al. | The fracture density and fractal dimension prediction based on support vector machine | |
| Huohai et al. | Injection-mining scheme optimization of underground gas storage based on agent model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20201208 |