New area oil reservoir key development index prediction method based on big data

Abstract

The invention provides a new area oil reservoir key development index prediction method based on big data, and the method comprises the steps: 1, analyzing key development indexes of a development scheme, and further analyzing main control factors affecting the development indexes; 2, developed oil reservoir instance data are collected, development indexes and main control factor data are extracted, and an initial sample set is constructed; step 3, adopting a clustering analysis algorithm and combining an oil reservoir type classification standard to carry out constraint, and carrying out sample data classification; 4, a multi-target recommendation algorithm is adopted for training, and an optimal prediction model of the development indexes is obtained; and 5, predicting key development indexes of the oil reservoir in the new area by using the prediction model. The new area oil reservoir key development index prediction method based on big data can accurately and intelligently predict the new area oil reservoir key development index by using new area oil reservoir development instance data, and improves the new area oil reservoir development success rate, guides new area productivity construction and obtains greater economic benefits by referring to successful development experience and development technology.

Description

Technical Field

[0001] The present invention relates to the technical field of artificial intelligence for reservoir development, and particularly to a method for predicting key development indicators of a new reservoir based on big data. Background Art

[0002] Oilfield development is a high-risk and high-investment task. In the past, business experts designed reservoir development plans and determined development indicators such as development methods and well pattern densities through reservoir engineering design or numerical simulation methods. Such methods take at least two months and at most more than half a year from data preparation to plan decision-making. Therefore, how to achieve rapid, correct, and accurate decision-making for the development of new reservoirs has long troubled oilfield enterprises and decision-makers. This patent uses past development examples and adopts deep learning calculation methods in a data-driven manner to achieve the prediction of key development indicators of new reservoirs.

[0003] In the Chinese patent application with the application number: CN201710092174.9, a method for optimizing the design of the economically optimal well pattern in a new offshore reservoir area is provided, including: Step 1, evaluating the reserves of a single sand body; Step 2, optimizing the development combination of producible sand bodies; Step 3, deploying the basic well pattern according to the reserve limit; Step 4, evaluating the economic benefits of a single well and optimizing the number and positions of economically efficient oil wells; Step 5, optimizing the maximum injection-production well spacing and optimizing the number and positions of the best water wells; Step 6, selecting the oil production process according to different wells to improve the productivity of a single well; Step 7, optimizing the scale, type, and position of the offshore platform. This patent uses an empirical method for the optimization design of the well pattern and relies on expert experience.

[0004] In the Chinese patent application with the application number: CN202110213855.2, a prediction analysis system for oilfield development indicators and its prediction method are provided. The system includes a data analysis and processing control unit and a database unit; the data analysis and processing control unit analyzes and processes the data in the database unit and establishes a model, and predicts the later-stage cumulative oil production N and recoverable oil volume N of the oilfield through model solution R ; the data analysis and processing control unit includes a data processing module, a prediction model establishment module, a model solution unit, and a model prediction unit. In this method, the prediction model uses the traditional empirical formula method to predict the cumulative oil production and recoverable oil reserves of the oilfield, and does not involve the prediction of other indicators such as well pattern density.

[0005] In the paper "Construction of a Rapid Evaluation Model for Development Indicators of Different Types of Reservoirs in New Areas of Sinopec" published on pages 77 - 81 of the journal "Oil & Gas Geology and Recovery Efficiency", in 2019, Volume 26, Issue 4, statistical analysis methods were used to deeply analyze the production capacity construction projects in new areas of Sinopec from 2007 to 2015, determine the main controlling factors affecting the design of key reservoir engineering parameters for development plans of 8 types of reservoirs, and establish an evaluation model for reservoir parameters and single - well production capacity development indicators corresponding to the 8 types of reservoirs. The method described in this literature mainly uses simple parameters such as reservoir types in established production capacity blocks to establish a conceptual design model library for development plans, establish a set of rapid evaluation models, and achieve rapid optimization of oilfield development planning and annual deployment. However, when using this method with a conceptual model library, attention must be paid to the reasonable scope of application during model application, and the selection of average reservoir parameters must be cautious, otherwise large errors will occur.

[0006] Through comprehensive analysis, it can be seen that the above - mentioned technologies all use traditional methods to design development plans and calculate development indicators, which are quite different from the technical idea of the present invention and fail to solve the technical problems we want to solve. Therefore, we have invented a method for predicting key development indicators of new - area reservoirs based on big data. Summary of the Invention

[0007] The purpose of the present invention is to provide a method that can utilize a large amount of development instance data of new - area reservoirs and can quickly and accurately predict the key development indicators of new - area reservoirs based on big - data methods.

[0008] The purpose of the present invention can be achieved by the following technical measures: A method for predicting key development indicators of new - area reservoirs based on big data, which includes:

[0009] Step 1: Analyze the key development indicators of the development plan and further analyze the main controlling factors affecting the development indicators;

[0010] Step 2: Collect the data of developed reservoir instances, extract the development indicator and main controlling factor data, and construct an initial sample set;

[0011] Step 3: Use the clustering analysis algorithm and combine it with the reservoir - type classification standard for constraint to classify the sample data;

[0012] Step 4: Use the multi - objective recommendation algorithm for training to obtain the optimal prediction model for development indicators;

[0013] Step 5: Use the prediction model to predict the key development indicators of new - area reservoirs.

[0014] The purpose of the present invention can also be achieved by the following technical measures:

[0015] In Step 1, by researching the experts for developing the plan and the experts for reviewing the plan, the key development indicators of the new area reservoir development plan are determined. The key development indicators include well pattern density, single well productivity, and ultimate recovery factor.

[0016] In Step 1, by collecting the examples of developing the new area reservoir development plan and conducting statistical analysis, the main controlling factors affecting the key development indicators are obtained. The initially selected main controlling factors include reservoir type, reservoir physical properties, fluid physical properties, oil reservoir properties, and economy. These factors have an impact.

[0017] In Step 1, in order to ensure the effect of model training, the characteristic parameters with high correlation with the target are screened through a radar chart, and finally the main controlling factors are determined to include reservoir type, crude oil viscosity, oil-bearing area, geological reserves, reservoir permeability, crude oil price, etc.

[0018] In Step 1, the classification criteria for reservoir types are usually considered from aspects such as reservoir properties and fluid properties based on the purpose of development, mainly including characteristic parameters such as reservoir lithology, porosity, permeability, and crude oil viscosity.

[0019] In Step 2, collect the new area reservoir development plans in recent years, conduct in-depth analysis, configure extraction rules, extract the key development indicators and their main influencing factors from the reservoir development examples and store them to construct a sample set.

[0020] In Step 2, technologies such as natural language understanding, context understanding, and picture OCR recognition are used to conduct in-depth analysis of the documents.

[0021] In Step 3, in order to improve the accuracy of the model, the K-Means clustering algorithm based on theoretical constraints is used for clustering analysis to lay a foundation for establishing a prediction model for classification.

[0022] In Step 3, collect the industry standards for dividing reservoir types by parameters such as permeability and viscosity, and build the judgment criteria into the K-Means clustering machine learning algorithm to achieve theoretical constraints on the classification results and improve the classification reliability.

[0023] In Step 4, for each type of sample in Step 3, the multi-objective prediction deep learning algorithm is used. The main controlling factors such as crude oil viscosity, oil-bearing area, geological reserves, reservoir permeability, and crude oil price are used as characteristic parameters, and the development indicators such as well pattern density, single well productivity, and ultimate recovery factor are used as labels. The multi-task network structure based on MMOE is used for training to obtain the optimal prediction model for each type.

[0024] In Step 5, cluster the target new area reservoir, select the corresponding prediction model, quickly predict the development indicators of the target block reservoir development plan, and guide the development.

[0025] The purpose of the present invention can also be achieved through the following technical measures: a key development indicator prediction system for new oil reservoirs based on big data, which adopts a key development indicator prediction method and labeling method for new oil reservoirs based on big data to predict key development indicators of new oil reservoirs.

[0026] The method for predicting key development indicators of new oil reservoirs based on big data in the present invention uses big data and data mining technology, based on a sample set constructed based on developed oil reservoir examples, and through clustering machine learning and multi-task deep learning algorithms based on theoretical constraints, to achieve fast and accurate intelligent recommendation of new oil reservoir development indicators. The method for predicting key development indicators of new oil reservoirs based on big data adopts document deep parsing and key parameter automatic extraction technology to automatically extract reservoir parameters and key development indicators, avoiding the huge manual workload caused by manual data collection. The clustering analysis algorithm based on theoretical model constraints is adopted, which fully considers theoretical experience and improves the reliability of automatic classification of similar samples, laying a foundation for subsequent improvement of model training accuracy. The multi-objective deep learning algorithm based on the MMOE model framework is used to train the model, comprehensively consider the interdependence between labels, optimize the prediction at the same time, and improve the model accuracy. The training model is used to predict the target block, and the key development indicator parameters are intelligently recommended. The recommended results have a compliance rate of more than 90% on the validation set. The prediction results are consistent with the experience of business experts, and the compliance rate with traditional methods is more than 80%. The present invention can utilize a large amount of new area oil reservoir development case data to quickly and accurately intelligently predict key development indicators of new area oil reservoirs, draw on successful development experience and development technology, improve the success rate of new area oil reservoir development, guide new area production capacity construction, and achieve greater economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 The present invention is a flowchart of a specific embodiment of the method for predicting key development indicators of new oil reservoirs based on big data. DETAILED DESCRIPTION

[0028] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0029] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations and / or combinations thereof.

[0030] As Figure 1 shown, Figure 1 Figure 1 Figure 1 is a flowchart of the method for predicting key development indicators of a new oil reservoir based on big data according to the present invention. The method for predicting key development indicators of a new oil reservoir based on big data includes:

[0031] Step 1, to formulate a scientific and reasonable development plan, analyze the key development indicators of the development plan, and further analyze the main controlling factors affecting the development indicators;

[0032] Step 2, collect the data of developed oil reservoir examples, extract the development indicator and main controlling factor data, and construct an initial sample set;

[0033] Step 3, adopt a clustering analysis algorithm, and combine with the oil reservoir type classification standard for constraint to realize the automatic classification of sample data;

[0034] Step 4, adopt a multi-objective recommendation algorithm for training to obtain an optimal prediction model for development indicators;

[0035] Step 5, use the prediction model to quickly predict the key development indicators of the new oil reservoir, realize intelligent recommendation, and guide the development.

[0036] In Step 1, by investigating the experts for plan compilation and plan review, determine the key development indicators of the new oil reservoir development plan, including well pattern density, single well productivity, ultimate recovery factor, etc. By collecting the examples of new oil reservoir development plan compilation, statistically analyze to obtain the main controlling factors affecting the key development indicators. The initially selected main controlling factors include factors such as oil and gas reservoir type, reservoir physical properties, fluid physical properties, oil layer properties, and economy. In order to ensure the effect of model training, screen the characteristic parameters with high correlation with the target through a radar chart. Finally, determine that the main controlling factors include oil and gas reservoir type, crude oil viscosity, oil-bearing area, geological reserves, reservoir permeability, crude oil price, etc. Among them, the classification standard of oil and gas reservoir type usually considers aspects such as reservoir properties and fluid properties, and mainly includes characteristic parameters such as reservoir lithology, porosity, permeability, and crude oil viscosity.

[0037] In Step 2, collect the new oil reservoir development plans of Shengli Oilfield in recent years, use technologies such as natural language understanding, context understanding, and picture OCR recognition to deeply analyze the documents, configure extraction rules, extract the key development indicators and their main influencing factors from the oil reservoir development examples and store them to construct a sample set.

[0038] In step 3, the reservoir instance data is distributed throughout the Shengli oilfield and involves a variety of reservoir types. In order to improve the accuracy of the model, the K-Means clustering algorithm based on theoretical constraints is used for clustering analysis to lay the foundation for the establishment of a prediction model for classification. The industry standards for classifying reservoir types based on parameters such as permeability and viscosity are collected, and the judgment criteria are built into the K-Means clustering machine learning algorithm to implement theoretical constraints on the classification results and improve the reliability of classification.

[0039] In step 4, for each type of sample in step 3, a multi-objective prediction deep learning algorithm is used. The main controlling factors such as crude oil viscosity, oil-bearing area, geological reserves, reservoir permeability, and crude oil price are used as feature parameters, and development indicators such as well network density, single well production capacity, and ultimate recovery rate are used as labels. The multi-task network structure based on MMOE is used for training to obtain the optimal prediction model of each type.

[0040] In step 5, the target new area reservoirs are clustered, the corresponding prediction model is automatically selected, and the development indicators of the target block reservoir development plan are quickly predicted to guide the development.

[0041] Compared with traditional reservoir engineering design and numerical simulation methods, this invention draws on the experience of previous development cases, fully considers industry experience models and theoretical understanding, and the intelligent prediction of development indicators has a more scientific basis. It has been verified that the prediction results can meet the requirements of engineering applications and improve work efficiency by more than 3 times compared with traditional research methods.

[0042] The following are several specific embodiments of the present invention.

[0043] Example 1

[0044] In the specific embodiment 1 of the present invention, the method for predicting key development indicators of new oil reservoirs based on big data includes the following steps:

[0045] 1. Convene business experts, organize discussions, and determine key development indicators and major influencing factors. Key development indicators include well pattern density, single well production capacity, ultimate recovery rate, etc., and major influencing factors include oil and gas reservoir type, crude oil viscosity, oil-bearing area, geological reserves, reservoir permeability, crude oil price, etc.

[0046] 2. Collect reservoir development plans for the new area of Shengli Oilfield in recent years, extract basic parameters and key development indicator data of reservoir plans from reservoir development examples, and construct a sample set.

[0047] 3. Use a clustering analysis algorithm based on industry reservoir classification standards to automatically classify similar samples in instance samples.

[0048] 4. Adopt the MMOE multi-task framework and deep learning algorithm for multi-objective training to obtain an optimal prediction model for each type of sample.

[0049] 5. Cluster the target reservoir, automatically select the corresponding prediction model for prediction, realize the prediction of key development indicators, and guide the development.

[0050] Example 2

[0051] In a specific Example 2 of applying the present invention, the method for predicting key development indicators of a new district reservoir based on big data includes:

[0052] Step 110, analyze and determine the key development indicators and basic reservoir parameters. The present invention selects key development indicators including well pattern density, single-well productivity, ultimate recovery rate, etc., and the main influencing factors include reservoir type, crude oil viscosity, oil-bearing area, geological reserves, reservoir permeability, crude oil price, etc.

[0053] Step 120, collect the data of developed reservoir examples and construct a sample set.

[0054] In the case of the present invention, 108 new district productivity construction plans of Shengli Oilfield in recent years are selected, and document deep parsing and key parameter automatic extraction technologies are used to automatically extract reservoir parameters and key development indicators to form a sample set.

[0055] Step 130, adopt a clustering analysis algorithm constrained by industry classification standards to realize the automatic classification of reservoir types.

[0056] Step 140, train with a multi-objective deep learning algorithm based on the MMOE framework to obtain an optimal prediction model.

[0057] Step 150, select a reservoir in a certain block as the test target. The lithology of the reservoir of this target is glutenite, the porosity is 12.2%, the permeability is 9.6 mD, the crude oil viscosity is 36.3 mPa·s, and the oil-bearing area is 1.07 km 2 , and the geological reserves are 792,000 tons. Cluster the test target reservoir, select the corresponding prediction model, quickly predict the key development indicators of the new district reservoir, the predicted well pattern density is 11.2 wells / km 2 , the single-well productivity is 4.6 t / d, and the ultimate recovery rate is 10.2%. The predicted values can be used as a reference for the development of this target reservoir.

[0058] Example 3

[0059] In a specific Example 3 of applying the present invention, the method for predicting key development indicators of a new district reservoir based on big data includes:

[0060] Step 110, analyze and determine the key development indicators and basic reservoir parameters. In the present invention, the selected key development indicators include well pattern density, single well productivity, ultimate recovery factor, etc., and the main influencing factors include reservoir type, crude oil viscosity, oil-bearing area, geological reserves, reservoir permeability, crude oil price, etc.

[0061] Step 120, collect the data of developed reservoir examples and construct a sample set.

[0062] In the case of the present invention, 108 new area productivity construction plans of Shengli Oilfield in recent years are selected, and through document in-depth analysis and key parameter automatic extraction technology, reservoir parameters and key development indicators are automatically extracted to form a sample set.

[0063] Step 130, adopt a clustering analysis algorithm constrained by industry classification standards to realize the automatic classification of reservoir types.

[0064] Step 140, train with a multi-objective deep learning algorithm based on the MMOE framework to obtain an optimal prediction model.

[0065] Step 150, select a reservoir in a certain block as the test target. The lithology of the reservoir of this reservoir is fine sandstone, the porosity is 27.2%, the permeability is 1368.5 mD, the crude oil viscosity is 2 mPa·s, and the oil-bearing area is 0.62 km 2 , and the geological reserves are 357,000 tons. According to the characteristic parameters, the target reservoir is automatically clustered to select the corresponding prediction model, and the key development indicators of the new area reservoir are quickly predicted. The predicted well pattern density is 10.8 wells / km 2 , the single well productivity is 7.9 t / d, and the ultimate recovery factor is 54%. The predicted values can be used as a reference for the development of the target reservoir.

[0066] Finally, it should be noted that the above are only the preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

[0067] Except for the technical features described in the specification, the rest are the known technologies of those skilled in the art.

Claims

1. A method for predicting key development indicators of a new oil reservoir based on big data, characterized in that, The method for predicting key development indicators of a new - area reservoir based on big data includes: Step 1: Analyze the key development indicators of the development plan, and further analyze the main controlling factors affecting the development indicators; Step 2: Collect the data of developed reservoir examples, extract the development indicator data and main controlling factor data, and construct an initial sample set; Step 3: Use the clustering analysis algorithm, combined with the classification standard of reservoir types for constraint, to classify the sample data; Step 4: Use the multi - objective recommendation algorithm for training to obtain the optimal prediction model of the development indicators; Step 5: Use the prediction model to predict the key development indicators of the new - area reservoir.

2. The prediction method for key development indicators of a new oil reservoir based on big data according to claim 1, wherein In Step 1, by investigating the experts in program compilation and program review, determine the key development indicators of the new - area reservoir development plan, including well pattern density, single - well productivity, and ultimate recovery ratio.

3. The method for predicting key development indicators of a new oil reservoir based on big data according to claim 2, wherein In Step 1, by collecting the examples of new - area reservoir development plan compilation, statistically analyze to obtain the main controlling factors affecting the key development indicators. The initially selected main controlling factors include reservoir type, reservoir physical properties, fluid physical properties, oil layer properties, and economy.

4. The method for predicting key development indicators of a new oil reservoir based on big data according to claim 3, characterized in that, In Step 1, to ensure the effect of model training, screen the characteristic parameters with high correlation with the target through a radar chart, and finally determine that the main controlling factors include reservoir type, crude oil viscosity, oil - bearing area, geological reserves, reservoir permeability, and crude oil price.

5. The method for predicting key development indicators of a new oil reservoir based on big data according to claim 4, wherein In Step 1, the classification standard of reservoir types, based on the purpose of development, usually considers aspects such as reservoir properties and fluid properties, mainly including characteristic parameters such as reservoir lithology, porosity, permeability, and crude oil viscosity.

6. The method for predicting key development indicators of a new oil reservoir based on big data according to claim 1, characterized in that In Step 2, collect the new - area reservoir development plans in recent years, conduct in - depth analysis, configure extraction rules, extract the key development indicators and their main influencing factors from the reservoir development examples and store them to construct a sample set.

7. The method for predicting key development indicators of a new oil reservoir based on big data according to claim 6, characterized in that, In Step 2, use technologies such as natural language understanding, context understanding, and picture OCR recognition to conduct in - depth analysis of the document.

8. The method for predicting key development indicators of a new oil reservoir based on big data according to claim 1, wherein In Step 3, to improve the accuracy of the model, use the K - Means clustering algorithm based on theoretical constraints for clustering analysis, laying a foundation for establishing a prediction model for classification.

9. The method for predicting key development indicators of a new oil reservoir based on big data according to claim 8, wherein, In Step 3, collect the industry standards for dividing reservoir types with parameters such as permeability and viscosity, and build the judgment criteria into the K - Means clustering machine learning algorithm to achieve theoretical constraints on the classification results and improve the classification reliability.

10. The method for predicting key development indicators of a new oil reservoir based on big data according to claim 1, characterized in that, In Step 4, for each type of sample in Step 3, use the multi - objective prediction deep - learning algorithm. Take the main controlling factors such as crude oil viscosity, oil - bearing area, geological reserves, reservoir permeability, and crude oil price as characteristic parameters, and take the development indicators such as well pattern density, single - well productivity, and ultimate recovery ratio as labels. Use the multi - task network structure based on MMOE for training to obtain the optimal prediction model for each type.

11. The method for predicting key development indicators of a new oil reservoir based on big data according to claim 1, wherein In Step 5, cluster the target new - area reservoir, select the corresponding prediction model, quickly predict the development indicators of the development plan of the target block reservoir, and guide the development.

12. A new area reservoir key development index prediction system based on big data, characterized in that, The system for predicting key development indicators of a new - area reservoir based on big data uses the method for predicting key development indicators of a new - area reservoir based on big data described in any one of claims 1 - 11 to predict the key development indicators of the new - area reservoir.

Images