CN110598326A - Well testing interpretation method based on artificial intelligence - Google Patents
Well testing interpretation method based on artificial intelligence Download PDFInfo
- Publication number
- CN110598326A CN110598326A CN201910869000.8A CN201910869000A CN110598326A CN 110598326 A CN110598326 A CN 110598326A CN 201910869000 A CN201910869000 A CN 201910869000A CN 110598326 A CN110598326 A CN 110598326A
- Authority
- CN
- China
- Prior art keywords
- well testing
- layer
- artificial intelligence
- interpretation
- neural network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/14—Fourier, Walsh or analogous domain transformations, e.g. Laplace, Hilbert, Karhunen-Loeve, transforms
- G06F17/141—Discrete Fourier transforms
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/14—Fourier, Walsh or analogous domain transformations, e.g. Laplace, Hilbert, Karhunen-Loeve, transforms
- G06F17/148—Wavelet transforms
-
- 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)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Analysis (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Computational Mathematics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Algebra (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Discrete Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to a well testing interpretation method based on artificial intelligence, which belongs to the field of well testing analysis; the problems that the interpretation process is complicated, the efficiency is too slow, certain interpretation errors exist and the like in the existing well testing interpretation are solved; the technical scheme is as follows: preprocessing a pressure derivative based on filtering and denoising, compiling a program to process a training sample, establishing a well testing interpretation model based on a convolutional neural network, improving a genetic algorithm to improve the fitting speed and the parameter fitting accuracy, adjusting an optimized network structure by the model through experimental and theoretical researches, identifying a characteristic section of a derivative curve, and intelligently diagnosing the model by a system by combining the characteristics of a flowing section and the data trend of a well testing model; the invention carries out well testing analysis based on artificial intelligence, can automatically adjust, self-learn and associate, optimizes the working process, improves the working efficiency and decision quality, avoids the influence of subjective judgment and realizes full-automatic well testing interpretation.
Description
Technical Field
The invention relates to a well testing interpretation method based on artificial intelligence, and belongs to the field of well testing analysis of oil and gas reservoirs.
Background
The well testing is an important means of reservoir engineering, and is a method for researching oil and gas reservoir geology and oil and gas well engineering parameters by taking oil and gas seepage mechanics as a theoretical basis and taking pressure, temperature and yield tests as means. Namely, the wells (oil wells, gas wells and water wells) are tested, the pressure and yield changes of the wells (the oil wells, the gas wells and the water wells) caused by changing the working system are measured, and the stratum parameters, the productivity and the well completion quality of the test wells, the dynamic problems related to the oil reservoirs and the test wells are researched through the analysis of the change processes, so that the effect of the yield increasing and improving of the test wells is analyzed.
Compared with the domestic and foreign well testing analysis methods, the method has some advantages and disadvantages. The foreign well testing analysis method has the advantages that: the commercialization degree is high; the universality is emphasized, and the applicability is wide; integrating the functions of software; the numerical well testing function is achieved; and the software is maintained and updated timely. The disadvantages are that: the adaptability to the actual oil field in China is high; test methods and test process problems; testing data quality problems; the specificity of the actual formation; the specialization of the mining conditions. The domestic well testing analysis method has the advantages that: emphasis is placed on specific problem or project development; strong pertinence and strong adaptability. The disadvantages are that: the study of unconventional well testing technology is not considered; the development scale is small, and the expandability is poor; the functions are single, and the number of models is small; software maintenance and update lags behind the development of computer software technology.
The application of traditional programming in well test interpretation, well test evaluation is quite common. Since the introduction of computer technology into well test interpretation, well test interpretation has been a combination of automated and manual processing. Most of the existing well testing interpretation means are manual and computer-aided modes, but most of the existing well testing interpretation means have the limitations of complicated interpretation process, low efficiency, certain interpretation errors and the like due to experience and other reasons in the artificial fitting process, and the existing auxiliary software is written by non-intelligent programs, does not have the functions of self-adjustment, self-learning and self-association, is only limited to solve a part of work which is easy to do, only can help a field engineer to accelerate the calculation process of the field engineer, and has poor solution uniqueness. The most difficult part, model identification, injury identification and solution uniqueness problems, can only be accomplished by a very small number of experts. Along with the development of artificial intelligence and the innovation of oil field intellectualization, the artificial intelligence technology can optimize the working process, improve the working efficiency and the decision quality, avoid the influence of subjective judgment and realize full-automatic well test interpretation, so that the well test interpretation method based on the artificial intelligence has great practical significance.
Disclosure of Invention
The invention aims to: in order to solve the problems of complicated interpretation process, low efficiency, certain interpretation error and the like in the existing well testing interpretation, the invention carries out well testing analysis based on artificial intelligence, can self-regulate, self-learn and self-associate, optimizes the working process, improves the working efficiency and decision quality, avoids the influence of subjective judgment and realizes full-automatic well testing interpretation.
In order to achieve the purpose, the invention provides a well testing interpretation method based on artificial intelligence, which comprises the following steps: introducing measured pressure data and yield data, processing the data based on filtering and denoising, and preprocessing a pressure derivative by improving and optimizing a wavelet transform algorithm to form a continuous smooth diagnosis curve; implementing a complete convolutional neural network with Tensorflow, using the convolutional neural network to identify a handwritten digit data set (MNIST); writing a program to process a training sample, and establishing a well testing interpretation model based on a convolutional neural network; the genetic algorithm is improved to improve the fitting speed and the parameter fitting accuracy, the model adjusts an optimized network structure through experimental and theoretical research, the characteristic section of a derivative curve is identified, and the model is intelligently diagnosed by the system by combining the flowing section characteristic and the data trend of the well testing model.
In the well testing interpretation method based on artificial intelligence, Fourier transform is adopted for preprocessing the pressure derivative, and characteristic quantities independent of the translational change of well testing data are extracted.
In the above well testing interpretation method based on artificial intelligence, the main steps of establishing the well testing interpretation model based on the convolutional neural network include: defining Weight and bias of convolutional layer; defining a pooling layer; building a convolution layer; establishing a full connection layer; establishing an output layer; an optimization method; and carrying out sample training and establishing a well testing interpretation model based on the convolutional neural network.
In the above well testing interpretation method based on artificial intelligence, the step of establishing the convolutional layer mainly comprises: defining a first layer of convolution, and defining Weight of the layer; defining bias; defining a first convolutional layer h _ conv1 of the convolutional neural network as conv2d (x _ image, W _ conv1) + b _ conv1, and performing nonlinear processing, namely, activating function processing on h _ conv 1; carrying out pooling treatment; in the same way, a second convolutional layer is defined, the input of this layer being the output of the above pooling layer; then, defining a second convolution layer of the convolution neural network; and finally performing pooling operation.
In the above well testing interpretation method based on artificial intelligence, the optimization method defines the cost function by using the cross entropy loss function, and optimizes by using tf.
Compared with the prior art, the invention has the following beneficial effects: (1) the well testing interpretation process is simple, the efficiency is high, and the interpretation error is small; (2) well testing analysis is carried out based on artificial intelligence, and self-adjustment, self-learning and self-association can be realized; (3) the working flow is optimized, the working efficiency and the decision quality are improved, the influence of subjective judgment is avoided, and the full-automatic well testing interpretation is realized.
Drawings
In the drawings:
FIG. 1 is a schematic diagram of a convolutional neural network.
FIG. 2 is a schematic diagram of the structure of the convolutional layer and the pooling layer.
FIG. 3 is a diagram of a software extraction data interface programmed using the method provided by the present invention.
FIG. 4 is a diagram of a software training interface programmed using the method provided by the present invention.
FIG. 5 is a diagram of a software fit interpretation interface programmed using the method provided by the present invention.
Detailed Description
The present invention will be further described with reference to the following embodiments and drawings.
The invention provides a well testing interpretation method based on artificial intelligence, which comprises the following steps: introducing measured pressure data and yield data, processing the data based on filtering and denoising, and preprocessing a pressure derivative by improving and optimizing a wavelet transform algorithm to form a continuous smooth diagnosis curve; implementing a complete convolutional neural network with Tensorflow, using the convolutional neural network to identify a handwritten digit data set (MNIST); writing a program to process a training sample, and establishing a well testing interpretation model based on a convolutional neural network; the genetic algorithm is improved to improve the fitting speed and the parameter fitting accuracy, the model adjusts an optimized network structure through experimental and theoretical research, the characteristic section of a derivative curve is identified, and the model is intelligently diagnosed by the system by combining the flowing section characteristic and the data trend of the well testing model.
In the well testing interpretation method based on artificial intelligence, Fourier transform is adopted for preprocessing the pressure derivative, and characteristic quantity independent of the translational change of well testing data is extracted.
In the well testing interpretation method based on artificial intelligence, the main steps of establishing the well testing interpretation model based on the convolutional neural network comprise: defining Weight and bias of convolutional layer; defining a pooling layer; building a convolution layer; establishing a full connection layer; establishing an output layer; an optimization method; and carrying out sample training and establishing a well testing interpretation model based on the convolutional neural network.
The invention mainly establishes a well testing interpretation model based on a Convolutional Neural Network (CNN), as shown in figure 1, the convolutional neural network is one of deep learning, and the weight value of the convolutional neural network shares a network structure to make the convolutional neural network more similar to a biological neural network, thereby reducing the complexity of the network model and the number of the weight values. The first few layers are usually alternating convolutional layers and downsampled layers, and the last few layers near the output layer are usually fully connected networks. The training process of the convolutional neural network mainly learns network parameters such as convolutional kernel parameters and interlayer connection weights of convolutional layers, and the prediction process mainly calculates category labels based on input images and the network parameters.
The basic structure of the CNN is composed of an input layer, a convolutional layer (convolutional layer), a pooling layer (also called a sampling layer), a full link layer, and an output layer. Fig. 2 is a schematic diagram showing the structure of the convolutional layer and the pooling layer of the one-dimensional CNN, where the topmost layer is the pooling layer, the middle layer is the convolutional layer, and the bottommost layer is the input layer of the convolutional layer.
In the well testing interpretation method based on artificial intelligence, the step of establishing the volume layer mainly comprises the following steps: defining a first layer of convolution, and defining Weight of the layer; defining bias; defining a first convolutional layer h _ conv1 of the convolutional neural network as conv2d (x _ image, W _ conv1) + b _ conv1, and performing nonlinear processing, namely, activating function processing on h _ conv 1; carrying out pooling treatment; in the same way, a second convolutional layer is defined, the input of this layer being the output of the above pooling layer; then, defining a second convolution layer of the convolution neural network; and finally performing pooling operation.
Pooling (Pooling) is an important operation in convolutional neural networks that enables feature reduction while maintaining local invariance of features. The convolution kernel and pooling kernel in CNN are equivalent to the realization of the receptive field in the Hubel-Wiesel model in engineering, the convolution layer is used for simulating simple cells of the Hubel-Wiesel theory, and the pooling layer simulates complex cells of the theory. The size (number of neurons) DWindow of each output feature plane of each pooling layer in CNN is:
in equation (1), the size of the pooling kernel is DWindow, and the pooling layer reduces the amount of computation of the network model by reducing the number of connections between convolution layers, i.e., by pooling to reduce the number of neurons.
In the well testing interpretation method based on artificial intelligence, the step of establishing the full connection layer mainly comprises the following steps: defining a full connection layer; multiplying the flattened h _ pool2_ flat with W _ fc1 of the layer; consider the problem of overfitting, add a dropout process.
After passing through a plurality of convolutional layers and pooling layers, 1 or more than 1 fully-connected layer is connected, similar to the MLP, and each neuron in the fully-connected layer is fully connected with all neurons in the layer before the neuron.
In the well testing interpretation method based on artificial intelligence, the step of establishing the output layer mainly comprises the following steps: and constructing an output layer. The input is 1024, the final output is 10 (since the mnist dataset is ten classes 0-9), the prediction is the final predictor; the output is classified with a softmax classifier (multi-classification, output is probability of individual classes).
The invention relates to a well testing interpretation method based on artificial intelligence, wherein an optimization method defines a cost function by using a cross entropy loss function, and optimizes by using tf.
In the well testing interpretation method based on artificial intelligence, the training steps mainly comprise: defining session and initializing all variables; the model was trained 1000 times, and checked every 50 times for accuracy.
And finally, writing well testing interpretation software by using a C + + and Python integration module, wherein an extracted data interface diagram, a software training interface diagram and a fitting interpretation interface diagram are respectively shown in fig. 3, 4 and 5.
Compared with the prior art, the invention has the following beneficial effects: (1) the well testing interpretation process is simple, the efficiency is high, and the interpretation error is small; (2) well testing analysis is carried out based on artificial intelligence, and self-adjustment, self-learning and self-association can be realized; (3) the working flow is optimized, the working efficiency and the decision quality are improved, the influence of subjective judgment is avoided, and the full-automatic well testing interpretation is realized.
Claims (5)
1. A well testing interpretation method based on artificial intelligence is characterized by comprising the following steps:
importing measured pressure data and output data, and preprocessing a pressure derivative based on filtering and denoising to form a continuous smooth diagnosis curve;
implementing a complete convolutional neural network with Tensorflow, using the convolutional neural network to identify a handwritten digit data set (MNIST);
writing a program to process a training sample, and establishing a well testing interpretation model based on a convolutional neural network;
the genetic algorithm is improved to improve the fitting speed and the parameter fitting accuracy, the model adjusts an optimized network structure through experimental and theoretical research, the characteristic section of a derivative curve is identified, and the model is intelligently diagnosed by the system by combining the flowing section characteristic and the data trend of the well testing model.
2. The artificial intelligence based well testing interpretation method of claim 1, wherein: the preprocessing of the pressure derivative adopts Fourier transform, and characteristic quantity independent of the translational change of well testing data is extracted.
3. The artificial intelligence based well testing interpretation method of claim 1, wherein: the main steps of establishing the well testing interpretation model based on the convolutional neural network comprise:
defining Weight and bias of convolutional layer;
defining a pooling layer;
building a convolution layer;
establishing a full connection layer;
establishing an output layer;
an optimization method;
and carrying out sample training and establishing a well testing interpretation model based on the convolutional neural network.
4. The artificial intelligence based well testing interpretation method of claim 3, wherein: the steps of establishing the convolutional layer mainly comprise:
defining a first layer of convolution, and defining Weight of the layer;
defining bias;
defining a first convolutional layer h _ conv1 of the convolutional neural network as conv2d (x _ image, W _ conv1) + b _ conv1, and performing nonlinear processing, namely, activating function processing on h _ conv 1;
carrying out pooling treatment;
in the same way, a second convolutional layer is defined, the input of this layer is the output of the previous pooling layer;
then, defining a second convolution layer of the convolution neural network;
finally, performing pooling operation, and so on.
5. The artificial intelligence based well testing interpretation method of claim 3, wherein: the optimization method is to define the cost function by using a cross entropy loss function, and optimize by using tf.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910869000.8A CN110598326A (en) | 2019-09-16 | 2019-09-16 | Well testing interpretation method based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910869000.8A CN110598326A (en) | 2019-09-16 | 2019-09-16 | Well testing interpretation method based on artificial intelligence |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110598326A true CN110598326A (en) | 2019-12-20 |
Family
ID=68859678
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910869000.8A Pending CN110598326A (en) | 2019-09-16 | 2019-09-16 | Well testing interpretation method based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110598326A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112926805A (en) * | 2021-04-12 | 2021-06-08 | 中国石油大学(北京) | Intelligent vertical well testing interpretation method and device based on deep reinforcement learning |
| CN113807532A (en) * | 2020-06-12 | 2021-12-17 | 百度（美国）有限责任公司 | AI model transfer method using layer randomization |
| CN116341393A (en) * | 2023-05-26 | 2023-06-27 | 中国石油大学(华东) | Automatic unsteady state well test interpretation method, device, equipment and medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109360120A (en) * | 2018-10-22 | 2019-02-19 | 常州大学 | A kind of electric immersible pump well operating condition intelligent diagnosing method based on convolutional neural networks |
| EP3493120A1 (en) * | 2017-12-01 | 2019-06-05 | Koninklijke Philips N.V. | Training a neural network model |
-
2019
- 2019-09-16 CN CN201910869000.8A patent/CN110598326A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3493120A1 (en) * | 2017-12-01 | 2019-06-05 | Koninklijke Philips N.V. | Training a neural network model |
| CN109360120A (en) * | 2018-10-22 | 2019-02-19 | 常州大学 | A kind of electric immersible pump well operating condition intelligent diagnosing method based on convolutional neural networks |
Non-Patent Citations (4)
| Title |
|---|
| 刘启国等: "开关井不同程度应力敏感效应对井底压力动态的影响", 《石油与天然气地质》 * |
| 强琳等: "应用人工神经网络识别试井解释模型", 《西安石油学院学报》 * |
| 成绥民等: "神经网络专家系统在试井解释中的应用", 《西安石油学院学报(自然科学版)》 * |
| 段友祥等: "卷积神经网络在储层预测中的应用研究", 《通信学报》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113807532A (en) * | 2020-06-12 | 2021-12-17 | 百度（美国）有限责任公司 | AI model transfer method using layer randomization |
| CN113807532B (en) * | 2020-06-12 | 2024-06-07 | 百度（美国）有限责任公司 | AI model transfer method using layer randomization |
| CN112926805A (en) * | 2021-04-12 | 2021-06-08 | 中国石油大学(北京) | Intelligent vertical well testing interpretation method and device based on deep reinforcement learning |
| CN112926805B (en) * | 2021-04-12 | 2023-11-10 | 中国石油大学(北京) | Intelligent interpretation method and device for vertical well test based on deep reinforcement learning |
| CN116341393A (en) * | 2023-05-26 | 2023-06-27 | 中国石油大学(华东) | Automatic unsteady state well test interpretation method, device, equipment and medium |
| CN116341393B (en) * | 2023-05-26 | 2023-08-04 | 中国石油大学(华东) | Automatic unsteady state well test interpretation method, device, equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109272123B (en) | Sucker-rod pump working condition early warning method based on convolution-circulation neural network | |
| CN111460728B (en) | Method and device for predicting residual life of industrial equipment, storage medium and equipment | |
| CN112284736B (en) | Convolutional neural network fault diagnosis method based on multi-channel attention module | |
| CN110598326A (en) | Well testing interpretation method based on artificial intelligence | |
| CN108932314A (en) | A kind of chrysanthemum image content retrieval method based on the study of depth Hash | |
| CN114282443B (en) | Residual service life prediction method based on MLP-LSTM supervised joint model | |
| CN115545321A (en) | On-line prediction method for process quality of silk making workshop | |
| CN114218872B (en) | DBN-LSTM semi-supervised joint model-based residual service life prediction method | |
| CN108921285A (en) | Single-element classification method in sequence based on bidirectional valve controlled Recognition with Recurrent Neural Network | |
| CN112861912A (en) | Deep learning-based method and system for identifying indicator diagram of complex working condition of pumping well | |
| CN115329853B (en) | Equipment parameter prediction and knowledge transfer method based on multi-source domain migration | |
| CN112539054A (en) | Production optimization method for ground pipe network and underground oil reservoir complex system | |
| CN111649779A (en) | Oil well oil content and total flow rate measuring method based on dense neural network and application | |
| CN109444604A (en) | A kind of DC/DC converter method for diagnosing faults based on convolutional neural networks | |
| CN117277279A (en) | Deep learning short-term load prediction method based on particle swarm optimization | |
| CN108596044B (en) | Pedestrian detection method based on deep convolutional neural network | |
| CN112664185A (en) | Indicator diagram-based rod-pumped well working condition prediction method | |
| CN115860197A (en) | Data-driven coal bed gas yield prediction method and system | |
| CN110533249A (en) | A kind of smelter energy consumption prediction technique based on integrated shot and long term memory network | |
| CN114202060A (en) | Method for predicting methylene blue adsorption performance of biomass activated carbon based on deep neural network | |
| CN111639609A (en) | Intelligent identification system for metal fracture types based on machine vision and deep learning | |
| CN116821828A (en) | Multi-dimensional time sequence prediction method based on industrial data | |
| CN116247658A (en) | Photovoltaic power generation amount prediction method based on DLRM deep learning model | |
| Hu et al. | Research on the fault identification method of oil pumping unit based on residual network | |
| CN114897139B (en) | Bearing fault diagnosis method for ordered stable simplified sparse quantum neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20191220 |