An intelligent identification method of wellbore fluid flow pattern based on DAS monitoring

Abstract

The present invention discloses a method for intelligently identifying the flow state of wellbore fluid based on DAS monitoring, which belongs to the technical field of petroleum exploration and production, and includes the following steps: designing a simulated physical experimental device, laying optical fibers on the wellbore; adjusting the wellbore angle and flow controller; using DAS to collect acoustic data in the wellbore and preprocessing the data; comparing the waterfall chart generated by DAS data with different observed flow states, and obtaining the flow law of wellbore fluid under the influence of different factors; using a feature extraction algorithm to extract flow state characteristic parameters; using a machine learning algorithm to process the feature set, constructing a wellbore flow state intelligent identification model, and forming a method for intelligently identifying the flow state of wellbore fluid. The present invention adopts the above-mentioned method for intelligently identifying the flow state of wellbore fluid based on DAS monitoring to realize real-time dynamic monitoring of the flow of fluid in the wellbore throughout the life cycle of oil and gas wells, provide technical support for downhole multiphase fluid flow monitoring and identification, and bring economic benefits and safety guarantees to the oil and gas production industry.

Description

Technical Field

[0001] The invention relates to the technical field of petroleum exploration and exploitation, and in particular to a method for intelligently identifying the flow state of wellbore fluid based on DAS monitoring. Background Art

[0002] In the field of oil and gas production, the monitoring of the flow state of multiphase fluid in the wellbore is crucial for the observation of the production status of oil and gas wells, the evaluation of production capacity and the dynamic analysis of reservoirs. Compared with traditional multiphase flow monitoring methods, distributed optical fiber acoustic wave monitoring (DAS) technology has the advantages of simple structure, large dynamic range, high sensitivity, ability to provide real-time and continuous measurement data, no need to preheat or separate the measured medium, and low equipment maintenance requirements and labor costs. Unstable flow state in the wellbore may lead to safety problems. For example, slug flow may cause pressure fluctuations and unstable production of oil and gas wells, and may even lead to accidents such as blowouts in severe cases. Therefore, it is necessary to use distributed optical fiber acoustic wave sensing (DAS) to monitor, identify and analyze the flow state of wellbore fluid, which has important theoretical value and practical significance for the intelligent identification of multiphase flow state monitoring in the wellbore. Real-time and continuous monitoring can also provide guidance for on-site safety operations. However, there is currently a lack of intelligent identification methods for real-time dynamic and continuous monitoring of wellbore fluid flow state during the entire life cycle of oil and gas wells in China. It is necessary to explore and form an intelligent identification method for wellbore fluid flow state based on DAS monitoring.

[0003] Many researchers have conducted extensive research on the problem of flow state monitoring and identification. With the advancement of technology, the method of identifying the flow state of fluid in the wellbore has been continuously developed and improved. In the 1970s, with the development of optical fiber communication technology, optical fiber sensing technology developed rapidly. At the same time, DAS technology has been widely used in the oil field, making great contributions to the progress of the oil industry. In the early 21st century, DAS monitoring of wellbore flow state identification began to be applied and studied. Misjudgment of wellbore flow state can cause huge losses, which may even be difficult to make up for. Therefore, accurately realizing real-time dynamic and continuous monitoring of fluid flow in the wellbore throughout the life cycle of oil and gas wells has immeasurable economic and practical value in engineering practice. However, the traditional flow state monitoring and identification method is difficult to deploy, has high maintenance costs, is easily affected by the environment, has low spatial coverage and data processing lags, making it difficult to achieve real-time dynamic and continuous monitoring of fluid flow in the wellbore throughout the life cycle of oil and gas wells. Summary of the invention

[0004] The purpose of the present invention is to provide a method for intelligently identifying the flow state of wellbore fluid based on DAS monitoring, so as to solve the problem of how to realize real-time dynamic and continuous monitoring of the flow conditions of fluid in the wellbore throughout the life cycle of oil and gas wells.

[0005] To achieve the above object, the present invention provides a method for intelligently identifying the flow state of wellbore fluid based on DAS monitoring, comprising the following steps:

[0006] S1. Design a simulated physical experimental device, lay optical fibers on the wellbore, and connect the DAS system module, wellbore, gas supply module, liquid supply module, and collection module;

[0007] S2, adjust the wellbore angle, start the operation console, industrial control computer, field instrument and software system, set the pneumatic valve, and adjust the flow controller;

[0008] S3, using DAS to collect acoustic data in the wellbore and preprocessing the collected data;

[0009] S4. Compare the waterfall diagram generated by DAS data with the different observed flow patterns to obtain the flow patterns of wellbore fluid under the influence of different factors;

[0010] S5. Based on the preprocessed data, flow characteristic parameters are extracted using a feature extraction algorithm;

[0011] S6. Use machine learning algorithms to test and verify the feature set, build a wellbore flow pattern intelligent identification model, and form a wellbore fluid flow pattern intelligent identification method.

[0012] Preferably, the specific operation of S1 is:

[0013] According to the designed simulated physical experimental device plan, optical fiber is laid on the wellbore, pulled into straight lines on the upper and lower surfaces of the pipeline, and wound along the pipeline wall. Each pipeline is wound in the order of lower layer-upper layer-spiral, and the DAS system modules are connected to the optical fiber and computer respectively.

[0014] Preferably, the specific operation of S2 is:

[0015] S21, start four power distribution cabinets to supply power to all electrical equipment of the test device, adjust the shaft to the required angle, and fix the shaft;

[0016] S22, set the pneumatic valve, adjust the flow controller, set the liquid volume to a fixed value, and the liquid volume remains unchanged during the simulation experiment;

[0017] S23. Use pneumatic valves to control the flow equipment to continuously adjust the flow ratio and set the air intake to keep the fluid stable in the pipeline for about 3 minutes.

[0018] Preferably, the specific operation of S3 is:

[0019] S31, reading data, and generating a waterfall chart with the data collected by DAS;

[0020] S32, using FK filtering to convert the DAS data from the time and spatial position of the recorded data in the optical cable, i.e., the corresponding time-space domain, into the frequency and wave number domain, wherein the FK filtering is implemented by a two-dimensional fast Fourier transform FFT algorithm, and the formula is:

[0021]

[0022] Where t represents the time of input data, x represents the position of input data, f represents the frequency, k 1 represents the wave number, i is the imaginary unit;

[0023] S33, using Min-Max normalization to normalize the data, the Min-Max normalization formula is expressed as:

[0024]

[0025] Among them, X is the original data, X min and X max are the minimum and maximum values ​​in the data set, respectively, and X norm is the normalized data.

[0026] Preferably, the specific operation of S4 is:

[0027] According to the different flow states observed in the simulation experiment, the corresponding DAS waterfall diagram and FK diagram are analyzed to qualitatively identify the flow state of the wellbore fluid. At the same time, the flow state of the fluid in the wellbore at different angles and different flow rates is analyzed, and the flow law of the fluid in the wellbore is obtained using DAS.

[0028] Preferably, the specific operation of S5 is:

[0029] S51, using wavelet transform to process the original signal to achieve low variance representation of the data;

[0030] S52. Apply the short-time Fourier transform (STFT) to the acoustic data so that the acoustic signal is divided into small parts. The STFT formula is expressed as:

[0031]

[0032] Where w(m) is a finite window selected from the sliding signal x(m+n), n represents the time index, k 2 is the frequency, m represents the different time offsets in the summation process, and N represents the number of samples.

[0033] Preferably, the specific operation of S6 is:

[0034] The feature set obtained after preprocessing is input into the convolutional neural network algorithm for testing and verification, and a wellbore flow pattern intelligent identification model is constructed to form an intelligent identification method for wellbore fluid flow pattern.

[0035] Therefore, the present invention adopts the above-mentioned intelligent identification method of wellbore fluid flow state based on DAS monitoring, which has the following beneficial effects:

[0036] (1) The present invention designs a simulation physical experiment scheme, which can simulate the changes in the flow of fluid in the wellbore as the wellbore angle, flow pattern, and flow rate change during the horizontal well production process. Based on this, DAS acquisition dynamic data is obtained, and the data is used to analyze the flow law of fluid in the wellbore under different influencing factors. A DAS data preprocessing and feature extraction method is proposed, and a wellbore flow pattern intelligent recognition model is constructed, forming a wellbore fluid flow pattern intelligent recognition method.

[0037] (2) The intelligent identification method of the present invention can realize real-time dynamic and continuous monitoring of fluid flow in the wellbore throughout the life cycle of oil and gas wells, provide technical support for downhole multiphase fluid flow monitoring and identification, provide guidance for on-site wellbore intelligent analysis, and bring economic benefits and safety guarantees to the oil and gas extraction industry.

[0038] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a schematic diagram of the steps of an embodiment of a method for intelligently identifying the flow state of wellbore fluid based on DAS monitoring of the present invention;

[0040] Figure 2 It is a flow chart of an embodiment of a method for intelligently identifying flow patterns of wellbore fluids based on DAS monitoring according to the present invention;

[0041] Figure 3 It is a schematic diagram of an optical fiber winding method on a wellbore according to an embodiment of a method for intelligently identifying the flow state of wellbore fluid based on DAS monitoring of the present invention;

[0042] Figure 4 It is a schematic diagram of a waterfall diagram generated by a DAS system of a wellbore fluid flow pattern intelligent identification method based on DAS monitoring in the embodiment of the present invention when a slug flow is at a wellbore angle of 15°;

[0043] Figure 5 It is a schematic diagram of a convolutional neural network (CNN) network architecture of an embodiment of a method for intelligently identifying wellbore fluid flow patterns based on DAS monitoring of the present invention. DETAILED DESCRIPTION

[0044] The technical solution of the present invention is further described below through the accompanying drawings and embodiments.

[0045] Unless otherwise defined, technical or scientific terms used in the present invention shall have the common meanings understood by one having ordinary skills in the field to which the present invention belongs.

[0046] like Figure 1 As shown, the present invention provides a method for intelligently identifying the flow state of wellbore fluid based on DAS monitoring, comprising the following steps:

[0047] S1. Design a simulated physical experimental device, lay optical fibers on the wellbore, and connect the DAS system module, wellbore, gas supply module, liquid supply module, and collection module. The specific operations are as follows:

[0048] According to the designed simulated physical experimental device plan, optical fiber is laid on the wellbore, pulled into straight lines on the upper and lower surfaces of the pipeline, and wound along the pipeline wall. Each pipeline is wound in the order of lower layer-upper layer-spiral, and the DAS system modules are connected to the optical fiber and computer respectively.

[0049] The air supply module includes a water pump, a water tank, a cooling tower, an air compressor, a large air storage tank, an air filter, and an air dryer. The liquid supply module includes an oil tank, a water tank, an oil-water mixing tank, a metering pump, a plunger pump, and a pressure regulating tank. The acquisition module includes a high-speed camera, a pressure sensor, a differential pressure sensor, a gas-liquid instantaneous metering tank, and a digital thermometer.

[0050] S2, adjust the wellbore angle, start the operation console, industrial control computer, field instrument and software system, set the pneumatic valve, adjust the flow controller, specifically including the following steps:

[0051] S21, start four power distribution cabinets to supply power to all electrical equipment of the test device, adjust the shaft to the required angle, and fix the shaft;

[0052] S22, set the pneumatic valve, adjust the flow controller, set the liquid volume to a fixed value, and the liquid volume remains unchanged during the simulation experiment;

[0053] S23. Use pneumatic valves to control the flow equipment to continuously adjust the flow ratio and set the air intake to keep the fluid stable in the pipeline for about 3 minutes.

[0054] S3. Use DAS to collect acoustic data in the wellbore and pre-process the collected data. The specific operations are as follows:

[0055] S31, reading data, and generating a waterfall chart with the data collected by DAS;

[0056] S32, using FK filtering to convert the DAS data from the time and spatial position of the recorded data in the optical cable, that is, the corresponding time-space domain, into the frequency and wave number domain. The FK filtering is implemented by a two-dimensional fast Fourier transform FFT algorithm, and the formula is:

[0057]

[0058] Where t represents the time of input data, x represents the position of input data, f represents the frequency, k 1 represents the wave number, i is the imaginary unit;

[0059] S33. Use Min-Max normalization to normalize the data. The Min-Max normalization formula is expressed as:

[0060]

[0061] Among them, X is the original data, X min and X max are the minimum and maximum values ​​in the data set, respectively, and X norm is the normalized data.

[0062] S4. Compare the waterfall chart generated by DAS data with the different observed flow patterns to obtain the flow law of wellbore fluid under the influence of different factors. The specific operations are as follows:

[0063] According to the different flow states observed in the simulation experiment, the corresponding DAS waterfall diagram and FK diagram are analyzed to qualitatively identify the flow state of the wellbore fluid. At the same time, the flow state of the fluid in the wellbore at different angles and different flow rates is analyzed, and the flow law of the fluid in the wellbore is obtained using DAS.

[0064] S5. Based on the preprocessed data, the flow characteristic parameters are extracted using a feature extraction algorithm. The specific operations are as follows:

[0065] S51, using wavelet transform to process the original signal to achieve low variance representation of the data;

[0066] S52. Apply the short-time Fourier transform (STFT) to the acoustic data so that the acoustic signal is divided into small parts. The STFT formula is expressed as:

[0067]

[0068] Where w(m) is a finite window selected from the sliding signal x(m+n), n represents the time index, k 2 is the frequency, m represents the different time offsets in the summation process, and N represents the number of samples.

[0069] S6. Use machine learning algorithms to test and verify the feature set, build a wellbore flow pattern intelligent recognition model, and form a wellbore fluid flow pattern intelligent recognition method. The specific operations are:

[0070] The feature set obtained after preprocessing is input into the convolutional neural network algorithm for testing and verification, and a wellbore flow pattern intelligent identification model is constructed to form an intelligent identification method for wellbore fluid flow pattern.

[0071] Embodiment 1

[0072] like Figure 2 As shown, Figure 2 This is a flow chart of a method for intelligently identifying the flow state of wellbore fluids based on DAS monitoring provided by the present invention. First, according to the actual experimental requirements, a simulated physical experimental device is designed to carry out the experiment, optical fiber is laid on the wellbore, and the DAS system module, wellbore, gas supply module, liquid supply module and acquisition module are connected. Then, the wellbore angle is adjusted to 15° and 30°, and the fluid flow value required for the experiment is set.

[0073] Subsequently, DAS was used to collect acoustic data in the wellbore. After reading the data, a waterfall diagram was generated from the data collected by DAS. FK filtering was used to convert the DAS data from the time-space domain to the frequency-wave domain for noise reduction, and the data was normalized using Min-Max.

[0074] Then, according to the different flow state information recorded in the experiment, the waterfall diagram and FK diagram presented by the DAS system module after receiving the data information collected by the optical fiber are compared, and the flow state of the wellbore fluid is qualitatively identified while analyzing the flow law of the fluid in the wellbore under different influencing factors. Furthermore, based on the pre-processed data, wavelet transform (WT) and short-time Fourier transform (STFT) are applied to the acoustic data to extract the flow state characteristic parameters.

[0075] Finally, the feature set was input into the convolutional neural network algorithm for testing and verification, and a wellbore flow pattern intelligent recognition model was constructed, forming a wellbore fluid flow pattern intelligent recognition method. The above process constitutes a complete wellbore fluid flow pattern intelligent recognition method based on DAS monitoring.

[0076] Embodiment 2

[0077] The present invention provides a method for intelligently identifying the flow state of wellbore fluid based on DAS monitoring, which includes the following steps:

[0078] S1. Design a simulated physical experimental device, lay optical fibers on the wellbore, and connect the DAS system module, wellbore, gas supply module, liquid supply module, and collection module. The specific method is as follows:

[0079] (1) Figure 3 As shown, Figure 3This is a schematic diagram of the fiber winding method on the wellbore provided by the present invention. The experiment uses a single-core single-mode optical fiber with a total length of 103.4m. The optical fiber is arranged along the pipeline 1-2-3-4, and each pipeline is wound in the order of lower layer-upper layer-helix. Fixed winding of the optical fiber prevents excessive movement of the optical fiber and helps reduce unnecessary noise when acquiring data. The optical fiber joints must be protected because they are very fragile. They can be cleaned with a fiber end face cleaning pen before starting the experiment.

[0080] (2) Connect one end of the DAS system module to the optical fiber and the other end to the computer.

[0081] (3) The air supply module includes a water pump, a water tank, a cooling tower, an air compressor, a large air storage tank, an air filter, and an air dryer. The liquid supply module includes an oil tank, a water tank, an oil-water mixing tank, a metering pump, a plunger pump, and a pressure regulating tank. The acquisition module includes a high-speed camera, a pressure sensor, a differential pressure sensor, a gas-liquid instantaneous metering tank, and a digital thermometer.

[0082] (4) The main part of the experimental pipe section is composed of a 60 mm diameter organic glass tube. The gas-liquid two-phase flow experiment is carried out at room temperature, and the experimental media are water and air.

[0083] Table 1 Optical fiber winding length data

[0084]

[0085] S2. According to the simulated physical experimental device of S1, adjust the wellbore angle, start the operation console, industrial control computer, field instrument and software system, set the pneumatic valve, and adjust the flow controller. The specific method is as follows:

[0086] (1) Start four power distribution cabinets to supply power to all electrical equipment in the experimental device, adjust the shaft angle to 15°, fix the shaft, check whether the experimental device is normal, and prepare to start the experiment.

[0087] (2) Start the operation console, industrial control computer, field instruments and software system to prepare to observe and record flow conditions.

[0088] (3) Set the pneumatic valve, adjust the flow controller, and use the pneumatic valve opening setting control device to set the liquid volume to 0.2m 3 / h, after keeping the liquid volume stable, fill the pipeline with water and then let in air, and slowly increase the gas volume from a low level to the required level. When the fluid in the wellbore presents a bubbling flow, maintain it for about 3 minutes.

[0089] (4) While keeping the liquid volume unchanged, increase the air intake to keep the fluid stable in the pipeline. When the fluid in the wellbore presents slug flow and annular flow, maintain the flow for a period of time, observe and record.

[0090] S3. According to the experimental preparation process of S2, the acoustic data in the wellbore is collected by DAS, and the collected data is preprocessed. The specific method is as follows:

[0091] (1) Read the data and generate waterfall charts of different flow states using the data collected by DAS.

[0092] (2) The original acoustic data set in the time domain is converted into the frequency-wave number domain using the two-dimensional fast Fourier (2D-FFT) algorithm to achieve FK filtering and noise reduction. The two-dimensional fast Fourier (2D-FFT) formula is:

[0093]

[0094] Where t represents the time of input data, x represents the position of input data, f represents the frequency, k 1 represents the wave number, and i is the imaginary unit.

[0095] (3) Min-Max normalization is used to normalize the data, eliminate the dimension effect, and scale the data to the interval [0,1]. The Min-Max normalization formula is expressed as:

[0096]

[0097] Among them, X is the original data, X min and X max are the minimum and maximum values ​​in the data set, respectively, and X norm is the normalized data.

[0098] S4. Based on the data preprocessing in S3, the waterfall diagram generated by DAS data is compared with the different flow states observed to obtain the flow law of wellbore fluid under the influence of different factors. The specific method is as follows:

[0099] (1) According to the different flow patterns observed in the simulation experiment, the corresponding DAS waterfall diagram and FK diagram are analyzed to qualitatively identify the flow pattern of the wellbore fluid, such as Figure 4 As shown, Figure 4 It is a schematic diagram of a waterfall diagram generated by a DAS system provided by the present invention when a slug flow is at a wellbore angle of 15°.

[0100] (2) Based on the waterfall diagram generated by DAS data, it can be concluded that the microseismic differences of different flow types at different inclination angles and different flow rates are obvious, and the similarities of the same flow type at different angles are also obvious.

[0101] S5. According to the qualitative analysis of the waterfall chart in S4, the flow characteristic parameters are extracted using the feature extraction algorithm based on the preprocessed data. The specific method is as follows:

[0102] (1) The raw signal is processed using wavelet transform (WT) to achieve low-variance representations of the data. These low-variance representations are used as input to the classifier. The proposed framework has two wavelet filter banks, the first one has 8 wavelets per octave and the second one has 1 wavelet per octave.

[0103] (2) Apply the short-time Fourier transform (STFT) to the acoustic data to divide the acoustic signal into small parts so that they can be overlapped and windowed. Then perform DFT calculation on each part separately, and the calculation formula is:

[0104]

[0105] Where w(m) is a finite window selected from the sliding signal x(m+n), n represents the time index, k 2 is the frequency, m represents different time offsets in the summation process, and N represents the number of samples. A sliding Hamming window can be used to reduce spectral leakage. The sliding Hamming window length is 512 samples, and the window overlaps with 511 samples in time.

[0106] S6. Based on the time-frequency feature extraction of S5, the feature set is tested and verified using a machine learning algorithm, a wellbore flow pattern intelligent recognition model is constructed, and a wellbore fluid flow pattern intelligent recognition method is formed. The specific method is as follows:

[0107] The feature set is input into the convolutional neural network algorithm to classify the bubble flow, slug flow, and annular flow data sets. The processed DAS data is divided into training set, validation set, and test set at 15%, 35%, and 50% to ensure that the model can effectively evaluate the data. Conventional CNN usually consists of input layer, convolution layer, activation layer, pooling layer, and output layer. Figure 5 Come and see, Figure 5 It is a diagram of the convolutional neural network (CNN) network architecture, which includes convolutional layers, group normalization layers, activation layers, anti-overfitting layers, pooling layers, and fully connected layers. These operations are repeated between layers, and each layer learns to monitor different input data features.

[0108] The network is trained by inputting DAS signals of different categories, initializing the weights of the convolutional layer, and using the Adam optimizer with default parameters. The parameter settings of the improved network include the number of classification groups, input data dimensions, and data length. The network takes the time series of the original DAS signal as input, but ensures that the input data length is the same, and the number can be different, and outputs a classification labeling result.

[0109] The accuracy of the proposed CNN classifier on the training set, test set and validation set reached 97.40%, 94.57% and 94.94% respectively, forming an intelligent recognition model for wellbore flow regime.

[0110] According to the principle of distributed fiber acoustic sensing (DAS) technology, the present invention uses a simulated physical experimental device to design an intelligent identification method for wellbore fluid flow state based on DAS monitoring. At the same time, DAS acquisition dynamic data is obtained, waterfall charts and FK charts are generated, DAS data preprocessing and feature extraction methods are proposed, and a reliable machine learning method is constructed to intelligently identify the wellbore fluid flow state model, thus forming an intelligent identification method for wellbore fluid flow state.

[0111] Therefore, the present invention adopts the above-mentioned intelligent identification method of wellbore fluid flow state based on DAS monitoring. The intelligent identification method in the present invention can realize real-time dynamic and continuous monitoring of fluid flow in the wellbore throughout the life cycle of oil and gas wells, provide technical support for downhole multiphase fluid flow monitoring and identification, provide guidance for on-site wellbore intelligent analysis, and bring economic benefits and safety guarantees to the oil and gas extraction industry.

[0112] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solution of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solution to deviate from the spirit and scope of the technical solution of the present invention.

Claims

1. A method for intelligently identifying the flow pattern of wellbore fluid based on DAS monitoring, characterized in that: The following steps are involved: S1. Design a simulated physical experimental device, lay optical fibers on the wellbore, and connect the DAS system module, wellbore, gas supply module, liquid supply module, and collection module; S2, adjust the wellbore angle, start the operation console, industrial control computer, field instrument and software system, set the pneumatic valve, and adjust the flow controller; S3, using DAS to collect acoustic data in the wellbore, and preprocessing the collected data. The specific operations of S3 are: S31, reading data, and generating a waterfall chart with the data collected by DAS; S32, using FK filtering to convert the DAS data from the time and spatial position of the recorded data in the optical cable, i.e., the corresponding time-space domain, into the frequency and wave number domain, wherein the FK filtering is implemented by a two-dimensional fast Fourier transform FFT algorithm, and the formula is: ； in t Indicates the time of input data, x Indicates the location of the input data, f Indicates frequency, k 1 represents the wave number; S33, using Min-Max normalization to normalize the data, the Min-Max normalization formula is expressed as: ； in, X is the original data, is the minimum value in the data set, is the maximum value in the data set, is the normalized data; S4. Compare the waterfall chart generated by DAS data with the different observed flow states to obtain the flow law of wellbore fluid under the influence of different factors. The specific operation of S4 is: According to the different flow patterns observed in the simulation experiment, the corresponding DAS waterfall diagram and FK diagram are analyzed to qualitatively identify the flow pattern of the wellbore fluid. At the same time, the flow pattern of the fluid in the wellbore at different angles and different flow rates is analyzed, and the flow law of the fluid in the wellbore is obtained using DAS. S5. Based on the preprocessed data, the flow state characteristic parameters are extracted using a feature extraction algorithm. The specific operation of S5 is as follows: S51, using wavelet transform to process the original signal to achieve low variance representation of the data; S52. Apply the short-time Fourier transform (STFT) to the acoustic data so that the acoustic signal is divided into small parts. The STFT formula is expressed as: ； in w ( m ) is the sliding signal x ( m + n ) to select a limited window of a segment, n Indicates the time indicator, k 2 is the frequency, m In the summation process, different time offsets are represented. N Indicates the number of samples; S6. Use machine learning algorithms to test and verify the feature set, build a wellbore flow pattern intelligent identification model, and form a wellbore fluid flow pattern intelligent identification method.

2. The method for intelligently identifying the flow state of wellbore fluid based on DAS monitoring according to claim 1, characterized in that: The specific operation of S1 is: According to the designed simulated physical experimental device plan, optical fiber is laid on the wellbore, pulled into straight lines on the upper and lower surfaces of the pipeline, and wound along the pipeline wall. Each pipeline is wound in the order of lower layer-upper layer-spiral, and the DAS system modules are connected to the optical fiber and computer respectively.

3. The method for intelligently identifying the flow state of wellbore fluid based on DAS monitoring according to claim 2 is characterized in that: The specific operation of S2 is: S21, start four power distribution cabinets to supply power to all electrical equipment of the test device, adjust the shaft to the required angle, and fix the shaft; S22, set the pneumatic valve, adjust the flow controller, set the liquid volume to a fixed value, and the liquid volume remains unchanged during the simulation experiment; S23. Use pneumatic valves to control the flow equipment to continuously adjust the flow ratio and set the air intake to keep the fluid stable in the pipeline for 3 minutes.

4. The method for intelligently identifying the flow state of wellbore fluid based on DAS monitoring according to claim 1, characterized in that: The specific operation of S6 is: The feature set obtained after preprocessing is input into the convolutional neural network algorithm for testing and verification, and a wellbore flow pattern intelligent identification model is constructed to form an intelligent identification method for wellbore fluid flow pattern.

Images