Downhole annular flow channel fluid detection device and method based on electrical capacitance tomography

Abstract

The present invention provides a downhole annular flow channel fluid detection device and method based on electrical capacitance tomography. The device comprises an annular capacitance detection sensor array, a system circuit shield, a system power supply, a test circuit, a downhole microprocessor, and an intelligent surface analysis module. The test circuit comprises a multi-channel capacitance switching switch, a C / V conversion unit, a downhole power amplifier module, and an annular capacitance information acquisition unit. The system power supply respectively supplies power to the downhole microprocessor, the annular capacitance detection sensor array, and the test circuit. The annular capacitance detection sensor array is connected to the test circuit. The present invention enables non-contact rapid measurement of downhole annular flow channels during oil drilling. Based on the imaging results, gas, oil, and water intrusion in the downhole annular flow channels can be promptly determined. Accidents such as overflows and well leakage can be monitored and predicted in advance. The device is of great significance for downhole annular flow channel drilling fluid detection and downhole overflow and well leakage monitoring during oil drilling.

Description

Technical Field

[0001] The present invention relates to the field of multiphase fluid measurement in annular flow channels of oil wells, and in particular to a device and method for detecting fluid in annular flow channels of oil wells based on electrical capacitance tomography. Background Art

[0002] Nowadays, whether drilling on land or offshore, early gas intrusion and overflow detection is a crucial aspect of annular well control safety. When large amounts of methane gas invade the annular pipe, if it is not detected promptly and appropriate annular well control measures are not implemented, overflows or even serious accidents such as lost circulation will occur. A major factor in the 2010 blowout accident at the Deepwater Horizon drilling site in the Gulf of Mexico was the failure to promptly detect and confirm the intrusion of methane gas. Current early overflow detection technologies used in drilling are primarily based on parameters such as changes in mud inlet and outlet flow rates and pipeline pressure. Most of these parameters are measured at the wellhead, resulting in a significant delay in overflow detection results.

[0003] Accidents such as overflows and lost circulation are a major source of safety concerns for oil well drilling platforms, and a comprehensive solution currently lacks both domestically and internationally. Research on effective detection and rapid early warning of these incidents is crucial for ensuring safe drilling platform operations and protecting the marine ecosystem. To address this issue, numerous overflow detection methods based on downhole data have been developed. These methods can more accurately and promptly measure the extent of gas intrusion, allowing sufficient time for incident response. However, these methods require upgrading and modifying existing drilling processes and equipment, which is very costly. Therefore, most methods remain in the research phase. Providing more timely and reliable gas intrusion detection technology at a low cost, without changing existing drilling processes, and thus enabling faster implementation in commercial drilling applications, is a critical issue of both research and commercial value in the current development of oil drilling technology.

[0004] In terms of device structure, capacitance tomography technology is currently widely used in single-pipeline central flow basin fluid detection, but has never been applied to multi-pipeline annular flow basin fluid detection. In multi-pipeline annular flow basins, if traditional excitation methods are used, there is a detection polar angle range, which makes the flow basin outside the detection range of 180° enter the detection blind spot, which will lead to an increase in the edge effect of the capacitor plate, and more electric field lines will cross the inner boundary of the field, causing the sensitivity of the inner boundary of the field to show plate-like enhancement, while the sensitivity of the outer field is weakened, which is not conducive to improving imaging quality. For multi-pipeline annular flow basins, it is necessary to change the detection angle direction, angle and size of the capacitor plate, and at the same time innovate the excitation scheme to solve the problem of blind spots in annular flow basin detection.

[0005] Another challenge is that compared to other process tomography techniques, such as optical tomography and radiation tomography, ECT has lower spatial resolution and poorer image quality. Two main factors contribute to ECT's low spatial resolution: first, the number of independent capacitance measurements is far less than the number of pixels in the reconstructed image, a "pathological" characteristic; and second, the linear simplification of the nonlinear relationship between the capacitance vector and the dielectric distribution by traditional reconstruction algorithms results in a loss of projection information.

[0006] Currently, ECT systems commonly use sensors with 8-, 12-, or 16-electrode structures. Improving spatial resolution simply by increasing the number of capacitance measurements is difficult. Furthermore, algorithms used for ECT image reconstruction can be primarily categorized into linear and nonlinear models. Linear model algorithms linearly simplify the nonlinear mapping between capacitance and dielectric distribution, describing the physical relationship between the two using a sensitivity field matrix. Linear model algorithms can be further categorized into direct algorithms, iterative algorithms, and intelligent optimization algorithms. Direct algorithms, such as the LBP algorithm and the Tikhonov algorithm, directly use the pseudo-inverse of the sensitivity field matrix for backprojection. These algorithms offer low computational complexity and fast reconstruction speed, but low reconstruction accuracy. Iterative algorithms, such as the Landweber algorithm and the conjugate gradient method, use linear iterations to approximate the sensitivity field matrix. These algorithms offer improved accuracy but slow reconstruction speed. Intelligent optimization algorithms, such as simulated annealing and genetic algorithms, transform the inverse problem into an optimization problem and use a global optimization algorithm to search for the optimal approximation of the sensitivity field inverse matrix. These algorithms offer high accuracy but are difficult to converge, resulting in significant limitations in practical applications. Summary of the Invention

[0007] The purpose of the present invention is to solve the defects of the above-mentioned prior art and provide a downhole annular flow channel fluid detection device and method based on electrical capacitance tomography. The downhole annular flow channel fluid detection method based on electrical capacitance tomography has the advantages of non-contact, non-invasive, simple and flexible structure, low cost and high sensitivity.

[0008] This device eliminates the need for extensive upgrades and replacements to drilling equipment and can more accurately and promptly measure the extent of gas intrusion. It is a direct means of detecting multiphase flow in the annular passages of oil wells. Using this device to detect annular passages can promptly identify leaks and overflows, providing early warning to drilling operators and minimizing the occurrence of well kicks and leaks, thereby avoiding further losses and buying more time for subsequent well control, thereby reducing the incidence of accidents.

[0009] At the same time, in response to the problems existing in fluid detection in multi-pipeline annulus flow areas, a new electro-capacitive plate design and innovative excitation scheme are proposed to solve the problem of enhanced sensitivity at the inner boundary of the field and weakened sensitivity at the outer field, thereby improving the validity of detection data and improving imaging quality.

[0010] The image reconstruction algorithm used in this device is based on the extreme learning machine (ELM) machine learning method. Combined with the Landweber iteration principle, a new machine learning algorithm is proposed: the L-ELM-Landweber algorithm. The L-ELM algorithm improves the ELM algorithm because the number of training samples is much smaller than the number of pixels of the reconstructed image, and the output weight β is an ill-conditioned, sparse, and non-positive definite matrix. The solution to the ECT positive problem is a nonlinear problem. The solution to the linear positive problem of reconstructing the image based on the traditional Landweber iterative algorithm will inevitably introduce reconstruction errors. The ELM network is used to solve the ECT nonlinear positive problem, predict the ECT capacitance measurement value, and use the ELM-Landweber algorithm to reduce the image reconstruction error. The fusion of the L-ELM algorithm and the ELM-Landweber algorithm into the L-ELM-Landweber algorithm can effectively solve the problems of slow reconstruction speed, low reconstruction accuracy, and difficult convergence of results.

[0011] The present invention adopts the following technical solutions:

[0012] A downhole annular flow channel fluid detection device based on electrical capacitance tomography includes an annular capacitance detection sensor array, a test circuit, a system power supply, a downhole microprocessor, and an intelligent surface analysis module. The annular capacitance detection sensor array consists of 16 annular capacitance detection electrode plates, a system circuit shield, filler material, and insulating pipes. Its function is to detect the distribution of the medium within the sensitive field. The annular capacitance detection sensor array detects the dielectric constant distribution of two-phase or multi-phase fluids and converts it into the sensor's output capacitance. The downhole microprocessor controls the multi-channel capacitance switching switch to sequentially select one capacitor plate as the excitation capacitor plate and the remaining capacitor plates as the receiving capacitor plates. An excitation signal is applied to the excitation capacitor plates. The C / V conversion module converts the tiny capacitance between the plates into a corresponding voltage value, and the downhole power amplifier module amplifies the converted voltage value. The annular capacitance information acquisition unit transmits the amplified voltage value to the downhole microprocessor, which reconstructs the downhole annular flow channel cross-section information based on the L-ELM-landweber algorithm and transmits it to the intelligent ground analysis module. Ground staff can then use real-time image information to monitor annular flow channel gas intrusion, overflow, well leakage and other accidents in advance.

[0013] The system power supply supplies power to the downhole microprocessor, the annular space capacitance detection sensor array and the test circuit respectively, wherein the annular space capacitance detection sensor array is connected to the test circuit, the test circuit is connected to the downhole microprocessor, the downhole microprocessor is connected to the intelligent ground analysis module, the system power supply test circuit, and the downhole microprocessor are located between the system circuit shielding cover and the inner wall of the instrument.

[0014] The test circuit includes a multi-channel capacitance switching switch, a C / V conversion unit, a power amplification module, and an annular space capacitance information acquisition unit, wherein the multi-channel capacitance switching switch is connected to the C / V conversion unit, the C / V conversion unit is connected to the downhole power amplification module, and the downhole power amplification module is connected to the annular space capacitance information acquisition unit.

[0015] Due to the harsh downhole environment of oil drilling, where temperatures can exceed 200 degrees Celsius, the insulating pipe is made of high-temperature and high-pressure resistant organic glass (polymethyl methacrylate) and serves as the detection annulus flow channel. The system circuit shield is a metal (copper or steel) cylinder installed between the insulating pipe and the inner flow channel. It must be grounded during operation of the electrical capacitance tomography system to effectively prevent interference from external electromagnetic fields and changes in dielectric distribution caused by environmental changes from affecting the detection capacitor plates. The insulating pipe is a transparent, thin-walled organic glass tube. While observing the flow pattern, it also prevents the wall thickness from causing insensitivity in the detection area and severe nonlinear distortion, thereby affecting the quality of image reconstruction. The capacitor plate leads are generally made of shielded wire and are not too long. This prevents crosstalk between leads and also prevents severe attenuation of the tiny capacitance signal caused by long-distance transmission, which can cause inaccurate acquisition signals.

[0016] The annular capacitance detection sensor array consists of 16 capacitance detection plates, a system circuit shield, filler material, and insulated pipe. The capacitor plate array is fixed to the measurement subsections of the outer and inner flow channels. The plates are arranged at a 22.5° central angle with the inner channel's center as the center. The filler material thickness should be increased or decreased based on a 68:1 ratio of the wellbore diameter to the filler material thickness. For a well with a 6.75-inch radius, a 5mm filler insert is recommended. Because the detection flow area is the annular channel, not the inner flow channel, the plates should be bent outward from the pipe and set at a 25° plate angle. The plate width should be increased or decreased based on a 71:1 ratio of the wellbore diameter to the plate width. For a well with a 6.75-inch radius, a 15mm plate width is recommended for optimal measurement results.

[0017] A new two-dimensional ECT sensor excitation measurement mode is used, employing a traditional single-capacitor plate excitation measurement scheme. Excitation voltage is applied to capacitor plate I, and the capacitance values ​​between capacitor plates I-II, I-III, …, and I-VIII are measured, and so on, for a total of 28 data sets. For annular flow channel fluid detection, the capacitance values ​​between 16 capacitor plates are measured, and so on, for a total of 28 data sets. For annular pipeline fluid detection, a 16-electrode annular capacitance detection sensor array employs an eight-group capacitance annular wheel excitation measurement scheme. The angle between the excitation and detection capacitor plates is kept within a range of -90° to 90°. Exceeding this range results in significant errors in the data collected by the detection plates, resulting in distorted imaging.

[0018] In the first round of excitation, the 16 plates are divided into 8 groups, and every two capacitor plates are regarded as a new capacitor. Plates I-II are new capacitors. * , III-VI plates are new capacitors 2 * ,……, XV-XVI plates are new capacitors 8 * , for the new capacitor 1 * Apply the excitation voltage and measure the capacitance of the new capacitor in the range of -90° to -90°, that is, measure C 1*2* 、C 1*3* 、C 1*7* 、C 1*8* Capacitor value. Then stimulate the new capacitor 2 * When the angle is between -90° and 90°, the capacitance of the new capacitor is measured, and so on for a total of 16 sets of data.

[0019] In the second round of excitation, plates II-III are the new capacitor 1", plates IV-V are the new capacitor 2", ..., XVI-I is the new capacitor 8", and an excitation voltage is applied to the new capacitor 1", and its capacitance value is measured within the range of -90°-90°; and so on. In each round of excitation, the 16-electrode plate annular capacitance detection sensor array is switched to one capacitor plate, with each switching angle separated by 22.5°. Then, the sensors are grouped in pairs and re-excited for detection. A total of 16 rounds of switching are performed, and finally, this method detects a total of 32 groups of data.

[0020] The downhole microprocessor controls the multi-channel capacitor switching switch to control the data acquisition sequence. The multi-channel capacitor switching switch includes a single-chip CMOS chip. The single-chip CMOS chip structure ensures low power consumption of the switching circuit and distributes the excitation signal to the required excitation capacitor plate. The single-chip CMOS chip contains 16 independent channels, which are connected to the 16 capacitor plates in the annular space capacitance detection sensor array module in turn, and uses the decoding of four address lines to select one of the 16 input channels to be connected to the common output.

[0021] The distance between the capacitor plate used as the detection capacitor and the excitation capacitor in the annulus capacitance detection sensor array should be different. Overflow, well leakage and other accident information includes one or more of the overflow\well leakage time, overflow\well leakage rate, overflow\well leakage location and overflow\well leakage type. The downhole microprocessor reconstructs the fluid map in the annulus pipeline at the current moment and displays it in the intelligent ground analysis module for analysis. The staff can determine the current moment as the time when the overflow, well leakage and other accidents occurred based on the current image and determine the type of accident based on the dielectric constant. The microprocessor uses the collected data and the L-ELM-landweber algorithm to perform image reconstruction and cross-sectional phase content calculation to obtain the conductivity distribution information and cross-sectional phase content information of the annulus flow channel cross section, and transmits it to the intelligent ground analysis module. The following are the steps for image reconstruction:

[0022] The oil-gas-water three-phase flow is used as the medium. The relative dielectric constant of the oil phase is set to 8, the relative dielectric constant of the gas phase is set to 1, and the dielectric constant of water is 81.

[0023] Step 1. ECT finite element simulations of the annular flow channel were performed using Comsol software, targeting oil / gas and water / gas two-phase distributions. Empty and full field samples were generated. Five oil / gas two-phase distribution models were selected for simulation: single-bubble, two-bubble, three-bubble, stratified flow, and annular flow. 850 samples were collected for each of the five distribution models, for a total of 4250 samples. Simulated voltage measurements corresponding to each of the 4250 distribution models were calculated.

[0024] Step 2. Use the first 800 samples of each distribution model as training samples and the last 50 samples as test samples. Use the dielectric constant distribution of the 4000 training samples as the input of the L-ELM network and the measured capacitance value as the output of the L-ELM network to complete the training of the L-ELM network.

[0025] Step 3. Use the LBP algorithm to calculate the reconstructed image grayscale matrix corresponding to the sample and normalize it to obtain the initial values ​​of the 5 oil / gas two-phase distribution model images.

[0026] Step 4. Use the L-ELM network to predict the voltage value, substitute it into equations (1) and (2) to reconstruct the image.

[0027] G K+1 =G K +α K S T (ν m -ν ELM ) (1)

[0028] G0=G LBP (2)G Kis the image grayscale matrix of the k-th iteration; G0 is the initial value of the iteration and is obtained by the LBP algorithm; α k is the iteration factor of the kth step. ELM is the voltage value predicted by the L-ELM network; v m is the measured voltage value. G LBP It is the grayscale matrix of the reconstructed image after LBP algorithm processing.

[0029] Step 5. Use the image relative error formula (3) as the evaluation index. After revising the error, the downhole microprocessor reconstructs the downhole annulus flow channel cross-section information based on the L-ELM-landweber algorithm and the voltage value collected by the annulus information detection unit, and transmits it to the intelligent ground analysis module.

[0030]

[0031] g * is the calculated conductivity distribution; g is the set conductivity distribution.

[0032] IE is the image error evaluation index.

[0033] Step 6. Based on the real-time image information, ground staff can know the real-time distribution of oil and gas in the downhole fluid and detect abnormal fluid distribution.

[0034] Step 7. Similarly, finite element simulation experiments and L-ELM network training are performed for water / gas two-phase distribution and oil-gas-water three-phase distribution. The voltage values ​​collected by the annular information detection unit are used to reconstruct the downhole annular flow channel cross-section information and transmit it to the intelligent ground analysis module.

[0035] Based on the detection data, the downhole microprocessor transmits the phase content information of the annular flow channel section to the intelligent surface analysis module, which calculates the real-time dynamic phase content δ of the annular flow channel section. The formula is as follows:

[0036]

[0037] Among them: δ is the phase content, g i is the grayscale value of the i-th pixel point on the cross section, and N is the number of interface pixels.

[0038] Ground staff monitor the real-time image information in the annular flow area based on the annular intelligent ground analysis module, and use the real-time dynamic phase content δ observation judgment standard of the annular flow channel section to monitor the annular flow channel fluid anomalies, which can achieve early detection of accidents such as gas intrusion, overflow, and well leakage.

[0039] Beneficial effects of the present invention:

[0040] The present invention can obtain a cross-sectional reconstruction image of the annular flow channel in an oil well, and then use the cross-sectional phase content formula to calculate the phase content of the annular flow channel cross section, thereby understanding the distribution of solids, liquids, gases, and flows. In the field of oil well pipeline fluid measurement, there is no effective measurement method for the cross-sectional phase content of the annular flow channel. Due to the complex downhole working conditions, it is impossible to measure it directly. The present invention can obtain the dielectric constant image and cross-sectional phase content of the annular flow channel in a non-contact manner, which is of great significance for downhole annular flow channel measurement and downhole gas intrusion and overflow monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 This is a schematic diagram of the detection device of the present invention;

[0042] Figure 2 This is a structural diagram of the detection device of the present invention;

[0043] Figure 3 This is a top view of the detection device structure of the present invention;

[0044] Figure 4 This is a flowchart of the image reconstruction algorithm of the detection device of the present invention;

[0045] Figure 5 This is a voltage information table detected by the detection device of the present invention.

[0046] In the figure: 1-well wall, 2-instrument inner wall, 3-instrument outer wall, 4-system circuit shielding cover, 5-annulus capacitance detection sensor array, 6-system power supply, 7-test circuit, 8-downhole microprocessor, 9-intelligent surface analysis module, 10-inner flow channel, 11-filling material, 12-annulus flow channel, 13-insulation pipe;

[0047] 70-Multi-channel capacitance switching switch, 71-C / V conversion unit, 72-Downhole power amplification module, 73-Annular space capacitance information acquisition unit. DETAILED DESCRIPTION

[0048] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention are described clearly and completely below. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of the present invention.

[0049] like Figure 1-2 As shown, the downhole annulus flow channel fluid detection device based on electrical capacitance tomography of the present invention includes an instrument inner wall 2, an instrument outer wall 3, a system circuit shielding cover 4, an annulus capacitance detection sensor array 5, a system power supply 6, a test circuit 7, a downhole microprocessor 8, and an intelligent ground analysis module 9.

[0050] Among them, the test circuit 7 includes a multi-channel capacitor switching switch 70, a C / V conversion unit 71, a downhole power amplifier module 72, and an annular space capacitance information acquisition unit 73. The downhole microprocessor 8 controls the multi-channel capacitor switching switch 70 in the test circuit 7 to select one capacitor plate in the annular space capacitance detection sensor array 5 as an excitation capacitor, and the remaining capacitor plates as detection capacitor plates. The detection signal is converted into a corresponding voltage value by the C / V conversion unit 71 in the test circuit 7, and is collected by the annular space capacitance information acquisition unit 73 in the test circuit 7, and then transmitted to the downhole microprocessor 8; the downhole microprocessor 8 uses the voltage value collected by the annular space information detection unit according to the L-ELM-Iandweber algorithm, adopts the image relative error formula (3) as an evaluation index, and reconstructs the downhole annular space flow channel cross-section information and transmits it to the intelligent ground analysis module 9; the ground staff monitors the real-time image information in the annular space flow field according to the intelligent ground analysis module 9, and observes the real-time dynamic phase content δ of the annular space flow channel cross-section, so as to achieve early monitoring of accidents such as gas intrusion, overflow, and well leakage.

[0051] The system power supply 6 generates an excitation signal and simultaneously powers the device. The wellbore 1, the instrument inner wall 2, and the instrument outer wall 3 isolate the annular capacitance detection sensor array 5, the system power supply 6, the test circuit 7, and the downhole microprocessor 8 from the internal flow channel fluid. This is pressure-resistant and radiation-proof, and the material is non-magnetic. The annular capacitance detection sensor array 5 is isolated from the system power supply 6, the test circuit 7, and the downhole microprocessor 8 by its system circuit shield 4. The system power supply 6, the test circuit 7, and the downhole microprocessor 8 are all installed between the system circuit shield 4 and the instrument inner wall 2 to prevent circuit operation from interfering with the annular capacitance detection sensor array 5 and affecting the monitoring of incidents such as overflows and well leaks.

[0052] like Figure 3 As shown, in the annular space capacitance detection sensor array 5, 16 annular space capacitance detection electrode plates are fixed in the measuring short section between the inner flow channel and the annular space flow channel, with the center of the inner flow channel as the center point, and the central angle between each capacitor plate is 22.5°. Among them, the 16 annular space capacitance detection electrode plates are located between the system circuit shielding cover 4 and the insulating pipe 13, and the system circuit shielding cover 4 and the insulating pipe 13 are also filled with filling material 11. The annular space capacitance detection electrode plate is made of copper foil or stainless steel, and can be placed inside the pipe according to different needs. 16 annular space capacitance detection electrode plates, a system circuit shielding cover 4, filling material 11 and insulating pipe 13 constitute an annular space capacitance detection sensor.

[0053] The annular capacitance detection sensor array 5 adopts an 8-group capacitor wheel excitation measurement scheme. In the first round of excitation, the 16 plates are divided into 8 groups, and every 2 capacitor plates are regarded as a new capacitor. Plates I-II are new capacitors. * , III-VI plates are new capacitors 2 * ,……, XV-XVI plates are new capacitors 8 * , for the new capacitor 1 * Apply the excitation voltage and measure the capacitance of the new capacitor in the range of -90° to -90°, i.e. measure Capacitor value. Then stimulate the new capacitor 2 * When the angle is between -90° and 90°, the capacitance of the new capacitor is measured, and so on for a total of 16 sets of data.

[0054] During the second round of excitation, plates II-III became capacitor 1", plates IV-V became capacitor 2", and so on, and plates XVI-I became capacitor 8". Excitation voltage was applied to capacitor 1, and its capacitance was measured within the -90° to 90° angle range. This process continued in this manner, switching the 16-plate annular capacitance detection sensor array by one capacitor plate each time the excitation cycle began. Each switching angle was 22.5° apart, and the array was then divided into two groups and re-excited for testing. This cycle was repeated 16 times, resulting in a total of 32 sets of detection data.

[0055] The downhole microprocessor 8 controls the multi-channel capacitor switching switch 70 to control the data acquisition sequence. The multi-channel capacitor switching switch 70 includes a single-chip CMOS chip. The single-chip CMOS chip structure ensures low power consumption of the switching circuit and distributes the excitation signal to the required excitation capacitor plate. The single-chip CMOS chip contains 16 independent channels, which are connected to the 16 capacitor plates in the annular space capacitance detection sensor array 5 in turn, and uses the decoding of four address lines to select one of the 16 input channels to be connected to the common output.

[0056] The distance between the capacitor plate serving as the detection capacitor and the excitation capacitor in the annular capacitance detection sensor array 5 should be different. Accident information such as overflow and well leakage includes one or more of the overflow / well leakage time, overflow / well leakage rate, overflow / well leakage location and overflow / well leakage type. The downhole microprocessor 8 reconstructs the fluid map in the annular pipeline at the current moment and displays it in the intelligent ground analysis module 9 for analysis. The staff can determine the current moment as the time when the overflow, well leakage and other accidents occurred based on the current image and determine the type of accident based on the dielectric constant. The data collected by the downhole microprocessor 8 uses the L-ELM-landweber algorithm to perform image reconstruction and cross-sectional phase content calculation to obtain the cross-sectional conductivity distribution information and cross-sectional phase content information of the annular flow channel, and transmit it to the intelligent ground analysis module 9. The following are the image reconstruction steps:

[0057] The oil-gas-water three-phase flow is used as the medium. The relative dielectric constant of the oil phase is set to 8, the relative dielectric constant of the gas phase is set to 1, and the dielectric constant of water is 81.

[0058] Step 1. Use Comsol software to perform ECT finite element simulation of the annular flow channel, with the simulation objects being oil / gas two-phase distribution and water / gas two-phase distribution. Generate empty field and full field samples. Select five oil / gas two-phase distribution models for simulation: single bubble, two bubbles, three bubbles, stratified flow, and annular flow. Each of the five distribution models has 850 samples, for a total of 4250 samples. Calculate the simulated voltage measurement values ​​corresponding to the 4250 distribution models.

[0059] Step 2. Use the first 800 samples of each distribution model as training samples and the last 50 samples as test samples. Use the dielectric constant distribution of the 4000 training samples as the input of the L-ELM network and the measured capacitance value as the output of the L-ELM network to complete the training of the L-ELM network.

[0060] Step 3. Use the LBP algorithm to calculate the reconstructed image grayscale matrix corresponding to the sample and normalize it to obtain the initial values ​​of the five oil / gas two-phase distribution model images.

[0061] Step 4. Use the L-ELM network to predict the voltage value, substitute it into equations (1) and (2) to reconstruct the image.

[0062] G K+1 =G K +α k s T (ν m -ν ELM ) (1)

[0063] G0=G LBP (2)G K is the image grayscale matrix of the k-th iteration; G0 is the initial value of the iteration and is obtained by the LBP algorithm; α k is the iteration factor of the kth step; S is the sensitivity matrix; ν ELM is the voltage value predicted by the L-ELM network; ν m is the measured voltage value, s T is the transpose of the sensitivity matrix S.

[0064] Step 5. Use the image relative error IE formula (3) as the evaluation index. After revising the error, the downhole microprocessor uses the voltage value collected by the annular information detection unit according to the L-ELM-landweber algorithm to reconstruct the downhole annular flow channel cross-section information and transmit it to the intelligent ground analysis module.

[0065]

[0066] g * is the calculated conductivity distribution; g is the set conductivity distribution.

[0067] Step 6. Based on the real-time image information, ground staff can know the real-time distribution of oil and gas in the downhole fluid and detect abnormal fluid distribution.

[0068] Step 7. Similarly, finite element simulation experiments and L-ELM network training are performed for water / gas two-phase distribution and oil-gas-water three-phase distribution. The voltage values ​​collected by the annular information detection unit are used to reconstruct the downhole annular flow channel cross-section information and transmit it to the intelligent ground analysis module.

[0069] Ground staff monitor the real-time image information in the annular flow area based on the intelligent ground analysis module 9, and observe the real-time dynamic phase fraction δ of the annular flow channel section to achieve early detection of accidents such as gas intrusion, overflow, and well leakage.

[0070] This embodiment provides a method for detecting fluid in an annular flow channel in a downhole system based on electrical capacitance tomography, including: a system power supply 6 generates an AC excitation signal, which controls a multi-channel capacitance switching switch 70 in a test circuit 7 through a downhole microprocessor 8 to select one of the capacitor plates in the annular capacitance detection sensor array 5 as an excitation capacitor, and the remaining capacitor plates as detection capacitor plates; the detection signal is converted into a corresponding voltage value by a C / V conversion unit 71 in the test circuit 7, which is a small capacitance value that changes between the plates. The voltage value is then collected by an annular capacitance information collection unit 73 in the test circuit 7 and then transmitted to the downhole microprocessor 8.

[0071] The present invention uses a downhole annular flow channel fluid detection method based on the L-ELM-Landweber algorithm and electrical capacitance tomography, and realizes downhole overflow monitoring by transmitting the dielectric constant change and phase content information of the annular flow channel to the ground staff for post-processing based on the information.

[0072] Example

[0073] Based on the simulation software, the gas / liquid fluid simulation modeling of the water injection well of polymer flooding oil recovery is carried out to obtain the detection electrical signal, such as Figure 5 As shown, the detection data is sent to the microprocessor; image reconstruction is performed according to the L-ELM-landweber algorithm. The entire annular cross-section domain under test contains 1,014,000 pixels with a grayscale value of 55,060. The phase content δ in the downhole annular flow channel fluid detection method using electrical capacitance tomography is calculated using formula (4), and the phase content of the downhole annular flow channel is 5.43%.

[0074] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A downhole annulus flow channel fluid detection device based on electrical capacitance tomography, comprising an annulus capacitance detection sensor array, a system circuit shield, a system power supply, a test circuit, a downhole microprocessor, and an intelligent surface analysis module, characterized in that: The test circuit includes a multi-channel capacitance switching switch, a C / V conversion unit, a downhole power amplification module, and an annular capacitance information acquisition unit. The system power supply supplies power to the downhole microprocessor, the annular capacitance detection sensor array, and the test circuit respectively. The annular capacitance detection sensor array is connected to the test circuit, which is connected to the downhole microprocessor. The downhole microprocessor is connected to the intelligent surface analysis module. The system power supply, test circuit, and downhole microprocessor are located between the system circuit shield and the inner wall of the instrument. The downhole microprocessor uses the L-ELM-landweber algorithm to reconstruct the detection information of the downhole annular flow channel fluid during oil drilling, and completes the timely judgment of gas invasion, oil invasion, and water invasion in the annular flow channel of oil drilling. The annular capacitance detection sensor array consists of 16 capacitor plates, a system circuit shield, filling material, and an insulating pipe. With the center of the inner flow channel as the center point, the central angle between each capacitor plate is 22.5°. The 16 annular capacitance detection plates are located between the system circuit shield and the insulating pipe, and the space between the system circuit shield and the insulating pipe is also filled with filling material. The annular capacitance detection sensor array requires that the angle between the excitation capacitor plate and the detection capacitor plate be within the range of -90° to 90°. If this range is exceeded, the data collected by the detection capacitor plate will have large errors and the imaging effect will be distorted. Therefore, a new two-dimensional ECT sensor excitation measurement mode is adopted for the annular capacitance detection sensor array with 16 capacitor plates, using an 8-group capacitor wheel excitation measurement scheme for data acquisition; 8-group capacitor wheel-cut excitation measurement scheme: select one capacitor plate group as the excitation capacitor plate, and the remaining four capacitor plate groups in the range of -90°-90° as the detection capacitor plates to measure four capacitance measurement values. Use the adjacent capacitor plate group as the excitation capacitor plate and the remaining three capacitor plate groups as the detection capacitor plates to obtain three capacitance measurement values. Repeat this cycle until the last two capacitor plate groups measure one capacitance value, thus obtaining 16 capacitance measurement values. Then wheel-cut 22.5° for a new group and similar measurement, ultimately obtaining 32 groups of data.

2. The downhole annulus flow channel fluid detection device based on electrical capacitance tomography according to claim 1 is characterized in that: In the annular space capacitance detection sensor array, the thickness of the filling material is arranged according to the ratio of 68:1 between the length of the oil well diameter and the thickness of the filling material; the oil well with a radius of 6.75 inches is embedded in the filling material by 5 mm; the capacitor plate should be bent toward the outside of the pipeline and the plate angle should be set to 25°, among which the width of the capacitor plate is arranged according to the ratio of 71:1 between the circumference of the oil well diameter and the width of the capacitor plate, and the width of the plate for the oil well with a radius of 6.75 inches is 15 mm.

3. The downhole annulus flow channel fluid detection device based on electrical capacitance tomography according to claim 1, characterized in that: The insulating pipe is made of high temperature and high pressure resistant organic glass polymethyl methacrylate, and the system circuit shielding cover is a cylinder made of copper or steel installed between the insulating pipe and the inner flow channel.

4. The downhole annulus flow channel fluid detection device based on electrical capacitance tomography according to claim 1, characterized in that: The image reconstruction speed must be no less than 24 frames per second.

5. A method for detecting a downhole annular flow channel fluid using the electrical capacitance tomography-based downhole annular flow channel fluid detection device according to claim 1, characterized in that: The oil-gas-water three-phase flow is set as the medium. The relative dielectric constant of the oil phase is set to 8, the relative dielectric constant of the gas phase is set to 1, and the dielectric constant of water is 81. The specific processing method for reconstructing the fluid information image in the annular flow channel is as follows: Step 1. Use COMSOL software to perform ECT finite element simulation of the annular flow channel. The simulation objects are oil / gas two-phase distribution and water / gas two-phase distribution. Empty field and full field samples are generated. Five oil / gas two-phase distribution models, including single bubble, two-bubble, three-bubble, stratified flow, and annular flow, are selected for simulation. Each of the five distribution models has 850 samples, for a total of 4250 samples. The simulation voltage measurement values ​​corresponding to the 4250 distribution models are calculated. Step 2. Use the first 800 samples of each distribution model as training samples, and the last 50 samples as test samples. Use the dielectric constant distribution of the 4000 training samples as the input of the L-ELM network, and use the measured capacitance value as the output of the L-ELM network to complete the training of the L-ELM network. Step 3. Use the LBP algorithm to calculate the grayscale matrix of the reconstructed image corresponding to the sample and normalize it to obtain the initial values ​​of the five oil and gas two-phase distribution model images; Step 4. Use the L-ELM network to predict the voltage value and substitute it into equations (1) and (2) to reconstruct the image: G K+1 =G K +a K S T (n m -n BLM ) (1) G0=G LBP (2) G K is the image grayscale matrix of the k-th iteration; G0 is the initial value of the iteration and is obtained by the LBP algorithm; α k is the iteration factor of the kth step, S is the sensitivity matrix, s T is the transpose of the sensitivity matrix S, ν ELM is the voltage value predicted by the L-ELM network; ν m To measure the voltage value; Step 5. Use the image relative error IE formula (3) as the evaluation index. After correcting the error, the downhole microprocessor uses the voltage value collected by the annular information detection unit according to the L-ELM-Landweber algorithm to reconstruct the downhole annular flow channel cross-section information and transmit it to the intelligent ground analysis module: g * is the calculated conductivity distribution, g is the set conductivity distribution; Step 6. Ground staff detect abnormal fluid distribution based on real-time image information and the real-time distribution of oil and gas in the oil well fluid; Step 7. Conduct finite element simulation experiments on water / gas two-phase distribution and oil-gas-water three-phase distribution, and L-ELM network training. Use the voltage values ​​collected by the annular information detection unit to reconstruct the downhole annular flow channel cross-section information and transmit it to the intelligent ground analysis module.

6. The method for detecting fluid in an annular channel of a well based on electrical capacitance tomography according to claim 5, characterized in that: In the intelligent ground analysis module, the phase fraction δ of the annular flow channel section is as follows: δ is the phase content, g i is the grayscale value of the i-th pixel point on the cross section, and N is the number of interface pixels; Ground staff monitor the real-time image information in the annular flow area based on the intelligent ground analysis module, and use the real-time dynamic phase fraction δ observation judgment standard of the annular flow channel section to monitor the annular flow channel fluid anomalies, thereby achieving early detection of gas intrusion, overflow, and well leakage accidents.

Images