Gas well leakage point detection method

Abstract

The invention provides a gas well leakage point detection method, and relates to the technical field of gas well measurement, and the gas well leakage point detection method is realized by adopting a first ultrasonic measuring instrument, a second ultrasonic measuring instrument and a gyroscope. The coordinates of the gas well structure parameters are determined through angle parameters of the gyroscope and the diving depth of the second ultrasonic measuring instrument, supervised learning is used for connecting the gas well structure parameters according to the coordinates, and a three-dimensional model is constructed; and calculating a leakage path at the joint of the corrosion hole and / or the thread by using the three-dimensional model and a pixel value, and calculating the minimum cross section and the instantaneous leakage rate. The supervised learning adopts the structure parameters to construct a three-dimensional model, and the increment of the gas well structure parameters is used as the loss parameter of the dynamic simulation of the three-dimensional model. And the position and quantity of leakage at the joint of the corrosion hole and the thread are accurately determined.

Description

Technical Field

[0001] The invention relates to the technical field of gas well measurement, in particular to a gas well leakage point detection method. Background Art

[0002] In the prior art, ultrasonic measuring instruments are mainly used to monitor the annular liquid level and determine whether the pipeline is leaking by the change in liquid volume. However, this method has limitations. It cannot accurately locate the specific location and number of leaks, nor can it distinguish whether the leak is from a corrosion hole or a threaded connection. In addition, although the existing mathematical model can detect multiple leaks, the positioning accuracy of the corrosion hole is low, especially in the case of micro-leakage, it is difficult to effectively identify leaks from corrosion holes or threaded connections.

[0003] Existing technologies mainly rely on mathematical models in a two-dimensional coordinate system to establish the relationship between pressure and leakage rate through equivalent diameter and depth. This method requires processing a large amount of data and performing complex mathematical derivations, resulting in low detection efficiency and the inability to provide real-time feedback on the dynamic changes of the leakage point. Due to the complex gas well environment, the pressure, temperature and structural parameters in the pipeline will change over time. The existing static mathematical model cannot update the location and leakage amount of the leakage point in real time, causing the detection results to lag behind the actual situation.

[0004] In addition, existing technologies mainly focus on leak monitoring inside the pipeline, ignoring the leakage risk of external structures such as casing and cement layers. Gas wells have complex structures, and leakage of casing and cement layers will also have a serious impact on gas well safety and production. Existing methods fail to fully cover all potential leakage points of gas wells, limiting their effectiveness in practical applications.

[0005] Existing technologies have obvious deficiencies in the accuracy, real-time and comprehensiveness of leak detection, and urgently need to be improved to cope with the diverse leakage risks in complex gas well environments.

[0006] In view of this, this application is hereby filed. Summary of the invention

[0007] In view of the above problems existing in the prior art, a gas well leakage point detection method is provided, the purpose of which is to solve at least one of the above problems.

[0008] The technical solution to achieve the purpose of the present invention is: The present invention provides a gas well leakage point detection method, which is implemented by using a first ultrasonic measuring instrument, a second ultrasonic measuring instrument and a gyroscope, and the specific steps are as follows: In the first step, first, calibrate the ultrasonic emission points of the first ultrasonic measuring instrument and the second ultrasonic measuring instrument to the same horizontal plane. Subsequently, the first ultrasonic measuring instrument is used to measure the depth of the annular liquid surface. At the same time, the second ultrasonic measuring instrument has a built-in gyroscope and dives into the pipeline to measure the structural parameters of the gas well. The second ultrasonic measuring instrument obtains an image cross-section parallel to the annular liquid surface, determines the liquid surface position, and uses this to calibrate its diving depth. By comparing the liquid surface depth measured by the first ultrasonic measuring instrument with the diving depth of the second ultrasonic measuring instrument, the two are calibrated with each other. Next, the second ultrasonic measuring instrument measures the distance along the wellhead direction and in the opposite direction to further calibrate its diving depth. In this process, the measured structural parameters include the radius and curvature parameters of the pipeline, annulus, casing and cement layer. Through the above steps, the accuracy and consistency of the measurement data are ensured, providing a reliable basis for subsequent leak detection and positioning.

[0009] In the second step, the coordinates of the gas well structural parameters can be accurately determined by using the angle parameters of the gyroscope and the diving depth of the second ultrasonic measuring instrument; the data of the coordinates are connected through the supervised learning algorithm to construct a three-dimensional model of the gas well; the three-dimensional model can be dynamically adjusted according to the gas well structural parameters obtained in real time, and a real-time feedback signal is generated; the feedback signal is used to control the acquisition frequency and diving speed of the second ultrasonic measuring instrument, thereby optimizing the data acquisition process and ensuring measurement accuracy and efficiency; it should be noted that the angle parameters provided by the gyroscope combined with the diving depth of the second ultrasonic measuring instrument can accurately map the spatial coordinates of the internal structure of the gas well. The supervised learning algorithm constructs a high-precision three-dimensional model by analyzing and processing these coordinate data. The model can reflect the changes in the gas well structure in real time, and dynamically adjust the working parameters of the second ultrasonic measuring instrument according to these changes, such as the frequency and diving speed of the acquisition signal, so as to achieve efficient and accurate data acquisition and monitoring. This method not only improves the measurement efficiency, but also provides more reliable technical support for gas well leakage detection and structural analysis.

[0010] The third step is to use the three-dimensional model to calculate the curvature increment and predict the suspected leakage points that may exist in the gas well. Specifically, by analyzing the changing trend of the curvature in the three-dimensional model, the area with significantly increased curvature is identified, which may correspond to a potential leakage point. After the suspected leakage point is determined, the structural parameters of the gas well are continuously collected in the interval around the suspected leakage point. At the same time, ensure that the collected image cross-sections are continuously distributed in the axial direction of the gas well to fully cover the suspected area and capture detailed information of the leakage point. By combining the curvature analysis of the three-dimensional model with continuous data collection, the leakage point can be located more accurately, and reliable data support can be provided for subsequent leakage detection and repair.

[0011] The fourth step is to use the 3D model to accurately calculate the leakage path of the corrosion hole and / or threaded connection by analyzing the change of pixel values. Specifically, based on the high-resolution data of the 3D model, the pixel value change characteristics on the leakage path are identified to determine the specific direction and range of the leakage. At the same time, by calculating the minimum cross-sectional area in the leakage path and combining the principles of fluid dynamics, the instantaneous leakage volume is further estimated. By combining the pixel analysis of the 3D model with the geometric characteristics of the leakage path, the severity of the leakage can be quantified more accurately, and a scientific basis can be provided for the repair and risk assessment of the leakage point.

[0012] Compared with the prior art, the beneficial effects of the present invention include: (1) Select the image cross-sectional parameters obtained by the second ultrasonic measuring instrument as the parameters for three-dimensional modeling, improve the accuracy of three-dimensional modeling, reduce duplicate data, reduce the amount of data, and improve the efficiency of three-dimensional modeling; at the same time, correct the cross-sectional parameters of the image measured by the second ultrasonic measuring instrument by using the non-cross-sectional parameters of the image measured by the second ultrasonic measuring instrument, firstly, improve the accuracy of the image cross-sectional parameters obtained by the second ultrasonic measuring instrument, and secondly, improve the utilization rate of the non-cross-sectional parameters of the image measured by the second ultrasonic measuring instrument; (2) Supervised learning receives data from the second ultrasonic measuring instrument in real time to generate a three-dimensional model, and the three-dimensional model generates a real-time control signal to control the second ultrasonic measuring instrument to perform measurement; (3) The formation process of corrosion holes is simulated through loss parameters, the probability of corrosion holes appearing in the pipeline is predicted, and the geometric shape parameters of the corrosion holes measured in real time are verified; and based on the graphical distribution of pixel values, the area where corrosion holes will form or the threaded connection will leak can be predicted; (4) Accurately determine the location and quantity of leaks at corrosion holes and threaded connections. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0014] Figure 1 It is the overall schematic diagram of the gas well; Figure 2 It is a schematic diagram of a gas well provided with a first ultrasonic measuring instrument and a second ultrasonic measuring instrument; Figure 3 It is a schematic diagram of the area where the second ultrasonic measuring instrument ultrasonically measures the gas well mechanism parameters; Figure 4 It is a schematic diagram for determining the leakage direction in a three-dimensional model; Among them, 101-a first layer of casing, 102-a second layer of casing, 103-a third layer of casing, 104-a fourth layer of casing, 105-a fifth layer of casing, 201-a pipeline, 301-annulus, 401-a sealer, 501-a first ultrasonic measuring instrument, 502-a second ultrasonic measuring instrument, and 601-a leakage channel. DETAILED DESCRIPTION

[0015] To make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be described clearly and completely in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them.

[0016] Therefore, the following detailed description of the embodiments of the present invention is not intended to limit the scope of the invention claimed for protection, but merely represents some embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0017] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features and technical solutions in the embodiments may be combined with each other.

[0018] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0019] The present invention is further described in detail below with reference to the embodiments.

[0020] like Figures 1 to 4 As shown, the present invention provides a gas well leakage point detection method, which is implemented by using a first ultrasonic measuring instrument, a second ultrasonic measuring instrument and a gyroscope, and the specific steps are as follows: In the first step, the coordinates of the gas well structural parameters can be accurately determined by using the angle parameters of the gyroscope and the diving depth of the second ultrasonic measuring instrument. Through the supervised learning algorithm, these coordinate data are connected to construct a three-dimensional model of the gas well. The three-dimensional model can be dynamically adjusted according to the gas well structural parameters obtained in real time and generate a real-time feedback signal. The feedback signal is used to control the acquisition frequency and diving speed of the second ultrasonic measuring instrument, thereby optimizing the data acquisition process and ensuring measurement accuracy and efficiency; It should be noted that the diving depth measurement accuracy of the second ultrasonic measuring instrument directly determines the precision and accuracy of the three-dimensional modeling. If the diving depth measurement value is too small, it will cause the gas well structural parameters to overlap along the axial direction in the three-dimensional model, resulting in inaccurate modeling. Figure 4As shown in the figure, collecting a set of gas well structural parameters is equivalent to collecting a picture. If the cross-sectional area is not properly connected, the cross-sectional area of ​​the previous picture and the next picture will overlap, resulting in the misalignment of the gas well structural parameters and the actual position of the gas well. On the contrary, if the diving depth measurement value is too large, it will cause the gas well structural parameters to be discontinuous or separated along the axial direction in the three-dimensional model. In this case, the discontinuous area is usually filled with the gas well structural parameters of the previous or next picture, which will also cause the gas well structural parameters to be misaligned with the actual position. In order to solve this problem, the method of mutual calibration between the first ultrasonic measuring instrument and the second ultrasonic measuring instrument is adopted to improve the measurement accuracy of both. Specifically, the first ultrasonic measuring instrument is used to measure the volume and instantaneous leakage of the annular liquid, while the second ultrasonic measuring instrument is used to accurately measure the diving depth in the pipeline. Through bidirectional measurement calibration, that is, the second ultrasonic measuring instrument measures the distance along the wellhead direction and the opposite direction, the measurement accuracy of the diving depth is further improved. This method not only effectively avoids the problem of overlapping or discontinuity of gas well structural parameters, but also significantly improves the accuracy and reliability of three-dimensional modeling, providing more accurate data support for gas well leakage detection and structural analysis; The second step is to use the angle parameters of the gyroscope and the diving depth of the second ultrasonic measuring instrument to determine the coordinates of the gas well structural parameters, and use supervised learning to connect the gas well structural parameters according to the coordinates and build a three-dimensional model; the three-dimensional model generates a real-time feedback signal according to the real-time gas well structural parameters to control the frequency and diving speed of the second ultrasonic measuring instrument to collect signals; the angle parameters of the gyroscope and the diving depth of the second ultrasonic measuring instrument can be used to accurately determine the coordinates of the gas well structural parameters; through the supervised learning algorithm, the data of the coordinates are connected to build a three-dimensional model of the gas well; the three-dimensional model can be dynamically adjusted according to the gas well structural parameters obtained in real time, and a real-time feedback signal is generated; the feedback signal is used to control the acquisition frequency and diving speed of the second ultrasonic measuring instrument, thereby optimizing the data acquisition process and ensuring measurement accuracy and efficiency; It should be noted that when the second ultrasonic measuring instrument measures the connection between the cross sections of the image, it uses the angle parameters of the gyroscope for adjustment, and combines the curvature inside the gas well for correction, and uses the continuity of the curvature to achieve precise adjustment. It should be noted that since the second ultrasonic measuring instrument may rotate in the pipeline, its measurement of the gas well is carried out in sections, so the positional relationship between the data needs to be connected to ensure consistency. When the cross sections of the image collected by the second ultrasonic measuring instrument along the axis of the gas well are directly connected, angle misalignment may occur. In order to solve this problem, the rotation angle of the second ultrasonic measuring instrument is measured by a gyroscope, and the angle information is loaded into the measurement image, so that the gas well structural parameters have angle information. This method not only helps to accurately locate and connect the gas well structural parameters in the three-dimensional model, but also improves the ability to accurately locate the leakage point. In addition, the diving depth and rotation angle parameters of the second ultrasonic measuring instrument are loaded into the gas well structural parameters, and the gas well structural parameters are located and spliced ​​by combining the diving depth and rotation angle, so as to construct a high-precision three-dimensional model. The supervised learning algorithm uses real-time modeling technology to predict the changing trend of the gas well structural parameters that have been collected, and generates real-time feedback signals to dynamically control the acquisition frequency and diving speed of the second ultrasonic measuring instrument. For example, when a significant change in radius or curvature (such as increasing or decreasing) is detected, as well as areas where corrosion holes or threaded connections may appear, the diving speed of the second ultrasonic measuring instrument is reduced and the acquisition frequency is increased to obtain more data and improve measurement accuracy; in areas without abnormalities in the gas well, the diving speed is increased and the acquisition frequency is reduced to improve measurement efficiency and reduce invalid data. The above method not only optimizes the accuracy and efficiency of three-dimensional modeling, but also provides more reliable technical support for gas well leakage detection and structural analysis; The third step is to use the 3D model to calculate the curvature increment and predict the possible suspected leakage points in the gas well. After the suspected leakage point is determined, the structural parameters of the gas well are continuously collected in the interval around the suspected leakage point. At the same time, the collected image cross sections are ensured to be continuously distributed in the axial direction of the gas well to fully cover the suspected area and capture detailed information of the leakage point. It should be noted that leakage points may include oil pipe leakage points, air pipe leakage points and casing leakage points. Figure 4As shown, the ultrasonic wave of the second ultrasonic measuring instrument forms a cross-sectional area in the area perpendicular to the axial direction of the gas well, which is the image cross-sectional area obtained by measuring the gas well. It should be noted that the image cross-sectional area has a certain thickness or height (for example, the thickness is 60 pixels or 300 pixels, the height is 100 pixels or 200 pixels, and one pixel corresponds to a thickness or height of 1 cm or 5 cm), and the ultrasonic wave of the second ultrasonic measuring instrument has a certain divergence angle (for example, 30° or 60°). Therefore, the area non-perpendicular to the axial direction of the gas well includes an upper deflection angle area and a lower deflection angle area. The gas well structural parameters obtained in the upper deflection angle area and the lower deflection angle area need to be converted into the gas well structural parameters in the cross-sectional area through trigonometric conversion. When the gas well structural parameters in the cross-sectional area are selected as the basis for three-dimensional modeling, the gas well structural parameters obtained in the lower deflection angle area can be used as the data basis for generating real-time feedback signals, thereby having a prediction function. At the same time, these data can also be used for correction. The correction method is based on the side-angle relationship of the triangle, and the position of the point on the gas well cross-sectional area is determined by calculation, such as the position of the point on the pipeline, annulus and casing. In addition, the gas well structural parameters in the upper deflection angle area, lower deflection angle area and cross-sectional area can be mutually verified to ensure the accuracy and correctness of the parameters. Through the above method, not only the accuracy of 3D modeling can be improved, but also more reliable data support can be provided for the detection and positioning of leakage points, while achieving real-time feedback and prediction functions, and optimizing measurement efficiency and accuracy; The fourth step is to use the three-dimensional model to accurately calculate the leakage path of the corrosion hole and / or threaded connection by analyzing the change of pixel values; at the same time, by calculating the minimum cross-sectional area in the leakage path and combining the principle of fluid dynamics, the instantaneous leakage volume is further estimated; It should be noted that the pixel value is between 0 and 255, and different areas (such as pipes, annuli, casings, and liquids in annuli) have different pixel values. The characteristics of the corrosion hole can be calculated by the difference in pixel values. For example, the area with pixel values ​​between 245 and 248 is identified as the pixel value of the corrosion hole, and these pixel values ​​are analyzed using a supervised learning algorithm to determine the size, location, and path of the corrosion hole. In particular, by identifying the continuity of the pixel values ​​in the corrosion hole path, it is determined whether it may cause liquid leakage. Similarly, at the threaded connection of the pipeline, the size and location of the leakage path can also be determined by calculating the pixel values, and whether it may cause liquid leakage. The pixel values ​​of the leakage path at the threaded connection also need to meet the conditions of continuous distribution. In addition, the minimum cross-sectional area of ​​the corrosion hole and the leakage path at the threaded connection can be calculated by the number of pixels, and the instantaneous leakage can be calculated in combination with the pressure difference. At the same time, based on the graphical distribution of pixel values, the area where corrosion holes or threaded connection leakage may be formed can be predicted. For example, the normal wall thickness of a pipe may correspond to 600 or 2,000 pixels, while in an area where a corrosion hole is about to form, the number of pixels may be significantly reduced to 10 or 15. This method can not only accurately identify the characteristics of corrosion holes and leaks at threaded connections through the analysis and calculation of pixel values, but also predict potential leakage risks, providing a scientific basis for the safe monitoring and maintenance of gas wells.

[0021] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the first step, preferably, the first pressure measuring device and the first temperature measuring device are used to measure the pressure and temperature parameters in the annulus, while the second pressure measuring device and the second temperature measuring device follow the second ultrasonic measuring instrument to measure the pressure and temperature parameters in the pipeline. By measuring the pressure and temperature parameters in the annulus and the pipeline respectively, the instantaneous leakage can be calculated more accurately, providing reliable data support for leakage detection and analysis.

[0022] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the first step, further, as Figure 1As shown, taking the arrangement of five layers of casing in a gas well as an example, the casing includes a first layer of casing 101 , a second layer of casing 102 , a third layer of casing 103 , a fourth layer of casing 104 and a fifth layer of casing 105 . A cement layer is arranged between each layer of casing, specifically as follows: a cement layer is arranged between the first layer of casing 101 and the second layer of casing 102, and the area therebetween is a non-overlapping area, wherein the outer layer of the first layer of casing 101 is also arranged with a cement layer; a cement layer is arranged between the second layer of casing 102 and the third layer of casing 103, and the area therebetween is a non-overlapping area, wherein the outer layer of the second layer of casing 102 is also arranged with a cement layer; a cement layer is arranged between the third layer of casing 103 and the fourth layer of casing 104, and the area therebetween is a non-overlapping area, wherein the outer layer of the third layer of casing 103 is also arranged with a cement layer; a cement layer is arranged between the fourth layer of casing 104 and the fifth layer of casing 105, and the area therebetween is a non-overlapping area, wherein the outer layer of the fourth layer of casing 104 is also arranged with a cement layer; a cement layer is also arranged on the outer layer of the fifth layer of casing 105. The second ultrasonic measuring instrument can measure the radius and curvature of the first layer of casing 101, the second layer of casing 102, the third layer of casing 103, the fourth layer of casing 104 and the fifth layer of casing 105, and at the same time measure the radius and curvature of the cement layer between each layer of casing and the outer cement layer. Among them, the curvature includes the curvature of the inner and outer walls of the pipeline, the curvature of the annulus interface, the curvature of the inner and outer walls of the casing and the curvature of the cement layer interface. It should be noted that the second ultrasonic measuring instrument constructs a three-dimensional model by measuring the radius and curvature of the casing and cement layer, and establishes a timeline for the measurement data. The loss parameters are obtained on the timeline, and through the dynamic simulation of the three-dimensional model, the points where the casing and cement layer may leak can be predicted, and the leakage points that have occurred can be found in real time. This method not only improves the accuracy of leak detection, but also provides dynamic and real-time technical support for the safety monitoring and maintenance of gas wells.

[0023] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the first step, further, as Figure 1 and Figure 2 As shown, the second ultrasonic measuring instrument 502 dives into the pipeline 201 to measure the structural parameters of the gas well. Its structural parameters will be used for three-dimensional modeling and real-time monitoring. A sealer 401 is set between the first layer of casing 101 and the pipeline 201, and liquid is injected into the annulus 301. The first ultrasonic measuring instrument 501 is used to measure the liquid level in the annulus 301, thereby providing real-time data support for leakage detection and liquid level changes. This design combines the ultrasonic measurement technology and the role of the sealer, which can effectively improve the measurement accuracy of the gas well structural parameters and the real-time performance of leakage monitoring.

[0024] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the first step, further, as Figure 3As shown, the first ultrasonic measuring instrument is fixed to measure the liquid level depth of the annulus 301, and the initial position of the second ultrasonic measuring instrument is located at the same horizontal plane as the first ultrasonic measuring instrument. It should be noted that the measurement starting point of the first ultrasonic measuring instrument corresponds to the cross section measured by the second ultrasonic measuring instrument, which is used as the initial position of the measurement. The inclination angle of the pipeline 201 is a, and liquid is injected into the annulus 301. The liquid surface line is measured by the first ultrasonic measuring instrument to obtain the liquid surface height. At the same time, the second ultrasonic measuring instrument measures the cross-sectional image and liquid surface line of the gas well, and records the cable length of the second ultrasonic measuring instrument diving into the pipeline 201. Using the cosine function, combined with the two known quantities of liquid surface height and cable length, the inclination angle a of the pipeline 201 can be calculated. It should be emphasized that measuring the inclination angle a of the pipeline helps to accurately calculate the pressure parameters of the liquid, thereby more accurately calculating the leakage amount or instantaneous leakage amount of the liquid in the annulus 301. Through the trigonometric relationship between the inclination angle a and the cable length, the depth h of the pipeline 201 can be further calculated. This method not only improves the measurement accuracy, but also provides reliable data support for the calculation and real-time monitoring of the leakage amount.

[0025] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the first step, further, the diving position of the second ultrasonic measuring instrument is measured by its cable length, and the image of the pipe joint collected by the second ultrasonic measuring instrument is used for correction. It should be noted that since the unit length of the pipeline is determined, and the length of the unit pipeline connected by the thread is usually fixed (even if different lengths of pipelines are used, the radius is the same, and the length of the unit pipeline connected by the thread can still be determined by recording the connection sequence). When the connection length between the unit pipes exceeds its fixed length, it indicates that the threaded connection may be loose or fall off, and there is a possibility of thread leakage. The image measured by the second ultrasonic measuring instrument can further confirm the shape and state of the pipe joint, so as to accurately judge the integrity and leakage risk of the threaded connection. This method combines cable length measurement and image correction to significantly improve the accuracy and reliability of thread leakage detection.

[0026] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the first step, further, the second pressure measuring device is equipped with a plurality of pressure measuring probes, which are distributed on the same cross section of the image measured by the second ultrasonic measuring instrument, and are symmetrically distributed. For example, 4 or 6 pressure measuring probes can be set: if 4 probes are used, the angle interval between each probe is 90°; if 6 probes are used, the angle interval between each probe is 60°. In addition, an odd number of probes, such as 7 or 9, can also be used, and the angle interval between each probe is equal. This symmetrical distribution design is intended to improve the accuracy of pressure measurement, and in combination with the image recognition function of the three-dimensional model, the azimuth positioning of the leakage point is achieved through the parameter changes of the pressure measuring probes at various azimuth angles. Taking the symmetrical point of the pipeline cross section as the center, assuming that 4 pressure measuring probes are used as an example: if the pressure measuring probe at the 0° position detects a sudden change in the pressure parameter, and the pressure parameter increments of the probes at the 90° and 270° positions are second, and the pressure parameter increment of the probe at the 180° position is the smallest, it can be judged that the pipeline leakage point is most likely to be located at the 0° position. This method can accurately determine the angular position of the leakage point through comparative analysis of multi-angle pressure parameters, significantly improving the efficiency and accuracy of leak detection.

[0027] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the first step, further, the second temperature measuring device is equipped with a plurality of temperature measuring probes, and the probes are distributed on the same cross section of the image measured by the second ultrasonic measuring instrument, and are symmetrically distributed. For example, 4 or 6 temperature measuring probes can be set: if 4 probes are used, the angle interval between each probe is 90°; if 6 probes are used, the angle interval between each probe is 60°. In addition, an odd number of probes, such as 7 or 9, can also be used, and the angle interval between each probe is equal. This symmetrical distribution design is intended to improve the accuracy of temperature measurement, and in combination with the image recognition function of the three-dimensional model, the azimuth positioning of the leakage point is achieved through the parameter changes of the temperature measuring probes at various azimuth angles. Taking the symmetrical point of the pipeline cross section as the center, assuming that 4 temperature measuring probes are used as an example: if the temperature measuring probe at the 0° position detects a sudden change in the temperature parameter, and the probe temperature parameter increments at the 90° and 270° positions are second, and the probe temperature parameter increment at the 180° position is the smallest, it can be judged that the pipeline leakage point is most likely to be located at the 0° position. This method can accurately determine the angular position of the leakage point through comparative analysis of multi-angle temperature parameters, significantly improving the efficiency and accuracy of leak detection.

[0028] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the first step, further, a pressure measuring probe and a temperature measuring probe are paired and set at the same point. By mutual verification of the pressure parameter change and the temperature parameter change of the leakage point by the pressure measuring probe and the temperature measuring probe, a pressure model and a temperature model can be established in the three-dimensional model respectively, and the data of the two can be used for mutual correction. This method can not only improve the accuracy of measurement, but also enhance the reliability of leakage point detection, and provide more comprehensive data support for the safety monitoring and maintenance of gas wells.

[0029] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the first step, further, a telescopic mechanism is provided on the second ultrasonic measuring instrument to drive the movement of the pressure measuring probe and the temperature measuring probe. It should be noted that since the second ultrasonic measuring instrument can move freely in the pipeline and the pipeline may have an inclination angle, the distance from each pressure measuring probe and temperature measuring probe to the inner wall of the pipeline is inconsistent, thereby affecting the accuracy of the measurement. By locating the spatial position of each pressure measuring probe and temperature measuring probe through the data of the three-dimensional model, and using the data of the three-dimensional model to accurately control the telescopic mechanism, each probe can be brought close to the inner wall of the pipeline, thereby improving the measurement accuracy of the pressure parameters and temperature parameters. This method not only optimizes the measurement process, but also significantly improves the accuracy and reliability of the data.

[0030] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the first step, further, the image data measured by the second ultrasonic measuring instrument 502 is adjusted to the center point of the cross section of the pipeline 201 by vector addition and subtraction. It should be noted that due to the inclination angle of the pipeline 201, the center point measured by the second ultrasonic measuring instrument 502 is often difficult to directly locate on the center point of the pipeline cross section. In order to improve the accuracy of three-dimensional modeling, the center point measured by the second ultrasonic measuring instrument 502 is corrected to the center point of the cross section of the pipeline 201 by vector addition and subtraction. Take the number axis as an example for explanation: Assuming that the diameter of the pipeline 201 is 4 (using dimensionless expression), the pipeline is located at -2 to 2 on the number axis. If the center measurement point of the second ultrasonic measuring instrument 502 is located at the -1 position of the number axis, the center measurement point is adjusted to the 0 point position of the number axis, that is, the center point of the pipeline cross section, by adding 1 on the negative side and subtracting 1 on the positive side. This method significantly improves the accuracy of three-dimensional modeling and the accuracy of measurement data through vector correction.

[0031] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the first step, further, the cross-sectional parameters of the image measured by the second ultrasonic measuring instrument 502 are used as the main parameters for three-dimensional modeling, and the cross-sectional parameters are corrected using its non-cross-sectional parameters. It should be noted that the second ultrasonic measuring instrument 502 will generate a large amount of measurement data. By using the cross-sectional parameters as the core parameters for three-dimensional modeling, the amount of data can be effectively reduced and the modeling process can be simplified. At the same time, by correcting the cross-sectional parameters using non-cross-sectional parameters, on the one hand, the accuracy of the cross-sectional parameters can be improved, and on the other hand, the utilization rate of the non-cross-sectional parameters can be improved. This method not only optimizes the data processing efficiency, but also significantly improves the accuracy and reliability of three-dimensional modeling.

[0032] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the second step, preferably, under the supervised learning framework, the historical data and real-time data of the gas well structure parameters are used to calculate the increment, and the increment is used as the loss parameter of the dynamic simulation of the three-dimensional model. Through this loss parameter, the formation process of the corrosion hole is simulated to predict the probability of the corrosion hole appearing in the pipeline. At the same time, the geometric shape parameters of the corrosion hole measured in real time are verified to optimize the prediction accuracy of the model. This method can effectively improve the accuracy of corrosion hole prediction and provide a reliable basis for pipeline maintenance.

[0033] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the second step, further, supervised learning uses structural parameters to build a three-dimensional model, and the increment of the gas well structural parameters is used as the loss parameter of the three-dimensional model dynamic simulation. The specific steps are as follows: Step A1: Collect structural parameters, temperature parameters and pressure parameters, and pre-process their parameters Collect data from the first ultrasonic measuring instrument, the second ultrasonic measuring instrument, the first pressure measuring device, the second pressure measuring device, the first temperature measuring device, and the second temperature measuring device. The data includes but is not limited to the radius, curvature, annular liquid depth, inclination angle of the pipeline, pressure and temperature parameters in the pipeline, etc. of the pipeline and annulus. Set the dynamic acquisition frequency according to the working conditions and environmental changes of the gas well to ensure the real-time and comprehensiveness of the data. Remove outliers (such as noise data caused by sensor failure) and redundant data to ensure data quality. Normalize the collected data to unify data of different dimensions into the same numerical range to avoid deviations caused by data dimension differences during model training. Introduce data enhancement techniques (such as rotation, scaling, noise addition, etc.) to increase the diversity of training data and improve the generalization ability of the model. For example, randomly rotate and scale the ultrasonic image to simulate the gas well structure from different perspectives.

[0034] Step A2: Structural parameter extraction Obtain high-resolution images from ultrasonic measuring instruments to capture the geometric features of the pipe and annulus. Use image processing techniques (such as edge detection and contour extraction) to extract structural parameters such as the radius and curvature of the pipe and annulus. Calculate the inclination angle of the pipe and the depth of the annulus liquid level through image analysis algorithms (such as Hough transform). Record the depth of the annulus liquid level and the inclination angle of the pipe, which will serve as the basic data for 3D modeling. Verify the accuracy of the extracted structural parameters through multi-sensor data fusion technology to ensure the reliability of subsequent modeling.

[0035] Step A3: Build and train the 3D model Based on the extracted structural parameters, supervised learning is used to build a 3D model. The model should reflect the overall shape, size and structural characteristics of the pipeline as detailed as possible. Use regression models (such as linear regression, support vector regression) or deep learning models (such as convolutional neural network CNN, long short-term memory network LSTM) to build the model. Combine the advantages of multiple models (such as CNN for image feature extraction and LSTM for time series prediction) to perform multi-model fusion to improve prediction accuracy and robustness. Use the extracted structural parameters, pressure and temperature parameters as input features of the model. Use the increment of the structural parameters as part of the loss function, that is, the difference between the model prediction value and the actual measurement value. Use the historical data set to train the model and adjust the model parameters through optimization algorithms (such as gradient descent) until the preset convergence condition is reached or the loss function is minimized. Use independent data sets to verify the generalization ability and performance of the model to ensure that the model performs well on unseen data. Introduce automated parameter adjustment tools (such as Grid Search, Random Search or Bayesian Optimization) to optimize model hyperparameters, reduce the workload of manual parameter adjustment and improve model performance.

[0036] Step A4: Dynamic Simulation Use the trained three-dimensional model for dynamic simulation. When new structural parameters, pressure and temperature parameters are actually measured, they are input into the model for prediction and update. The model dynamically adjusts according to real-time data, generates feedback signals, controls the acquisition frequency and diving speed of the second ultrasonic measuring instrument, and optimizes the data acquisition process. The model adjusts the structural model of the pipeline and annulus according to the changes in the input to ensure that it always matches the actual working conditions. Introduce a real-time feedback mechanism to adjust the data acquisition strategy in real time according to the model's prediction results to further improve the real-time performance and accuracy of the model. For example, when the model detects a significant change in curvature, it automatically reduces the diving speed of the second ultrasonic measuring instrument and increases the acquisition frequency to obtain more detailed data.

[0037] Step A5: Model Evaluation and Optimization Continuously monitor the prediction performance of the model and adjust the model parameters or model structure as needed to improve the model accuracy and stability. By monitoring the prediction error of the model in real time, timely discover and correct the deviation or error in the model. Use cross-validation to tune parameters to ensure that the model performs consistently and reliably on different data sets. Combine automated parameter adjustment tools and cross-validation to further optimize the model performance and ensure its applicability in complex gas well environments. Dynamically adjust the weight of the loss function according to the performance of the model to optimize the prediction accuracy of the model. For example, increase the weight of the corrosion hole area to improve the detection sensitivity of the leakage point. Randomly rotate, scale and add noise to the ultrasonic image to simulate the gas well structure under different working conditions and enhance the generalization ability of the model. Introduce synthetic data generation technology to generate additional training data by simulating gas well leakage scenarios to further improve the robustness of the model. Fuse the data of ultrasonic measuring instruments, pressure sensors and temperature sensors to improve the accuracy of the model through multi-source data complementarity. Use the Kalman filter algorithm to fuse multi-sensor data to reduce measurement errors and improve data reliability. In the dynamic simulation process, introduce a real-time feedback mechanism to dynamically adjust the data acquisition strategy according to the prediction results of the model. For example, when the model detects a suspected leakage point, it automatically increases the acquisition frequency of the area to obtain more detailed data. The diving speed and acquisition frequency of the second ultrasonic measuring instrument are controlled by real-time feedback signals to optimize the data acquisition process and ensure the real-time and accuracy of the model. Bayesian Optimization is used to perform hyperparameter tuning to automatically search for the optimal model parameter combination and reduce the workload of manual parameter adjustment. Combined with cross-validation and automated parameter adjustment tools, the generalization ability and prediction accuracy of the model are further improved. Through the above steps and technical detail optimization, a high-precision and high-reliability three-dimensional model can be constructed for the detection and prediction of gas well leakage points. The optimized scheme not only improves the generalization ability and real-time performance of the model, but also further enhances the prediction accuracy and robustness of the model through multi-model fusion, automated parameter adjustment and real-time feedback mechanism, providing strong technical support for the safe monitoring and maintenance of gas wells.

[0038] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in step A3 of the second step, further, the specific steps of optimizing the three-dimensional model by supervised learning are as follows: Step B1: Multi-source data integration, labeling, data cleaning and data enhancement Collect images (obtained by ultrasonic measuring instruments), point clouds (laser scanning data) and voxel data (3D grid representation) of the 3D model, and simultaneously record the real-time parameters of the pressure and temperature sensors. Add structured labels to the training data, including pipeline bounding boxes, corrosion hole key points, leakage path area annotations, etc., to ensure the accuracy of supervised learning. Eliminate sensor outliers (such as temperature / pressure data beyond the physical range) and image noise (processed by median filtering). Randomly rotate (±10°), scale (0.8~1.2 times), and translate (±5 pixels) the ultrasonic image to simulate different measurement perspectives. Add Gaussian noise (σ=0.01) to improve the model's anti-interference ability. Simulate leakage scenarios through fluid dynamics simulation to generate annotated synthetic data sets that cover extreme working conditions (such as high-pressure leakage and coexistence of multiple leakage points). Use the Kalman filter algorithm to fuse ultrasonic, pressure, and temperature data to reduce the error of a single sensor and generate a high-confidence training set.

[0039] Step B2: Image feature extraction, point cloud feature extraction and multimodal feature fusion The pre-trained ResNet-50 is used as the basic architecture, and the model is fine-tuned through transfer learning to adapt to the geometric characteristics of the gas well structure. The channel attention module (SE Block) is introduced at the end of the convolutional neural network (CNN) to weight the feature importance of different channels and enhance the sensitivity to the corrosion hole area. On the basis of the original PointNet, a local feature aggregation layer is added (the neighborhood point features are extracted through the KNN algorithm) to improve the recognition ability of tiny corrosion holes. Adaptive Max Pooling is used to dynamically adjust the size of the pooling area to adapt to leakage paths of different scales. The image features extracted by CNN, the point cloud features extracted by PointNet, and the pressure and temperature parameters are spliced ​​to form a multi-dimensional feature vector. For time series data (such as continuously collected pressure changes), the LSTM network (Long Short-Term Memory) is used to extract time-dependent features and fuse them with spatial features.

[0040] Step B3: Joint spatial-temporal model and loss function calculation to build the model A CNN-LSTM hybrid architecture is designed. The CNN branch processes image / point cloud data, and the LSTM branch processes time series sensor data. Finally, the output is fused through a fully connected layer. A learnable weight coefficient is introduced to automatically balance the contribution of image features and sensor features. The mean square error (MSE, used for structural parameter regression) and cross entropy loss (used for leakage point classification) are combined, and the weights of the two are dynamically adjusted through the hyperparameter λ. For the increment of gas well structural parameters, an incremental sensitive loss term is designed to enhance the model's ability to capture small changes.

[0041] Step B4: Model training and optimization The model was trained on a synthetic dataset to preliminarily learn the general characteristics of gas well structures. The model was fine-tuned using real gas well data to adapt to actual working conditions. A feedback loop was embedded in the training process to dynamically adjust the data sampling strategy (such as oversampling the leak point area) based on the performance of the validation set. Bayesian Optimization was used to search for the optimal hyperparameter combination, and the parameter space was defined, including learning rate (1e-5~1e-3), batch size (16~128), and λ (0.3~0.7). The validation set loss was monitored, and training was terminated early if it did not decrease for 5 consecutive epochs to prevent overfitting. The leak point detection accuracy (Accuracy), recall rate (Recall), and F1 score were used to comprehensively evaluate the model performance. A 5-fold cross validation was performed to ensure the stability of the model under different data distributions. The leak point location accuracy was significantly improved by combining image, point cloud, and sensor data. The CNN-LSTM hybrid model can respond to changes in gas well structure in real time and support online updates (delay <50ms). Bayesian optimization improves the efficiency of hyperparameter search by 40% and accelerates model convergence. Data enhancement and Kalman filtering enable the model to maintain a detection accuracy of more than 90% in a noisy environment. Through the above optimization, the three-dimensional model constructed by supervised learning can not only predict the location and amount of leakage points with high accuracy, but also provide dynamic decision support for gas well maintenance, significantly improving the real-time and reliability of safety monitoring.

[0042] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the third step, further, the geometric shape parameters of the corrosion hole are calculated, and the instantaneous leakage of the corrosion hole is calculated through a three-dimensional model; the specific steps are as follows: Step C1, real-time measurement of the geometric parameters of the corrosion hole, and calculation of the smallest cross section in the corrosion hole channel; it should be noted that the smallest cross section in the corrosion hole channel determines the instantaneous leakage. Specifically, the diameter of the corrosion hole is measured, the depth of the corrosion hole is measured, the shape characteristics of the corrosion hole (such as irregular shape, circular, etc.) are recorded, and the specific position and depth of the corrosion hole in the pipeline are determined; In step C2, the three-dimensional model corrects the geometric shape parameters of the corrosion hole through the loss parameters; it should be noted that the geometric shape parameters of the corrosion hole obtained by the second ultrasonic measuring instrument are used as loss parameters; the structural parameters in the three-dimensional model are adjusted through optimization algorithms (such as gradient descent, genetic algorithm, etc.), and supervised learning dynamically simulates the change process of the structural parameters in the three-dimensional model to correct the geometric shape parameters of the corrosion hole measured in real time; at the same time, the geometric shape parameters of the corrosion hole reversely correct the supervised learning to dynamically simulate the change process of the structural parameters in the three-dimensional model; improve the accuracy of the geometric shape parameters of the corrosion hole. Correct specific structural parameters: correct the radius of the pipeline, correct the radius of the annulus, correct the local curvature of the pipeline, correct the radius of the casing, and correct the local curvature of the casing; In step C3, the instantaneous leakage of liquid from the annulus to the pipeline is estimated based on the pressure and temperature parameters in the pipeline and the annulus; the calculation formula for the instantaneous leakage of the corrosion hole is as follows: Q 瞬时 =f(h 差 ,S,P1,T1,P2,T2) Q 瞬时 represents the instantaneous leakage of the corrosion hole, f represents the functional relationship, h 差 represents the height difference from the corrosion hole of pipeline 201 to the liquid level in annulus 301, S represents the smallest cross section in the corrosion hole channel, P1 represents the pressure value in annulus 301, T1 represents the temperature value in annulus 301, P2 represents the pressure value near the corrosion hole in the pipeline, and T2 represents the temperature value near the corrosion hole in the pipeline; it should be noted that the instantaneous leakage of the corrosion hole Q 瞬时 Based on the function relationship f, the calculation method of calculus is used to improve the calculation accuracy of the instantaneous leakage of the corrosion hole. In order to simplify the calculation and measurement equipment, P1, T1, P2, and T2 are replaced by proportional coefficients in the three-dimensional model, and their proportional coefficients are calculated by supervised learning.

[0043] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the third step, further, when the pipeline 201 has no leakage, its temperature parameters are collected and annotated into the three-dimensional model, and simulated by the three-dimensional model; the formula for the temperature parameter changing with the depth of the pipeline 201 is: T h =T0+K×h Among them, T h is the temperature at the depth h of the pipeline, T0 is the surface temperature, K is the geothermal warming rate (generally in °C / m), h is the depth (in meters), + represents addition, and × represents multiplication; It should be noted that, when there is no leakage in the pipeline 201 (which can be confirmed by measuring the liquid level in the annulus 301 with the first ultrasonic measuring instrument 501), the parameters of the pipeline depth and temperature are collected to establish a functional relationship between the pipeline depth and temperature, especially to verify the geothermal warming rate K; the above data are measured in the shut-in state to improve the stability and repeatability of the data measurement.

[0044] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the third step, further, the temperature at the corrosion hole in the pipeline 201 is measured in real time, and the temperature increment is calculated according to its depth parameter, and simulated by a three-dimensional model; the formula for calculating the temperature increment is: ΔT=T h -T h实时 T h实时 is the real-time measured temperature at the pipeline depth h, and ΔT represents the temperature increment. It should be noted that when a leak occurs in the pipeline 201, the liquid in the annulus 301 leaks into the pipeline 201. Since there is a temperature difference between the liquid in the annulus 301 and the natural gas in the pipeline 201, the heat absorption and release between the liquid and the natural gas are equal, and the instantaneous amount of leakage is reflected in the change of the temperature increment.

[0045] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the third step, further, a functional relationship between the instantaneous leakage and the temperature increment is established at the corrosion hole, and then the instantaneous leakage of the thread is derived from the temperature increment at the thread leakage through the functional relationship; the functional relationship between the instantaneous leakage and the temperature increment is established at the corrosion hole as follows: Q 瞬时 =f1(ΔT) f1 represents the functional relationship between instantaneous leakage and temperature increment; it should be noted that since thread leakage is difficult to detect by ultrasound, it is indirectly measured by conservation of thermal energy and establishment of functional relationship. At the thread leakage point, the instantaneous leakage of the thread leakage is calculated by the functional relationship f1 between the instantaneous leakage and temperature increment and the temperature increment ΔT.

[0046] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the third step, further, when the pipeline 201 has no leakage, its pressure parameters are collected and annotated into the three-dimensional model, and simulated by the three-dimensional model; the formula for the pressure parameter changing with the depth of the pipeline 201 is: P h =P0+ρ×g×h Among them, P his the pressure at the pipeline depth h, P0 is the surface pressure, ρ is the density of the formation fluid (usually the density of natural gas), g is the acceleration due to gravity (about 9.81 m / s²), h is the depth (in meters), + represents addition, and × represents multiplication; It should be noted that, when there is no leakage in the pipeline 201 (which can be confirmed by measuring the liquid level in the annulus 301 with the first ultrasonic measuring instrument 501), the parameters of the pipeline depth and pressure are collected to establish a functional relationship between the pipeline depth and pressure; the above data are measured in the shut-in state to improve the stability and repeatability of the data measurement.

[0047] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the third step, further, the pressure at the corrosion hole in the pipeline 201 is measured in real time, and the pressure increment is calculated according to its depth parameter, and simulated by a three-dimensional model; the formula for calculating the pressure increment is: ΔP=P h -P h实时 P h实时 is the real-time measured pressure at the pipeline depth h, and ΔP represents the pressure increment; it should be noted that when a leak occurs in the pipeline 201, the liquid in the annulus 301 leaks into the pipeline 201, and the instantaneous amount of the leak is reflected in the change of the pressure increment.

[0048] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the third step, further, a functional relationship between the instantaneous leakage and the pressure increment is established at the corrosion hole, and then the instantaneous leakage of the thread is derived from the pressure increment at the thread leakage through the functional relationship; the functional relationship between the instantaneous leakage and the pressure increment is established at the corrosion hole as follows: Q 瞬时 =f2(ΔP) f2 represents the functional relationship between instantaneous leakage and pressure increment; it should be noted that since thread leakage is difficult to detect by ultrasound, indirect measurement is performed by pressure change and establishing a functional relationship. At the thread leakage point, the instantaneous leakage of the thread leakage is calculated by the functional relationship f2 between the instantaneous leakage and pressure increment and the pressure increment ΔP.

[0049] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the third step, further, at the corrosion hole position, the instantaneous leakage, temperature increment and pressure increment are accurately calculated by mutual correction. It should be noted that these three variables are all caused by the leakage of the pipeline 201, and their changes are correlated. By mutual correction of three different measurement methods (instantaneous leakage, temperature increment and pressure increment), the accuracy and reliability of leakage calculation can be effectively improved. This multi-variable collaborative correction method not only enhances the credibility of the data, but also provides a more accurate basis for leak detection and positioning.

[0050] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the fourth step, further, the second ultrasonic measuring instrument measures the geometric shape parameters of the corrosion hole in real time, and the three-dimensional model corrects the geometric shape of the corrosion hole through the loss parameter, and estimates the instantaneous leakage of the liquid from the annulus to the pipeline in combination with the pressure and temperature parameters in the pipeline and the annulus. By calculating the increment of the temperature and pressure parameters in the pipeline, a one-to-one mapping relationship is established between the instantaneous leakage and the increment. The total instantaneous leakage is calculated according to the annular liquid level, and the sum of the leakage components of several leakage points in the pipeline is summarized. By comparing the difference between the total leakage and the sum of the leakage components, it is determined whether there is leakage outside the annulus. The three-dimensional model uses the casing and its inner and outer structural parameters (including the radius and curvature of the casing and cement layer) to verify the location and instantaneous leakage of the leakage outside the annulus to ensure the accuracy of the model. The first ultrasonic measuring instrument is used to measure the liquid level depth of the liquid in the annulus, and the second ultrasonic measuring instrument dives into the pipeline to perform stereoscopic measurement of the gas well by emitting ultrasonic waves to the surroundings, generating a large amount of data. In order to simplify the amount of data and improve the measurement accuracy, the initial measurement positions of the images obtained by the first and second ultrasonic measuring instruments are calibrated at the same time, and the cross section of the image measured by the second ultrasonic measuring instrument is selected as the calibration plane. The cross section is parallel to the annular liquid surface, ensuring that the initial measurement positions of the two measuring instruments are in the same horizontal plane, thereby improving the calculation accuracy of the pipeline inclination angle, further improving the calculation accuracy of the annular liquid surface depth and liquid volume, and optimizing the calculation of the hydraulic pressure difference between the liquid surface and the pipeline leakage point. When the inclination angle of the gas well and its pipeline reaches a critical value, the first ultrasonic measuring instrument may not be able to measure the annular liquid surface depth. At this time, by calculating the pipeline inclination angle, the second ultrasonic measuring instrument can replace the first ultrasonic measuring instrument to complete the measurement of the annular liquid surface depth. In addition, by recording the time difference between the cross section of the image measured by the second ultrasonic measuring instrument and the annular liquid surface, the total amount of liquid leakage in the annular space during this period can be calculated. This method significantly improves the accuracy and efficiency of leak detection and provides a reliable guarantee for the safe operation of the pipeline.

[0051] It should be noted that, through the precise measurement of the first ultrasonic measuring instrument and the second ultrasonic measuring instrument, combined with the supervised learning technology, a highly accurate three-dimensional model of the pipeline and its annulus can be constructed. The model not only covers the geometric dimensions, curvature and inclination angle of the pipeline and the annulus, but also is closer to the actual physical environment, significantly improving the accuracy and reliability of the model. By real-time monitoring of the pressure, temperature and geometric shape parameters of the corrosion holes in the pipeline and the annulus, and dynamically adjusting the three-dimensional model and simulation process using supervised learning technology, deviations or errors in the model can be discovered and corrected in a timely manner. In particular, when liquid leaks in the pipeline, the model can be adjusted quickly to more accurately estimate the leakage amount and locate the leakage point. By calculating the mapping relationship between the instantaneous leakage amount and the temperature and pressure parameter increments in the pipeline, and calculating the instantaneous total leakage amount in combination with the annular liquid level height, the leakage situation can be comprehensively evaluated. At the same time, according to the difference between the total leakage amount and the sum of the leakage components of each leakage point, it is possible to judge and locate whether there is a leak outside the annulus, thereby improving the accuracy and efficiency of leak detection and reducing potential safety risks. In addition, the structural parameters of the 3D model (such as the radius and curvature of the casing) can be used to further verify whether the leakage occurs in the annulus area between the casing and the pipeline, ensuring the reliability and effectiveness of the model. Through real-time monitoring and dynamic adjustment of the pipeline interior and annulus, potential leakage risks can be predicted and preventive maintenance measures can be taken to reduce production losses and environmental hazards caused by leakage. This method not only improves the accuracy of leak detection, but also provides a strong guarantee for the safe operation of the pipeline.

[0052] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the fourth step, further, in the leakage channel 601, a pixel overflow method is used to determine the flow direction of the liquid. Figure 4 As shown, the liquid Figure 4 When the pixel value of the liquid is consistent with the pixel value in the leakage channel 601, a pixel overflow area convex outward is formed. The shape of the convex area can be regular (such as Figure 4 By analyzing the pixel overflow area, the direction of the casing and / or cement layer leakage can be identified, so as to determine whether the liquid in the annulus is leaking to the formation outside the casing and / or cement layer, or whether the water in the formation is flowing back into the annulus. This method provides an intuitive and effective basis for determining the leakage direction.

[0053] It should be noted that for technical features that are not fully explained, conventional technical means are used.

[0054] The above embodiments are only used to illustrate the present invention and are not intended to limit the technical solutions described in the present invention. Although the present invention has been described in detail with reference to the above embodiments, the present invention is not limited to the above specific implementation methods. Therefore, any modification or equivalent replacement of the present invention; and all technical solutions and improvements thereof that do not depart from the spirit and scope of the invention are included in the scope of the claims of the present invention.

Claims

1. A gas well leakage point detection method, which is implemented by using a first ultrasonic measuring instrument (501), a second ultrasonic measuring instrument (502) and a gyroscope, characterized in that: The specific steps are as follows: In a first step, a first ultrasonic measuring instrument (501) is used to measure the depth of the liquid surface of the annulus (301), and at the same time, a second ultrasonic measuring instrument (502) with a built-in gyroscope dives into the interior of the pipeline (201) to measure the structural parameters of the gas well; the second ultrasonic measuring instrument (502) obtains an image cross section parallel to the liquid surface of the annulus (301); The second step is to determine the coordinates of the gas well structural parameters using the angle parameters of the gyroscope and the diving depth of the second ultrasonic measuring instrument (502); connect the data of the coordinates through a supervised learning algorithm to construct a three-dimensional model of the gas well; the three-dimensional model can be dynamically adjusted according to the gas well structural parameters obtained in real time, and generate a real-time feedback signal; the feedback signal is used to control the acquisition frequency and diving speed of the second ultrasonic measuring instrument (502); The third step is to calculate the curvature increment using the three-dimensional model to predict the suspected leakage point in the gas well. After the suspected leakage point is determined, the structural parameters of the gas well are continuously collected in the interval around the suspected leakage point. The fourth step is to use the three-dimensional model to accurately calculate the leakage path of the corrosion hole and / or threaded connection by analyzing the change of pixel values. At the same time, the instantaneous leakage amount is further estimated by calculating the minimum cross-sectional area in the leakage path.

2. A gas well leakage point detection method according to claim 1, characterized in that: In the first step, the image data measured by the second ultrasonic measuring instrument (502) is adjusted to the center point of the cross section of the pipeline (201) by means of vector addition and subtraction.

3. A gas well leakage point detection method according to claim 1 or 2, characterized in that: In the first step, the cross-sectional parameters of the image measured by the second ultrasonic measuring instrument (502) are used as the main parameters for three-dimensional modeling, and the cross-sectional parameters are corrected using the non-cross-sectional parameters.

4. A gas well leakage point detection method according to claim 1, characterized in that: In the second step, the increment is calculated using the historical data and real-time data of the gas well structural parameters, and the increment is used as the loss parameter for the dynamic simulation of the three-dimensional model.

5. A gas well leakage point detection method according to claim 1, characterized in that: In the third step, the geometric shape parameters of the corrosion hole are calculated, and the instantaneous leakage of the corrosion hole is calculated through the three-dimensional model.

6. A gas well leakage point detection method according to claim 1, characterized in that: In the fourth step, in the leakage channel (601), the pixel overflow method is used to determine the flow direction of the liquid.

Images