Wellhead device and station process detection evaluation management system, method and product

Abstract

The invention relates to the technical field of integrity detection of oil and gas wellhead devices and station process pipelines, and particularly discloses a wellhead device and station process detection evaluation management system, method and product. Wherein online ultrasonic scanning and data acquisition can be carried out on a wellhead device and a pipeline through the matched wellhead detection terminal and station pipeline detection terminal; three-dimensional ultrasonic field self-correction is adopted, a phased array sound field is accurately calibrated, high-precision three-dimensional acoustic imaging under a complex structure is achieved, good coupling between a probe and a detected body is ensured, and defect echoes are accurately obtained; in combination with a multi-order echo recognition algorithm and a deep learning model, intelligent analysis is performed on the acquired ultrasonic data, so that the defect recognition rate and analysis efficiency are greatly improved, and automatic extraction and classification of defect features are effectively realized; the defect grade and risk are automatically evaluated through an algorithm, and real-time quantification and dynamic evaluation of a detection result are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of integrity detection of oil and gas wellhead devices and station process pipelines, and in particular to a wellhead device and station process detection evaluation management system, method and product. Background Art

[0002] During oil and gas field development, integrity management of wellhead equipment and process pipelines at stations is a crucial step in ensuring safe production. Currently, domestic oil and gas fields rely primarily on empirical judgment and visual inspections to manage wellhead equipment and process pipelines at stations, lacking unified testing specifications and quantitative assessment standards. There is often no clear basis for the timing, location, replacement cycle, and scope of testing, making it difficult to accurately identify safety hazards. Furthermore, wellhead equipment and process pipelines at stations often suffer from defects such as wall thinning, cracks, or pitting due to corrosion, erosion, and vibration fatigue. Existing technologies lack specialized equipment records and continuous testing data, making it difficult to analyze corrosion trends and impact patterns.

[0003] Existing inspection and management of oil and gas wellhead equipment and pipelines at field stations primarily rely on manual inspections, traditional ultrasonic testing, and fixed online monitoring. For example, on-site inspections, leak detection using instruments such as sonar or radar, or fixed sensors installed at key pipeline locations to monitor parameters such as pressure and flow rate, are among the methods with practical limitations. Manual inspections are labor-intensive, susceptible to environmental and human factors, and lack real-time, continuous monitoring capabilities. Traditional ultrasonic testing (such as phased array ultrasonic testing) has been used for online inspection of wellhead equipment, but is severely limited by the complex wellhead structure and confined space. Existing phased array scanning probes struggle to maintain consistent scanning alignment with the flange wall, resulting in significant errors in flange end-face inspection. Phased array scanning devices are suitable for large-diameter pipes with ample space, but are difficult to install and effectively inspect on-site for complex structures such as small-diameter pipelines and valves at wellheads. While automated phased array systems, such as those from Olympus, offer high-precision inspection of pipeline welds during pipeline construction, these systems are designed for girth welds on larger-diameter pipes and lack comprehensive support for defect detection of specialized components in wellhead equipment or field station process pipelines. In addition, traditional methods often rely on manual analysis of echo signals, making it difficult to achieve quantitative assessment of defects and real-time risk grading, and data processing and management are inefficient.

[0004] In summary, existing technologies have problems in adapting to high-precision detection of complex structural components, real-time defect identification, and automation levels. It is necessary to introduce new detection terminals and intelligent analysis technologies to improve detection accuracy and efficiency. Summary of the Invention

[0005] The purpose of the present invention is to address the shortcomings of the prior art and to propose a wellhead device and site process inspection and evaluation management system, method and product. To achieve the above purpose, the embodiment of the present invention adopts the following technical solutions:

[0006] In a first aspect, an embodiment of the present invention proposes a wellhead device and a field station process detection and evaluation management system, including: a wellhead detection terminal, a field station pipeline detection terminal, a three-dimensional multi-point self-correction component, a signal conditioning and edge computing unit, a central computing unit, and a visualization management platform; wherein,

[0007] The wellhead detection terminal is installed on the periphery of the wellhead device and is arranged around the main body of the gas tree. It is used to collect acoustic echo data of the wellhead device in the circumferential direction and within the preset axial travel range;

[0008] The station pipeline inspection terminal is installed on the outer wall of the station process pipeline and is used to obtain acoustic echo data of the station process pipeline within the specified inspection window. The wellhead inspection terminal and the station pipeline inspection terminal both include an annular ultrasonic probe array and a scanning drive mechanism, which is used to drive the annular ultrasonic probe to rotate.

[0009] The three-dimensional multi-point self-correction component is coaxially arranged with the wellhead detection terminal and the station pipeline detection terminal, and multiple reflection surface groups are arranged at equal intervals along the circumference and at at least two different axial heights;

[0010] The signal conditioning and edge computing unit is used to drive the annular ultrasonic probe array to emit sound beams, which are reflected by multiple reflection surface groups. It is also used to obtain acoustic echo data detected by the wellhead detection terminal and the station pipeline detection terminal respectively. The acoustic echo data includes two-order or multi-order echo signals. After receiving the echo signals of each order, the echo signals are identified through a time-amplitude joint clustering algorithm. The sound velocity, sound range, and probe posture parameters are solved in real time based on the propagation time difference between the same sound beam and the different order echoes to complete three-dimensional multi-point self-calibration and obtain three-dimensional acoustic data based on the acoustic echo data.

[0011] The central computing unit is used to call the deep learning model based on 3D acoustic data to perform image segmentation and size quantification of wall thickness thinning, cracks and pits to obtain defect recognition data, and output the risk level according to the preset threshold based on the defect recognition data;

[0012] The visual management platform is used to display risk levels in a hierarchical manner and push them to the monitoring system through the application program interface.

[0013] Preferably, each reflection surface group of the three-dimensional multi-point self-correction component contains 6 to 8 reflection surfaces, which are arranged at equal intervals along the circumference, and each reflection surface has an inclination angle of 45° to 55°; two axially adjacent groups of reflection surfaces are staggered by 1 / 2 of the interval angle along the circumference to form a spiral distribution, so that the same sound beam can produce multiple orders of echoes at different relative heights.

[0014] Preferably, after identifying the echo signal, the signal conditioning and edge computing unit applies dual-threshold filtering of a distance threshold Δd and a phase threshold Δφ to the noise signal, where Δd≤0.5mm and Δφ≤5°.

[0015] Preferably, the multi-order echo signal includes a first echo and a second echo, and the signal conditioning and edge computing unit is used to calculate the sound speed through the formula v = 2×(d2-d1) / (t2-t1), where d1 and d2 are the sound paths of the first echo and the second echo, respectively, and t1 and t2 are the arrival times corresponding to the first echo and the second echo, respectively. The calculated sound speed is used to correct the real-time sound paths of all channels of the annular ultrasonic probe array.

[0016] Preferably, the central computing unit is further used to perform Bayesian posterior confidence calculation on the defect recognition data, combine the sound pressure-spectrum-time three-dimensional feature vector multimodal fusion and output the risk level according to a preset threshold.

[0017] In a second aspect, an embodiment of the present invention provides a method for detecting and evaluating a wellhead device and a station process. The method is based on the wellhead device and station process detection and evaluation management system proposed in the above embodiment, and includes the following steps:

[0018] S1, collects acoustic echo data from the wellhead detection terminal and the station pipeline detection terminal, calculates the real-time sound velocity and corrects the probe spatial coordinates;

[0019] S2, driving the annular ultrasonic probe array to complete circumferential and axial scanning along a spiral trajectory to emit an acoustic beam and obtain acoustic echo data, where the acoustic echo data includes two-order or multi-order echo signals;

[0020] S3, performs least squares linear detrending and multi-scale wavelet denoising on each order echo signal, and uses time-amplitude joint clustering algorithm to identify echo signals;

[0021] S4, calculates the sound velocity v = 2Δd / Δt based on the echo arrival time difference Δt and the acoustic path difference Δd, and uses v to correct the acoustic path in real time. The sound velocity, acoustic path, and probe posture parameters are solved in real time based on the propagation time difference between echoes of different orders of the same acoustic beam to complete three-dimensional multi-point self-calibration and obtain three-dimensional acoustic data based on the acoustic echo data.

[0022] S5: Input the 3D acoustic data into the deep learning model, perform image segmentation and size quantification on wall thickness reduction, cracks, and pits to obtain defect recognition data, and output the defect risk level according to the preset threshold based on the defect recognition data;

[0023] S6: When the ambient temperature or pressure changes and the sound velocity changes by more than 0.5%, the system automatically returns to S4 and re-executes the three-dimensional multi-point self-calibration;

[0024] S7 is used to display the defect risk level in a graded manner and push it to the monitoring system through the application program interface.

[0025] Preferably, in S3, the multi-scale wavelet denoising process adopts the db4 wavelet basis and performs SURE threshold soft threshold truncation on the high frequency coefficients after three-layer decomposition.

[0026] Preferably, in S5, the defect risk level is divided into three levels: A, B, and C according to the wall thickness thinning rate δ, where level A corresponds to δ≤10%, level B corresponds to 10%<δ≤25%, and level C corresponds to δ>25%.

[0027] Preferably, in S2 , the circumferential step angle Δθ of the spiral track is ≤5°, and the axial step distance Δz is ≤2 mm.

[0028] In a third aspect, an embodiment of the present invention proposes a computer program product, including a non-temporary computer-readable storage medium, in which computer-executable instructions are stored. When the instructions are executed by a processor, the processor executes the method proposed in the above embodiment.

[0029] Beneficial effects:

[0030] Through the supporting wellhead detection terminal and station pipeline detection terminal, online ultrasonic scanning and data collection can be performed on wellhead equipment and pipelines. Using three-dimensional ultrasonic field self-correction, the phased array sound field is precisely calibrated to achieve high-precision three-dimensional acoustic imaging under complex structures, ensuring good coupling between the probe and the object being inspected, and accurately acquiring defect echoes. Combining multi-order echo recognition algorithms and deep learning models, the collected ultrasonic data is intelligently analyzed, significantly improving the defect recognition rate and analysis efficiency, and effectively realizing the automatic extraction and classification of defect features. Algorithms automatically evaluate defect levels and risks, achieving real-time quantification and dynamic evaluation of detection results. This significantly improves the accuracy and real-time performance of wellhead equipment and pipeline defect identification, enhances the automation level of the detection process and data management efficiency, and solves the problems of traditional detection's difficulty in adapting to complex structures and lagging data analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present invention. The same reference numerals are used throughout the accompanying drawings to denote the same components. In the accompanying drawings:

[0032] Figure 1 A schematic flow chart of a wellhead device and a field station process detection and evaluation method provided in one embodiment of the present invention;

[0033] Figure 2 A schematic diagram of the structure of a wellhead device and a station process inspection and evaluation management system provided by one embodiment of the present invention;

[0034] Figure 3 A diagram showing a detection scenario for a wellhead device according to an embodiment of the present invention;

[0035] Figure 4 This is an echo principle diagram of the wellhead device and field station process detection and evaluation management system provided by one embodiment of the present invention;

[0036] Figure 5 This is a visual interface for the wellhead equipment and site process inspection and evaluation management system provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0037] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended only to explain the present invention and are not to be construed as limiting the present invention. It should be noted that similar reference numerals and letters represent similar items in the following drawings, and therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings.

[0038] Example:

[0039] See also Figures 2 to 5As shown, in the first aspect, an embodiment of the present invention proposes a wellhead device and a site process detection and evaluation management system, including: a wellhead detection terminal, a site pipeline detection terminal, a three-dimensional multi-point self-correction component, a signal conditioning and edge computing unit, a central computing unit and a visualization management platform; wherein the wellhead detection terminal is installed on the periphery of the wellhead device and is arranged around the gas production tree body, for collecting acoustic echo data of the wellhead device in the circumferential direction and within a preset axial stroke range; the site pipeline detection terminal is sleeved on the outer wall of the site process pipeline, for obtaining acoustic echo data of the site process pipeline within a specified detection window; the wellhead detection terminal and the site pipeline detection terminal both include an annular ultrasonic probe array and a scanning drive mechanism, and the scanning drive mechanism is used to drive the annular ultrasonic probe to rotate; the three-dimensional multi-point self-correction component is coaxially arranged with the wellhead detection terminal and the site pipeline detection terminal, respectively, and a plurality of reflection surface groups are arranged at equal intervals along the circumference and at at least two different axial heights; The signal conditioning and edge computing unit is used to drive the annular ultrasonic probe array to emit a sound beam, which is reflected by multiple reflection surface groups. It is also used to obtain the acoustic echo data of the sound beam reflected by the reflection surface group and detected by the wellhead detection terminal and the station pipeline detection terminal. The acoustic echo data includes two-order or multi-order echo signals. After receiving the echo signals of each order, the echo signals are identified by the time-amplitude joint clustering algorithm, and the sound velocity, sound range and probe posture parameters are solved in real time based on the propagation time difference between the different order echoes of the same sound beam to complete three-dimensional multi-point self-correction and obtain three-dimensional acoustic data based on the acoustic echo data. The central computing unit is used to call the deep learning model based on the three-dimensional acoustic data, perform image segmentation and size quantification on wall thinning, cracks and pits to obtain defect identification data, and output the risk level according to the preset threshold based on the defect identification data. The visual management platform is used to display the risk level in a graded manner and push it to the monitoring system through the application interface.

[0040] In specific implementation, the wellhead detection terminal and the station pipeline detection terminal are respectively equipped with an annular ultrasonic probe array for transmitting and receiving ultrasonic waves to scan and detect pipelines and wellhead structures. The probe array can adopt, but is not limited to, a multi-channel piezoelectric ceramic array or a phased array design. Each array is composed of multiple independent elements, and the array form can be linear, matrix or annular. For example, to cover a larger area, the probe can adopt a two-dimensional matrix array or a concentric annular array, and be equipped with a 32-channel or 64-channel independent signal interface to ensure high sampling rate and resolution of the echo signal. The elements in each array can be excited in sequence to achieve electronic scanning. Among them, the surface of the ultrasonic probe is covered with ruby ​​material (dielectric constant optimized) to reduce the absorption and scattering of ultrasonic waves by the sulfur / wax layer.

[0041] The three-dimensional multi-point self-correction component includes multiple groups of reflection surface units, which are used to calibrate the position and delay characteristics of the probe in space. Each group of reflection surfaces usually includes multiple (for example, 3 to 8) reflection plates made of high acoustic impedance materials, commonly used materials are stainless steel or aluminum alloy. These reflection plates can be plane reflection surfaces or spherical reflection surfaces, such as flat-bottomed spherical reflection plates, and are fixed by adjustable supports. The reflection plates are evenly distributed around the probe and installed at a predetermined distance coaxial with the probe. The inclination angle (for example, 45°) can be set to generate multi-directional echoes. Using multiple reflection plates with known geometric positions, the system automatically calibrates the time compensation of each probe element by emitting pulses and receiving echoes returned from each reflection plate to ensure three-dimensional positioning accuracy.

[0042] The signal conditioning and edge computing unit is responsible for preprocessing and preliminary analysis of the ultrasonic echoes collected by the probe. This unit is implemented based on an embedded platform (such as an ARM / DSP / FPGA hybrid structure), in which the FPGA is responsible for high-speed parallel data processing, and the ARM / DSP is responsible for system control and algorithm processing. The analog front end (AFE) amplifies and performs anti-aliasing filtering on the received echo signal, and then obtains the digital signal through the analog-to-digital converter. The digital signal is filtered, downsampled, and envelope detected by programmable logic. For example, the DSP processing module can be used to filter out noise and enhance the echo signal in real time. The processed echo signal is sent to the edge computing module. The algorithm module detects the signal peak and identifies multiple echoes, classifies and clusters the echo data according to the physical depth, and extracts the candidate locations and features of defects such as the inner solder layer. Finally, the processed feature data and echo image are transmitted to the central computing unit.

[0043] The central computing unit receives the aggregated echo data from each detection terminal and uses a deep learning model to reconstruct images and identify defects. The deep learning model can analyze the aggregated ultrasonic echo sequence or the reconstructed scanned image to achieve automatic segmentation detection and size quantification of defects. For example, the improved Mask R-CNN is used to perform pixel-level segmentation of defects in ultrasonic scanned images and accurately segment abnormal echo areas in scanned images. Subsequently, the defect size is calculated and the risk level is evaluated based on the shape and echo characteristics of the defect. The model output includes information such as the spatial position, geometric size, and risk level of the defect. Based on the analysis results of the central computing unit, the visual management platform displays the detection area and defect distribution in real time in the graphical interface, such as Figure 5 The visual interface shown in the figure provides early warning and reporting functions.

[0044] In summary, the examples detailed the collaborative working of the system's components and their technical principles: a multi-channel annular ultrasonic probe array for rapid scanning, a three-dimensional reflector for multi-point calibration, edge units for signal acquisition and preprocessing, and a central unit for defect segmentation and quantification based on deep learning. This coordinated effort enables efficient detection and precise location of internal defects in pipelines and wellhead structures, laying the foundation for subsequent maintenance.

[0045] This embodiment further expands upon the preferred structure, parameters, and processing methods of the system to improve three-dimensional calibration accuracy and defect risk identification efficiency. Preferably, each reflective surface group of the three-dimensional multi-point self-correction assembly includes 6 to 8 reflective surfaces, arranged at equal intervals along the circumference, with each reflective surface having an inclination angle of 45° to 55°. Two axially adjacent groups of reflective surfaces are staggered circumferentially by 1 / 2 of an angle to form a spiral distribution, so that the same sound beam generates multiple-order echoes at different specific heights. The "specific height" is used to emphasize the reflection points of the sound beam at different height levels in three-dimensional space, thereby achieving the generation of multiple-order echoes and enhancing the system's self-correction or signal processing capabilities.

[0046] During specific implementation, the three-dimensional multi-point self-correction component includes multiple groups of optimized reflection surface units, for example, each group uses 4 reflection surfaces, which together form a multi-directional calibration. The reflection surface material is preferably aluminum alloy or ceramic to obtain stable sound reflection. The geometric design of the reflection surface includes planes and inclined planes: for example, two plane reflection plates and two inclined wedge-shaped reflection plates are set in each group, where the plane plates are parallel to the probe surface and the wedge plates are inclined at about 45°. These reflection plates are fixed by precision brackets and arranged equidistantly in different circumferential directions in front of the probe (such as one every 90°). Using a flat-bottomed spherical reflection plate as a reference can further improve the echo sensitivity. By adjusting the number and inclination of the reflection plates, this embodiment can cover echoes in more directions and further improve the three-dimensional accuracy of the calibration. After identifying the echo signal, the signal conditioning and edge computing unit uses a dual threshold filtering of the distance threshold Δd and the phase threshold Δφ for the noise signal, Δd≤0.5mm, Δφ≤5°. The multi-order echo signal includes the first echo and the second echo, that is Figure 5 As shown, the first and second bottom surface echoes are shown. The signal conditioning and edge computing unit calculates the sound velocity using the formula v = 2 × (d2-d1) / (t2-t1), where d1 and d2 are the sound paths of the first and second echoes, respectively, and t1 and t2 are the arrival times of the first and second echoes, respectively. The calculated sound velocity is used to correct the real-time sound paths of all channels in the annular ultrasonic probe array. The central computing unit also calculates Bayesian posterior confidence for the defect identification data, combines the multimodal fusion of the three-dimensional sound pressure, spectrum, and time feature vectors, and outputs the risk level according to a preset threshold.

[0047] Deep learning model in the central computing unit: The model can perform fine-grained analysis of edge-preprocessed echo data to achieve segmentation and quantification of defect echo images or volume data. After obtaining the defect segmentation image, the system automatically measures the defect size (such as length, width, and depth) and compares it with the preset standard to determine the risk level. In addition, this embodiment adds a deep learning-based defect classification module that can subdivide defect types (such as corrosion pits, cracks, or weld defects) according to echo and size characteristics. In this way, risk identification efficiency is improved and manual participation is reduced.

[0048] The system described in this embodiment significantly improves 3D calibration and defect identification performance, while significantly reducing risk warning response time, enabling near-real-time monitoring. The visualization management platform dynamically annotates defects on the 3D pipeline model and uses color-coded markers based on risk level, helping operations and maintenance personnel quickly locate and address potential risks.

[0049] See also Figure 1 In a second aspect, an embodiment of the present invention proposes a method for detecting and evaluating a wellhead device and a station process. The method is based on the wellhead device and station process detection and evaluation management system proposed in the above embodiment, and includes the following steps:

[0050] S1, collects acoustic echo data from the wellhead detection terminal and the station pipeline detection terminal, calculates the real-time sound velocity and corrects the probe spatial coordinates;

[0051] S2, driving the annular ultrasonic probe array to complete circumferential and axial scanning along a spiral trajectory to emit an acoustic beam and obtain acoustic echo data, where the acoustic echo data includes two-order or multi-order echo signals;

[0052] S3, performs least squares linear detrending and multi-scale wavelet denoising on each order echo signal, and uses time-amplitude joint clustering algorithm to identify echo signals;

[0053] S4, calculates the sound velocity v = 2Δd / Δt based on the echo arrival time difference Δt and the acoustic path difference Δd, and uses v to correct the acoustic path in real time. The sound velocity, acoustic path, and probe posture parameters are solved in real time based on the propagation time difference between echoes of different orders of the same acoustic beam to complete three-dimensional multi-point self-calibration and obtain three-dimensional acoustic data based on the acoustic echo data.

[0054] S5: Input the 3D acoustic data into the deep learning model, perform image segmentation and size quantification on wall thickness reduction, cracks, and pits to obtain defect recognition data, and output the defect risk level according to the preset threshold based on the defect recognition data;

[0055] S6: When the ambient temperature or pressure changes and the sound velocity changes by more than 0.5%, the system automatically returns to S4 and re-executes the three-dimensional multi-point self-calibration;

[0056] S7 is used to display the defect risk level in a graded manner and push it to the monitoring system through the application program interface.

[0057] The implementation steps of this method include: acoustic data acquisition, probe motion control, signal denoising and recognition, sound velocity calculation and correction, defect identification and quantification, risk level output and visualization push. The specific operations are as follows:

[0058] A multi-channel ultrasonic transducer array is deployed to transmit and receive ultrasonic signals using a pulse-echo method, performing omnidirectional acoustic scanning of the pipeline's outer wall. (The probe array can be installed at the wellhead or on a downhole device.) The collected echo signals contain information about pipe wall thickness and defects, which are initially sampled and digitized in real time by an edge computing unit.

[0059] The probe's self-rotation and axial displacement mechanism enables spiral scanning of the pipe's outer wall. Specifically, the probe is driven by a motor to rotate and simultaneously advance axially to ensure continuous coverage of the pipe's outer wall. This mode is similar to the operation of the IRIS (Internal Rotational Detection System) probe, ensuring complete inspection of the entire pipe's outer wall. The probe's movement is ensured to always follow the pipe's centerline through self-calibration components (such as a positioning centering device), and its position is accurately recorded using an angle encoder and a linear encoder.

[0060] The collected raw acoustic signal is first subjected to real-time denoising in the edge computing unit. A wavelet threshold denoising algorithm is employed: the Daubechies-4 (db4) wavelet basis is used to perform a three-level wavelet decomposition of the signal. SURE (Stein Unbiased Risk Estimation) soft thresholding is then applied to each level of detail coefficients. SURE soft thresholding effectively suppresses noise while preserving signal characteristics. After denoising, feature extraction is performed on the echo signal, extracting information such as the echo arrival time, amplitude, and waveform shape from the scanned waveform to facilitate subsequent defect location.

[0061] The sound velocity calibration adopts a dual method: one is to install a reference calibration block or an area of ​​known thickness, and use the known distance and the measured ultrasonic flight time to calculate the sound velocity; the other is to measure the environmental parameters such as the medium temperature in real time and perform temperature compensation. Specifically, the system calculates the sound velocity in the current medium through an ultrasonic measurement of a fixed distance (such as the thickness of a reference block or a known section of pipe length). The sound velocity can be determined jointly using known distance, time of flight (TOF) and temperature values. During the actual detection process, the edge computing unit continuously monitors the corrected sound velocity parameters; if it is detected that the current sound velocity changes by more than a threshold of 0.5% compared to the last calibration value (for example, sound velocity fluctuations caused by temperature or material reasons), the recalibration step is automatically triggered to ensure the accuracy of subsequent thickness measurements.

[0062] The central computing unit performs an in-depth analysis of the denoised echo signal to identify and quantify defects. Threshold detection and timing analysis algorithms are used: for example, a zero-crossing threshold algorithm (Zero-CrossThresholding) is used, which calculates the pipe wall thickness from the zero-crossing point of the first echo signal that crosses the detection threshold. This algorithm is insensitive to coupling and echo amplitude fluctuations and can achieve higher thickness measurement accuracy. By comparing the current measured bottom surface echo moment with the calibrated sound velocity, the current remaining thickness of the pipe wall can be calculated (and compared with the nominal thickness to obtain the thinning rate). If there are discrete defects (cracks, pitting), they will generate additional echoes before the bottom surface echo, and the system can also identify their location and depth through multi-dimensional detection. The processing results output parameters such as defect location, type (pitting corrosion, groove corrosion, cracks, etc.) and maximum wall thickness thinning rate.

[0063] The detected defects are risk graded according to the thinning rate of the pipe wall (for example, divided into three levels of thinning rate A, B, and C), and the results are pushed graphically to the monitoring center interface. The specific classification can be defined as: Class A (low risk, thinning rate is less than the threshold α), Class B (medium risk, thinning rate is between thresholds α and β), and Class C (high risk, thinning rate is greater than the threshold β). The central computing unit displays the thickness measurement and defect grading results in the form of a grid graph, and different risk levels are distinguished by color coding. For example, the use of the corrosion grid view to output color areas according to the upper and lower limits of the residual thickness can clearly and intuitively show the severity of corrosion on the outer wall of the pipeline. The final results are pushed to the on-site operator or monitoring center through the platform to facilitate the arrangement of maintenance or pipeline management.

[0064] The above steps are respectively implemented by the system's probe array, self-calibration components, edge computing unit and central computing unit hardware, as well as the corresponding software algorithms.

[0065] A multi-channel ultrasonic probe assembly, consisting of a rotating turbine head, a focusing probe, and a centering mechanism, is installed outside the wellhead assembly or casing pipe. The turbine head rotates the probe, while an internal 45° reflector deflects the ultrasonic beam emitted by the probe toward the pipe wall. The centering mechanism ensures that the probe always moves along the central axis of the pipe. The probe array is connected to an edge computing unit via a cable, enabling real-time sampling of high-frequency signals.

[0066] The edge computing unit, located near the wellhead, houses a high-speed processor / FPGA and performs real-time preprocessing of collected raw data, including time series acquisition, signal filtering, waveform extraction, wavelet denoising, and local feature extraction. The edge unit runs embedded firmware and executes a denoising algorithm based on preset parameters (db4 wavelet, three-layer decomposition, and SURE threshold). It also performs tasks such as preliminary sound velocity calculation and threshold detection, and transmits key features and calibration data to the central computing unit.

[0067] The central computing unit, deployed in a monitoring center or cloud server, is responsible for in-depth data processing and decision-making. The central unit runs more complex software algorithms to perform quantitative defect analysis, risk assessment, and report generation on the processed data uploaded by the edge units. Its software modules include an ultrasonic thickness measurement algorithm (which can utilize the aforementioned zero-crossing threshold algorithm), corrosion rate statistics, a risk grading module, and a visualization engine. The central unit is responsible for pushing the final results to the human-machine interface and database.

[0068] Preferably, in S3, the multi-scale wavelet denoising process adopts the db4 wavelet basis and performs SURE threshold soft threshold truncation on the high frequency coefficients after three-layer decomposition.

[0069] In the signal denoising step, a Daubechies-4 (db4) wavelet basis is selected for three-level decomposition, and noise reduction is performed on the detail coefficients at each level using a soft threshold determined by the Stein Unbiased Risk Estimation (SURE) criterion. This method effectively preserves ultrasonic signal details and suppresses high-frequency noise, improving subsequent defect detection accuracy.

[0070] Preferably, in S5, the defect risk level is divided into three levels: A, B, and C according to the wall thickness thinning rate δ, where level A corresponds to δ≤10%, level B corresponds to 10%<δ≤25%, and level C corresponds to δ>25%.

[0071] Risk levels are set based on the local or overall thinning rate of the pipe wall. Thinning rates are preferably categorized into three levels: A (low risk, thinning rate ≤ X%), B (medium risk, X% < thinning rate ≤ Y%), and C (high risk, thinning rate > Y%), where X and Y are thresholds set by experience or regulations (for example, X = 20% and Y = 50%). The actual grading can be adjusted based on pressure vessel or pipeline safety evaluation standards.

[0072] Preferably, in S2 , the circumferential step angle Δθ of the spiral track is ≤5°, and the axial step distance Δz is ≤2 mm.

[0073] As the probe moves axially along the pipe, the spiral scan angle step is set to Δθ ≤ 5° and the axial step Δz ≤ 2mm to ensure scan density. This design allows for a complete 2D thickness map within a reasonable scanning time, improving both inspection coverage and resolution. The system monitors the deviation of the current medium sound velocity from the calibration value in real time, triggering a recalibration process when the relative change exceeds 0.5%. This threshold is chosen to account for the sensitivity of the material sound velocity to factors such as temperature; timely correction prevents accumulation of thickness calculation errors.

[0074] When deployed at an oil and gas wellsite, this method operates collaboratively according to the following process: First, the inspection device is deployed at the pipeline segment or wellhead to be inspected. The probe array is installed and connected to the self-calibration component. The edge computing unit is then installed at the nearest operating station and connected to the network interface of the central computing unit. When the inspection is initiated, the central unit sends acquisition control commands to the edge unit, triggering the probe's spiral scanning motion and beginning echo data acquisition. The edge unit controls the probe's motion according to preset time and stepping rules, while also performing data preprocessing. After a complete spiral scanning cycle, the edge unit uploads the processed waveform features and correction data to the central unit via a wired or wireless network. The central unit aggregates the thickness data from each scanned section, performs defect identification, thickness comparison, and thinning rate calculation, and then outputs a risk level based on predefined standards. The results are then pushed to operations and maintenance personnel on the monitoring platform in the form of a grid chart and alerts. Throughout this process, hardware and software modules work collaboratively, with data and control flows interacting between the edge computing unit and the central computing unit, forming a complete closed loop from field acquisition to backend decision-making.

[0075] The execution subject of the wellhead device and site process detection and evaluation method includes but is not limited to at least one of the electronic devices such as the server, terminal, etc. that can be configured to execute the method provided in the embodiment of the present application. In other words, the wellhead device and site process detection and evaluation method can be executed by software or hardware installed in the terminal device or the server device, and the software can be a blockchain platform. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc. The server can be an independent server, or it can be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0076] In a third aspect, an embodiment of the present invention proposes a computer program product, including a non-temporary computer-readable storage medium, in which computer-executable instructions are stored. When the instructions are executed by a processor, the processor executes the method proposed in the above embodiment.

[0077] The present invention also discloses a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps of the above-mentioned wellhead device and site process detection and evaluation method provided by the present invention. The computer program product should be understood as a software product that mainly implements its solution through a computer program, such as a program product integrated in the cloud or a software library.

[0078] Throughout this specification, reference to terms such as "one embodiment," "some embodiments," "example," "specific example," "one implementation," "a preferred implementation," or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with that embodiment or example are included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0079] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to the embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the claims and their equivalents.

Claims

1. Wellhead equipment and site process inspection and evaluation management system, characterized by: include: Wellhead detection terminal, station pipeline detection terminal, three-dimensional multi-point self-correction component, signal conditioning and edge computing unit, central computing unit and visualization management platform; among them, The wellhead detection terminal is installed on the periphery of the wellhead device and is arranged around the main body of the gas tree, and is used to collect acoustic echo data of the wellhead device in the circumferential direction and within a preset axial travel range; The station pipeline detection terminal is mounted on the outer wall of the station process pipeline and is used to obtain acoustic echo data of the station process pipeline within a specified detection window; the wellhead detection terminal and the station pipeline detection terminal both include an annular ultrasonic probe array and a scanning drive mechanism, and the scanning drive mechanism is used to drive the annular ultrasonic probe to rotate; The three-dimensional multi-point self-correction assembly is coaxially arranged with the wellhead detection terminal and the station pipeline detection terminal, and a plurality of reflection surface groups are arranged at equal intervals along the circumference and at at least two different axial heights; The signal conditioning and edge computing unit is used to drive the annular ultrasonic probe array to emit a sound beam, which is reflected by the multiple reflection surface groups; it is also used to respectively obtain acoustic echo data of the sound beam reflected by the reflection surface group and detected by the wellhead detection terminal and the station pipeline detection terminal, wherein the acoustic echo data includes two-order or multi-order echo signals, and after receiving the echo signals of each order, the echo signals are identified by a time-amplitude joint clustering algorithm, and the sound velocity, sound range and probe posture parameters are solved in real time based on the propagation time difference between the different order echoes of the same sound beam to complete three-dimensional multi-point self-correction, and three-dimensional acoustic data is obtained based on the acoustic echo data; The central computing unit is configured to call a deep learning model based on the three-dimensional acoustic data, perform image segmentation and size quantification on wall thickness reduction, cracks, and pits to obtain defect identification data, and output a risk level according to a preset threshold based on the defect identification data; The visual management platform is used to display the risk levels in a hierarchical manner and push them to the monitoring system through an application program interface.

2. The system according to claim 1, wherein Each reflection surface group of the three-dimensional multi-point self-correction component includes 6 to 8 reflection surfaces, which are arranged at equal intervals along the circumference, and each reflection surface has an inclination angle of 45° to 55°; two axially adjacent groups of reflection surfaces are staggered by 1 / 2 of the interval angle along the circumference to form a spiral distribution, so that the same sound beam can generate multiple-order echoes at different specific heights.

3. The system according to claim 1, wherein: After identifying the echo signal, the signal conditioning and edge computing unit applies dual-threshold filtering of a distance threshold Δd and a phase threshold Δφ to the noise signal, where Δd is ≤ 0.5 mm and Δφ is ≤ 5°.

4. The system according to claim 3, wherein: The multi-order echo signal includes a first echo and a second echo. The signal conditioning and edge computing unit is used to calculate the sound velocity using the formula v = 2×(d2-d1) / (t2-t1), where d1 and d2 are the sound paths of the first echo and the second echo, respectively, and t1 and t2 are the arrival times corresponding to the first echo and the second echo, respectively. The calculated sound velocity is used to correct the real-time sound paths of all channels of the annular ultrasonic probe array.

5. The system according to claim 1, wherein: The central computing unit is further used to perform Bayesian posterior confidence calculation on the defect identification data, combine the sound pressure-spectrum-time three-dimensional feature vector multimodal fusion and output the risk level according to a preset threshold.

6. A method for detecting and evaluating wellhead equipment and station processes, the method being based on the wellhead equipment and station process detection and evaluation management system according to any one of claims 1 to 5, characterized in that: The following steps are involved: S1, collecting acoustic echo data from the wellhead detection terminal and the station pipeline detection terminal, calculating the real-time sound velocity and correcting the probe spatial coordinates; S2, driving the annular ultrasonic probe array to complete circumferential and axial scanning along a spiral trajectory to emit an acoustic beam and obtain acoustic echo data, wherein the acoustic echo data includes two-order or multi-order echo signals; S3, performs least squares linear detrending and multi-scale wavelet denoising on each order echo signal, and uses time-amplitude joint clustering algorithm to identify echo signals; S4, calculates the sound velocity v = 2Δd / Δt based on the echo arrival time difference Δt and the acoustic path difference Δd, and uses v to correct the acoustic path in real time. The sound velocity, acoustic path, and probe posture parameters are solved in real time based on the propagation time difference between echoes of different orders of the same acoustic beam to complete three-dimensional multi-point self-calibration and obtain three-dimensional acoustic data based on the acoustic echo data. S5, inputting the 3D acoustic data into the deep learning model, performing image segmentation and size quantification on wall thickness reduction, cracks, and pits to obtain defect recognition data, and outputting a defect risk level according to a preset threshold based on the defect recognition data; S6: When the ambient temperature or pressure changes and the sound velocity changes by more than 0.5%, the system automatically returns to S4 and re-executes the three-dimensional multi-point self-calibration; S7 is used to display the defect risk level in a graded manner and push it to the monitoring system through the application program interface.

7. The method according to claim 6, wherein In S3, the multi-scale wavelet denoising process uses the db4 wavelet basis and performs SURE threshold soft threshold truncation on the high frequency coefficients after three-layer decomposition.

8. The method according to claim 6, wherein In S5, the defect risk level is divided into three levels: A, B, and C according to the wall thickness reduction rate δ, where level A corresponds to δ≤10%, level B corresponds to 10%<δ≤25%, and level C corresponds to δ>25%.

9. The method according to claim 6, wherein In S2 , the circumferential step angle Δθ of the spiral track is ≤5°, and the axial step distance Δz is ≤2mm.

10. A computer program product, characterized in that The invention comprises a non-transitory computer-readable storage medium, wherein the storage medium stores computer-executable instructions, and when the instructions are executed by a processor, the processor is caused to perform the method according to any one of claims 6 to 9.

Images