Welded pipe end welding seam transverse ultrasonic phased array automatic judgment method, system and equipment

Abstract

The invention belongs to the field of nondestructive testing, and relates to a method, a system and equipment for automatically judging a transverse ultrasonic phased array of a welding seam at the pipe end of a welded pipe. The problems that manual detection of the pipe end of the welded pipe is high in subjectivity and low in efficiency, and digital tracing cannot be achieved are solved. The method comprises the following steps: acquiring transverse ultrasonic phased array atlas data of a welding seam; selecting target image data containing the welding seam and the adjacent area thereof; identifying an original weld joint area, a weld joint transition area and a weld joint polishing area which are longitudinally distributed along the weld joint in the target image data, and distinguishing a weld joint body area and a weld toe area; identifying and filtering an inherent structure echo signal based on the difference characteristics of the sound path distance and the wave amplitude to obtain residual image data; and automatically judging whether the transverse defect of the welding seam is qualified or not based on a comparison result of the signal characteristics in the residual image data and the judgment threshold. According to the invention, the automatic and standardized detection and judgment of the transverse defect of the welding seam at the pipe end of the welded pipe are realized, and the objectivity and efficiency of detection and the traceability of a result are improved.

Description

Technical Field

[0001] This invention belongs to the field of non-destructive testing, and specifically relates to an automatic evaluation method, system and equipment for transverse ultrasonic phased array of weld seams at the pipe end of a welded pipe. Background Technology

[0002] In the manufacturing process of welded pipes for oil and gas transportation, to ensure the welding quality of circumferential welds, non-destructive testing is often required on the weld seams at the pipe ends (e.g., within 300 mm of the pipe end). Due to blind spots in online automated ultrasonic testing equipment, manual ultrasonic testing is currently widely used for re-inspection in this area. Manual testing relies on operators holding the probe, scanning, and observing the waveforms or images on the instrument screen in real time (e.g., A-scan waveforms, B-scan patterns), and relying entirely on personal experience to identify, locate, and judge defect signals.

[0003] This traditional manual inspection method has several significant drawbacks: First, the accuracy and consistency of the inspection results are highly dependent on the skill level, experience, and work status of the inspectors, making it subjective and prone to missed detections or misjudgments, thus hindering stable quality control. Second, manual scanning is slow, labor-intensive, and inefficient, becoming a bottleneck in the production process and increasing labor costs. Third, the inspection process and results are usually stored in paper records or simple electronic notes, resulting in scattered, unstructured data that is difficult to digitize, manage, and trace long-term, hindering quality statistical analysis, process optimization, and accountability.

[0004] While some technical solutions exist for automated ultrasonic testing of welded pipes, such as those targeting the machined areas of the pipe end face (e.g., bevels, blunt edges), these solutions primarily address the detection of multi-directional defects under specific end-face geometries. For the circumferentially extending weld seam itself, especially the complex testing environment at the pipe end weld seam area where different surface states (original weld, transition zone, and grinding zone) arise due to process requirements (e.g., grinding away excess weld height), existing automated solutions struggle to effectively distinguish between fixed echoes generated by the inherent geometry of the weld seam (e.g., weld toe) and true transverse defect echoes. Their evaluation algorithms typically lack adaptability to such complex structures. Therefore, achieving automated, intelligent, and highly reliable detection and evaluation of transverse defects in weld seams at pipe ends, and creating traceable digital records, remains a pressing technical challenge in this field. Summary of the Invention

[0005] To address the aforementioned problems in the prior art, namely the high subjectivity, low efficiency, and lack of digital traceability in manual inspection of welded pipe ends, this invention provides an automatic evaluation method, system, and equipment for transverse ultrasonic phased array of weld seams at welded pipe ends.

[0006] The first aspect of this invention proposes an automatic evaluation method for transverse ultrasonic phased array weld seams at the pipe end of a welded pipe, comprising:

[0007] Ultrasonic phased array scanning was performed on the weld seam within a set range at the end of the welded pipe to obtain transverse ultrasonic phased array spectrum data of the weld seam;

[0008] From the acquired ultrasonic phased array atlas data, target image data containing the weld and its adjacent area are selected; the target image data is preprocessed to suppress noise;

[0009] Identify multiple partitions with different surface states distributed along the longitudinal direction of the weld in the target image data. The partitions include at least the original weld area that has not been ground, the weld transition area between the original weld and the ground area, and the ground weld area. For each partition, distinguish its weld body area from the weld toe area at the weld edge.

[0010] For each identified weld toe region, based on the difference in sound path and amplitude, the inherent structural echo signal generated by the welded pipe geometry is identified from the target image data; all identified inherent structural echo signals are filtered out from the target image data to obtain residual image data.

[0011] A judgment threshold is set according to the preset defect acceptance standard; based on the comparison result between the signal features in the residual image data and the judgment threshold, the transverse defects of the welded pipe end are judged as qualified or unqualified.

[0012] Furthermore, target image data containing the weld and its adjacent area is selected from the acquired ultrasonic phased array atlas data. Specifically, based on the preset position range of the weld at the pipe end, image data corresponding to the preset position range is extracted from the ultrasonic phased array atlas data as the target image data.

[0013] Furthermore, identifying multiple partitions distributed along the longitudinal direction of the weld in the target image data includes:

[0014] In the target image data, a grayscale profile curve is extracted along a preset direction parallel to the weld seam, penetrating the region.

[0015] Calculate the first derivative of the grayscale profile curve and detect the extreme points where the absolute value of the first derivative exceeds a preset first threshold. The location of the extreme points is initially marked as a potential partition boundary indicated by the abrupt change in weld surface height.

[0016] Within the adjacent intervals defined by the potential partition boundary, the gray-level co-occurrence matrix of the target image data is calculated respectively, and the contrast feature value of the gray-level co-occurrence matrix is ​​extracted.

[0017] Based on the numerical relationship and trend of the contrast feature value in adjacent intervals, the specific boundaries of the original weld area, the weld transition area, and the weld grinding area are confirmed and finally defined; wherein, the contrast feature value corresponding to the weld grinding area is expected to be smaller than that of the original weld area.

[0018] Furthermore, for each identified weld toe region, the inherent structural echo signal is identified based on the differences in sound path and amplitude, including:

[0019] A reference acoustic path correction model corresponding to the weld toe region in different zones was established based on standard samples;

[0020] Calculate the degree of matching between the actual acoustic path of the echo signal in the corresponding region of the target image data and the theoretical acoustic path given by the reference acoustic path correction model;

[0021] Calculate the difference between the echo signal amplitude of the corresponding region in the target image data and the background noise amplitude of the adjacent region;

[0022] The echo signal is generated as a confidence score of the inherent structure echo signal based on a weighted combination of the matching degree and the difference degree.

[0023] Echo signals with a confidence score higher than a preset threshold are identified as inherent structure echo signals.

[0024] Furthermore, filtering out all the identified inherent structure echo signals from the target image data specifically involves locating the region in the target image data that is determined to be an inherent structure echo signal, and setting the signal amplitude within that region to a value consistent with the image background.

[0025] Furthermore, the target image data is preprocessed to suppress noise, including: performing image scaling on the target image data to unify the analysis scale, and performing Gaussian blur filtering to eliminate random noise.

[0026] Furthermore, the evaluation is based on the comparison result between the signal features in the residual image data and the evaluation threshold, including:

[0027] Adaptive threshold segmentation is applied to the residual image data to identify all potential defective signal connected components;

[0028] For each defect signal connected region, its geometric features and amplitude features are extracted. The geometric features include at least the projected length along the transverse direction of the weld and the projected height along the depth direction of the ultrasonic beam. The amplitude features include at least the maximum wave amplitude value within the region.

[0029] The geometric or amplitude features of each defective signal connected component are compared with a preset evaluation threshold. If the geometric or amplitude features do not exceed their corresponding threshold, they are judged as qualified; otherwise, they are judged as unqualified.

[0030] Furthermore, the matching degree between the actual sound path of the echo signal in the corresponding region of the target image data and the theoretical sound path given by the reference sound path correction model is calculated, specifically as follows:

[0031] For the original weld area, the weld transition area, and the weld toe area of ​​the weld grinding area, respectively, calculate the absolute error between the actual sound path and the theoretical sound path;

[0032] The absolute error is input into a preset error function to calculate the matching degree;

[0033] The error function for the weld transition region is configured to output a higher matching degree value for the same absolute error input than the error functions for the original weld region and the weld grinding region.

[0034] A second aspect of the present invention provides an automatic evaluation system for transverse ultrasonic phased array weld seams at the ends of welded pipes, based on an automatic evaluation method for transverse ultrasonic phased array weld seams at the ends of welded pipes. The system includes:

[0035] The data acquisition module is configured to perform ultrasonic phased array scanning on the weld seam within a set range at the end of the welded pipe to acquire ultrasonic phased array spectrum data of the weld seam in the transverse direction.

[0036] The image processing module is configured to select target image data containing the weld and its adjacent area from the acquired ultrasonic phased array atlas data; and to preprocess the target image data to suppress noise.

[0037] A partition recognition module is configured to recognize multiple partitions with different surface states distributed along the longitudinal direction of the weld in the target image data. The partitions include at least the original weld area that has not been ground, the weld transition area between the original weld and the ground area, and the ground weld area. For each partition, the module distinguishes between the weld body area and the weld toe area of ​​the weld edge.

[0038] The structural echo processing module is configured to, for each determined weld toe region, identify the inherent structural echo signal generated by the welded pipe geometry from the target image data based on the difference characteristics of sound path and amplitude; and filter out all identified inherent structural echo signals from the target image data to obtain residual image data.

[0039] The defect assessment module is configured to set an assessment threshold based on a preset defect acceptance standard; and to assess whether the transverse defects of the welded pipe end weld are qualified or unqualified based on the comparison result between the signal features in the residual image data and the assessment threshold.

[0040] In a third aspect, the present invention provides an apparatus comprising:

[0041] At least one processor;

[0042] and a memory communicatively connected to at least one of the processors;

[0043] The memory stores instructions that can be executed by the processor to implement an automatic evaluation method for transverse ultrasonic phased array weld seams at the pipe end of a welded pipe.

[0044] The beneficial effects of this invention are:

[0045] This invention automatically identifies different surface state zones of welds—the original weld zone, transition zone, and grinding zone—using image processing algorithms. Specifically targeting the weld toe area of ​​each zone, it intelligently distinguishes between inherent structural echoes and actual defect echoes using a sound path correction model based on standard samples and amplitude difference characteristics. This process reduces the influence of subjective human experience, effectively avoiding misjudgments caused by structural echo interference, and making defect detection and evaluation results more objective, accurate, and reliable.

[0046] The proposed partition identification method can adapt to the continuous changes in surface condition of the weld area caused by the high-residue grinding process. To address the gradual surface changes in the weld transition zone, a differentiated matching fault-tolerant mechanism is introduced into the structural echo identification algorithm, enhancing the universality and robustness of this invention for welds in different states.

[0047] The detection and evaluation process of this invention is completed automatically, including scanning, data analysis, defect identification, and conformity determination, eliminating the need for continuous high-intensity, high-focus manual operation and real-time evaluation by inspection personnel. This significantly improves the inspection speed of a single welded pipe, reduces reliance on highly skilled workers, and saves human resources and time costs.

[0048] Automated inspection processes naturally generate structured digital records, including the original phased array spectra, preprocessed images, intermediate results from algorithmic identification (such as partition boundaries and filtered structural echo locations), and final automated evaluation conclusions. This data can be stored completely and systematically, establishing an independent digital quality profile for each welded pipe. This enables full-process traceability of product quality and provides a solid data foundation for production quality monitoring, process improvement, and problem analysis.

[0049] The automatic defect evaluation module of this invention extracts multi-dimensional features of defect signals (such as lateral length, depth direction height, and amplitude) and compares them with preset standards to achieve quantitative analysis of transverse defects in welds. This multi-parameter fusion evaluation method is more in line with the comprehensive evaluation requirements for defects in actual engineering standards, thus improving the scientific rigor and authority of the evaluation results. Attached Figure Description

[0050] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0051] Figure 1 This is a flowchart illustrating an automatic evaluation method for transverse ultrasonic phased array weld seams at the pipe end of a welded pipe according to the present invention.

[0052] Figure 2 This is a schematic diagram of the partitioning process in the transverse ultrasonic phased array automatic evaluation method for welded pipe end welds of the present invention.

[0053] Figure 3 This is a flowchart illustrating the process of identifying inherent structure echo signals in an automatic evaluation method for transverse ultrasonic phased array weld seams at the pipe end of a welded pipe according to the present invention.

[0054] Figure 4 This is a flowchart of the automatic evaluation method for transverse ultrasonic phased array weld seams at the pipe end of a welded pipe according to the present invention, which is used to determine whether the weld is qualified. Detailed Implementation

[0055] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0056] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0057] The present invention provides a first embodiment of an automatic evaluation method for transverse ultrasonic phased array weld seams at the pipe end of a welded pipe, comprising:

[0058] Step S10: Perform ultrasonic phased array scanning on the weld seam within a set range at the end of the welded pipe to obtain ultrasonic phased array spectrum data of the weld seam in the transverse direction.

[0059] Step S20: Select target image data containing the weld and its adjacent area from the acquired ultrasonic phased array atlas data; preprocess the target image data to suppress noise;

[0060] Step S30: Identify multiple partitions with different surface states distributed along the longitudinal direction of the weld in the target image data. The partitions include at least the original weld area that has not been ground, the weld transition area between the original weld and the ground area, and the ground weld area. For each partition, distinguish its weld body area from the weld toe area of ​​the weld edge.

[0061] Step S40: For each determined weld toe region, based on the difference characteristics of sound path and amplitude, identify the inherent structural echo signal generated by the welded pipe geometry from the target image data; filter out all identified inherent structural echo signals from the target image data to obtain residual image data;

[0062] Step S50: Set an evaluation threshold according to the preset defect acceptance standard; based on the comparison result between the signal features in the residual image data and the evaluation threshold, make a qualified or unqualified judgment on the transverse defects of the welded pipe end weld.

[0063] To more clearly explain a method for automatic evaluation of transverse ultrasonic phased array weld seams at the pipe ends of welded pipes, the following section combines... Figure 1 The steps in the embodiments of the present invention are described in detail below:

[0064] Step S10: Perform ultrasonic phased array scanning on the weld seam within a set range at the end of the welded pipe to obtain ultrasonic phased array spectrum data of the weld seam in the transverse direction.

[0065] The specific implementation of step S10 is as follows. To achieve automated detection of transverse defects in the weld within a set range at the end of the welded pipe, it is first necessary to acquire high-quality raw ultrasonic phased array image data. In specific implementation, an integrated ultrasonic phased array automatic detection system is used. The core of this system includes a fully digital ultrasonic phased array instrument, a multi-crystal linear array probe, a high-precision scanning mechanism and its motion controller, and computer software for data acquisition and preliminary processing. According to the specifications of the welded pipe being inspected, such as the outer diameter and wall thickness, and the requirements of the inspection standards, the probe model is selected. Typically, a linear array probe with a center frequency between 5 MHz and 10 MHz and a number of crystals of no less than 32 is selected. Based on the calculation law of the sound beam refraction angle, the sound beam delay law for exciting the probe is pre-set in the instrument to generate an oblique sound beam that can effectively cover the weld area and is mainly sensitive to transverse defects. The phased array probe is securely mounted at the end of the scanning mechanism using a special clamp. The clamp is adjusted to ensure that the sound beam emission surface of the probe is parallel to and in close contact with the surface of the welded pipe. The surface of the pipe can be the inner or outer wall. At the same time, it is ensured that the length direction of the probe array is parallel to the axial direction of the welded pipe so that the emitted fan-shaped sound beam plane is perpendicular to the weld direction.

[0066] To ensure stable acoustic coupling, a constant flow water supply system or an automatic coupling agent spraying device is used to form a uniform and continuous water film or coupling agent layer between the probe and the pipe wall.

[0067] Before conducting the inspection, key scanning parameters are set in the control software. First, the start and end positions of the scan are defined to cover the target area at the pipe end. This area is typically 300 mm axially extending from the pipe end to meet the requirements of relevant standards for re-inspection of blind spots at the pipe end. Next, the scanning step resolution is set, usually no greater than 1 mm, to ensure a high circumferential defect detection rate. Simultaneously, basic parameters such as the instrument's pulse repetition frequency, gain, filtering range, and sampling rate are set. These parameters need to be optimized based on the pipe wall thickness, material acoustic properties, and the expected defect depth to ensure the received echo signal has sufficient signal-to-noise ratio and resolution. After starting the automatic scanning program, the motion controller drives the scanning mechanism, causing the probe to rotate at a constant speed circumferentially along the welded pipe body, while simultaneously performing precise feeding along the pipe's axial direction at the preset step resolution.

[0068] At each predetermined position point during probe movement, the phased array instrument excites the probe wafers to emit ultrasonic pulses according to a preset delay rule, and simultaneously receives the echo signals from all wafers. The instrument's internal processor processes the raw A-scan signals from all channels in real time, including amplification, filtering, and digitization. Then, according to a specific imaging algorithm, such as total focusing or sector scanning, it synthesizes a two-dimensional ultrasound image slice corresponding to that scan position, i.e., a B-scan image. The control software synchronously records the precise spatial coordinates corresponding to each image slice, including circumferential angles and axial positions.

[0069] After the probe completes a full coverage scan of the designated area, this invention integrates all spatially ordered two-dimensional image slices into a complete three-dimensional dataset—a transverse ultrasonic phased array atlas of the weld, characterizing the internal structure of the weld area at the pipe end of the inspected welded pipe. This atlas data is typically saved in a specific file format, such as HDF5 or a custom binary format, containing complete amplitude information and corresponding spatial location information, providing the raw data foundation for subsequent automatic image processing and defect assessment. These steps ensure the consistency and repeatability of the inspection process, laying a reliable data foundation for automated assessment.

[0070] Step S20: Select target image data containing the weld and its adjacent area from the acquired ultrasonic phased array atlas data; preprocess the target image data to suppress noise;

[0071] In this embodiment, target image data containing the weld and its adjacent area is selected from the acquired ultrasonic phased array atlas data. Specifically, based on a preset position range of the weld at the pipe end, image data corresponding to the preset position range is extracted from the ultrasonic phased array atlas data and used as the target image data.

[0072] In practical implementation, after successfully acquiring the ultrasonic phased array three-dimensional image data of the weld seam area at the pipe end, the present invention needs to perform a data refinement step to extract the core image area directly related to the weld quality assessment and optimize the data of the area.

[0073] First, the target image data is selected. Based on pre-inputted welded pipe specifications, particularly the preset location information of the weld seam at the pipe end, the invention performs intelligent cropping within the three-dimensional image data. This preset location typically refers to a weld seam area within a specific range at the pipe end, such as within 300 mm of the pipe end. Its precise boundaries can be set in the software according to pipeline standards or process requirements. In the three-dimensional coordinate system of the data, based on the probe's spatial calibration data and the pipe's geometric model, the invention automatically calculates and delineates a three-dimensional data sub-block corresponding to the weld seam area and its adjacent heat-affected zone.

[0074] The specific operation involves: in the axial dimension of the data matrix, that is, along the length of the tube, extracting all data slices starting from the tube end and extending to a preset distance, such as 300 mm; in the circumferential dimension, retaining all data based on the complete circumferential angle range covered by the probe scan; and in the depth dimension, extracting a data layer from the inner surface of the tube to the outer surface, including a certain margin, based on the tube wall thickness and the sound beam coverage.

[0075] By using this geometrically based 3D data extraction, the present invention excludes base material data from areas far from the weld, resulting in a target image data block that is significantly reduced in size but fully contains the weld to be analyzed and its adjacent microstructure.

[0076] Next, the captured target image data block is preprocessed to suppress noise. Preprocessing mainly includes two core operations. The first operation is image scaling. This invention resamples the target image data block to a standardized pixel resolution in both the axial and circumferential spatial dimensions using a bilinear interpolation algorithm. For example, it establishes a fixed proportional relationship between the actual physical size and the image pixels, aiming to unify the scale of images under different specifications of welded pipes or different scanning parameters.

[0077] The second operation is Gaussian blur filtering. This invention performs a convolution operation between a two-dimensional discrete Gaussian convolution kernel and the scaled image data. For example, it uses a convolution kernel with a size of 5 pixels by 5 pixels and a standard deviation of 1.5, performing convolution in both the axial and circumferential planes. This smoothing filtering process effectively suppresses high-frequency random noise in the image. After the above cropping and preprocessing steps, the output is target image data with uniform size, suppressed noise, and focused on the key areas of the weld.

[0078] Step S30: Identify multiple partitions with different surface states distributed along the longitudinal direction of the weld in the target image data. The partitions include at least the original weld area that has not been ground, the weld transition area between the original weld and the ground area, and the ground weld area. For each partition, distinguish its weld body area from the weld toe area of ​​the weld edge.

[0079] See Figure 2 In this embodiment, identifying multiple partitions distributed along the longitudinal direction of the weld in the target image data includes:

[0080] Step S31: Extract the grayscale value profile curve through the region from the target image data along a preset direction parallel to the weld.

[0081] Step S32: Calculate the first derivative of the grayscale profile curve and detect the extreme points where the absolute value of the first derivative exceeds a preset first threshold. The location of the extreme points is initially marked as a potential partition boundary indicated by the abrupt change in weld surface height.

[0082] Step S33: Within the adjacent intervals defined by the potential partition boundary, calculate the gray-level co-occurrence matrix of the target image data respectively, and extract the contrast feature value of the gray-level co-occurrence matrix.

[0083] Step S34: Based on the numerical relationship and trend of the contrast feature value in adjacent intervals, confirm and finally delineate the specific boundaries of the original weld area, the weld transition area, and the weld grinding area; wherein, the contrast feature value corresponding to the weld grinding area is expected to be smaller than that of the original weld area.

[0084] The specific implementation of step S30 is as follows. After obtaining the preprocessed target image data, an automatic identification process for the longitudinal partitioning of the weld is initiated. This process aims to objectively divide the different surface state regions caused by welding and grinding processes based on ultrasonic image characteristics, and further identify the internal structure of each region. First, in the target image data, a grayscale value profile curve is extracted along a preset path parallel to the weld length and passing through its centerline, traversing the entire area to be analyzed. This curve is composed of the average grayscale value of the image region corresponding to each point along the path, intuitively reflecting the distribution of the overall intensity of the ultrasonic echo along the longitudinal direction of the weld. Its fluctuation pattern is directly related to the changes in surface geometry.

[0085] To accurately pinpoint potential locations where surface conditions undergo abrupt changes, the first derivative of the grayscale profile curve is calculated. Analysis of the derivative sequence identifies all local extrema where the absolute value of the first derivative exceeds a preset first threshold. This first threshold can be adaptively set based on the overall noise level of the image, for example, set to 2 to 3 times the average fluctuation amplitude of the profile curve. These extrema correspond to coordinates where grayscale changes drastically, and physically, they typically indicate significant inflection points in the weld surface profile; their coordinates are initially marked as potential partition boundaries.

[0086] To further verify and accurately delineate the boundaries, the process transitions to texture feature analysis. Using each continuous interval defined by the initially marked boundary points as a unit, the gray-level co-occurrence matrix is ​​calculated for the corresponding sub-region in the target image data, and contrast feature values ​​are extracted. Texture contrast values ​​quantify the drastic changes in local gray levels of the image and are closely related to the microscopic roughness of the material surface. In ultrasonic images, rough, original weld seam areas typically exhibit higher contrast values, while smooth, polished surface areas correspond to lower contrast values.

[0087] The partitioning decision is based on quantization rules. First, the global average of the contrast feature values ​​of all intervals is calculated as the evaluation benchmark. Then, each interval is analyzed sequentially along the longitudinal direction of the weld:

[0088] If the average contrast value of a certain continuous interval is consistently higher than the benchmark value by more than 30%, and the standard deviation of the contrast value within the interval is less than a stability threshold, which can be set to 10% of the benchmark value or obtained by analyzing the statistical fluctuation range of known qualified samples, then the interval is determined to be the original weld area.

[0089] If, after the original weld area, a continuous interval appears where the average contrast value shows a clear monotonically decreasing trend and the absolute value of the linear fitting slope of the decrease exceeds a preset gradient threshold, for example, in this embodiment, the contrast value decreases by more than 5% of the baseline value per millimeter length, then the interval is determined to be a weld transition area.

[0090] Following the transition area, if a continuous range of average contrast values ​​that are consistently more than 50% lower than the baseline value and fluctuate steadily appears, it is identified as a weld grinding area.

[0091] After completing the macroscopic partitioning, within each defined image sub-block, the weld body and weld toe are further subdivided. Specifically, along the direction perpendicular to the weld, i.e., the pipe wall thickness direction, the lateral gradient of pixel grayscale values ​​is calculated column by column in the sub-block image, for example using the Sobel operator. On the weld cross-section, the weld toe location, due to its abrupt geometric change, forms a significant peak in gradient amplitude. By finding pixel columns whose gradient amplitude continuously exceeds a preset edge threshold (e.g., 60% of the maximum gradient amplitude within the sub-block), and combining this with their spatial symmetry, the weld toe regions on both sides can be identified and marked. The image portion located between the two weld toe regions is defined as the weld region. Through this progressive analysis from the whole to the part, from grayscale to texture, and then to the edge, a complete and accurate automated deconstruction of the complex weld structure is achieved.

[0092] Step S40: For each determined weld toe region, based on the difference characteristics of sound path and amplitude, identify the inherent structural echo signal generated by the welded pipe geometry from the target image data; filter out all identified inherent structural echo signals from the target image data to obtain residual image data;

[0093] See Figure 3 In this embodiment, for each determined weld toe region, the inherent structural echo signal is identified based on the difference in sound path and amplitude, including:

[0094] Step S41: Establish a reference acoustic path correction model for the weld toe region corresponding to different partitions based on the standard sample;

[0095] Step S42: Calculate the matching degree between the actual sound path of the echo signal in the corresponding region of the target image data and the theoretical sound path given by the reference sound path correction model;

[0096] Step S43: Calculate the difference between the echo signal amplitude of the corresponding region in the target image data and the background noise amplitude of the adjacent region;

[0097] Step S44: Based on the weighted combination of the matching degree and the difference degree, generate the confidence score of the echo signal as an inherent structure echo signal.

[0098] Step S45: Echo signals with confidence scores higher than a preset threshold are identified as inherent structure echo signals.

[0099] Specifically, filtering out all the inherent structure echo signals identified from the target image data involves locating the region in the target image data that is determined to be an inherent structure echo signal, and setting the signal amplitude within that region to a value consistent with the image background.

[0100] The specific implementation of step S40 is as follows. After completing the fine division of the weld area and determining the weld toe position of each zone, the core step is executed: identifying and filtering out the inherent echoes generated by the weld toe geometry, thereby highlighting the true defect signals. This process begins with a preliminary calibration stage, namely, establishing a reference acoustic path correction model based on a standard sample. The standard sample here is a pipe section that has been confirmed to be defect-free and has the same specifications and weld morphology as the pipe to be inspected. The same complete scanning and data processing procedure as steps S10 and S20 is performed on this standard sample to obtain high-quality ultrasonic images of its weld toe area. By statistically analyzing the data from multiple standard samples and repeated tests, a theoretical acoustic path database of the echo signals in the image is established for the weld toes of the three different zones: the original weld area, the transition zone, and the grinding zone. This acoustic path corresponds to the depth or time coordinate of the ultrasonic wave propagation path in the image, thus forming a set of reference acoustic path correction models for each zone. This model accurately defines the theoretical position of the echoes of various weld toe structures and their acceptable normal fluctuation range under ideal defect-free conditions.

[0101] Upon entering the evaluation stage of the actual welded pipe to be inspected, this invention extracts and analyzes all significant echo signals with energy exceeding the basic noise threshold within each specific weld toe region identified in the target image data. For each echo signal peak point to be analyzed, this invention performs a two-dimensional quantitative evaluation.

[0102] The first step is to calculate the sound path matching degree. For the original weld area, the weld transition area, and the weld toe area of ​​the weld grinding area, the absolute error between the actual sound path and the theoretical sound path is calculated respectively.

[0103] The absolute error is input into a preset error function to calculate the matching degree;

[0104] The error function for the weld transition region is configured to output a higher matching degree value for the same absolute error input than the error functions for the original weld region and the weld grinding region.

[0105] Specifically, the actual sound path coordinates of the peak point are read and compared with the theoretical sound path value of the corresponding weld toe in the same zone in the reference sound path correction model. The matching degree is obtained by calculating the absolute value of the difference between the two and mapping it to a value between 0 and 1 using a preset error function, where 1 represents a perfect match. In this embodiment, a Gaussian error function is used, so that the smaller the absolute error, the higher the matching degree score.

[0106] The second step is amplitude difference calculation. This involves calculating the difference between the peak amplitude of the echo signal and the average amplitude of the background noise in its neighboring area. In practice, a ring-shaped region or a rectangular neighborhood that does not contain other significant signals is selected, centered on the signal peak point. The average grayscale value of all pixels in this region is calculated as the background noise level. Then, the ratio of the signal peak amplitude to the background noise level is calculated as a quantitative indicator of the difference. A larger ratio generally indicates a more prominent signal.

[0107] The calculated matching and difference values ​​are weighted linearly to generate a comprehensive confidence score, which quantifies the likelihood that the echo signal belongs to an inherent structure echo. The weighting coefficients can be set based on the analysis of a large amount of experimental data. In this embodiment, the path matching degree is assigned a weight of 0.6, and the amplitude difference degree is assigned a weight of 0.4 to emphasize the decisive role of positional features in the discrimination. A higher confidence score indicates that the signal more closely matches the echo characteristics generated by a fixed weld toe geometry.

[0108] This invention predefines a confidence threshold for judgment, which can be determined through training and validation on standard sample data, for example, set to 0.7. The confidence score of the echo signal for each weld toe region is compared one by one with this threshold. All echo signals with scores higher than this threshold are ultimately judged as inherent structure echo signals.

[0109] After identifying and marking all weld toe region signals, the final filtering operation is performed. This precisely locates each pixel region covered by a signal identified as an inherent structural echo within the original target image data matrix. All pixel amplitudes (grayscale values) within these specific regions are uniformly set to a value consistent with the overall background statistical mean of the image, or, for simplification, directly set to zero. This operation is equivalent to precisely erasing these fixed, non-defective bright spots or strip-like responses from the image. After this filtering process, weak defect signals that might have been obscured or confused by strong structural echoes are clearly revealed. The final output is a residual image containing only potential defect signals, random noise, and other unstructured indicators, thus providing a purified and targeted analysis object for subsequent automatic defect assessment steps.

[0110] Step S50: Set an evaluation threshold according to the preset defect acceptance standard; based on the comparison result between the signal features in the residual image data and the evaluation threshold, make a qualified or unqualified judgment on the transverse defects of the welded pipe end weld.

[0111] See Figure 4 The evaluation is based on a comparison between the signal features in the residual image data and the evaluation threshold, including:

[0112] Step S51: Apply adaptive threshold segmentation to the residual image data to identify all potential defective signal connected components;

[0113] Step S52: For each defect signal connected region, extract its geometric features and amplitude features. The geometric features include at least the projection length along the transverse direction of the weld and the projection height along the depth direction of the ultrasonic beam. The amplitude features include at least the maximum wave amplitude value in the region.

[0114] Step S53: Compare the geometric features or amplitude features of each defect signal connected component extracted with a preset evaluation threshold; if the geometric features or amplitude features do not exceed their corresponding thresholds, they are judged as qualified, otherwise they are judged as unqualified.

[0115] The specific implementation of step S50 is as follows: Set an evaluation threshold based on a preset defect acceptance standard. This standard is usually a technical specification that clearly defines the quantitative limits for whether a transverse defect in a weld is acceptable. For example, a standard for an oil and gas pipeline project stipulates that the indicated length of any single transverse defect must not exceed 50 mm, and its reflected amplitude must not exceed 80% of the baseline on the distance-amplitude curve (DAC). Convert this textual specification into an algorithmically executable digital threshold. The conversion process depends on calibration parameters: First, based on the spatial resolution of the image, such as the parameter that each pixel corresponds to 0.5 mm obtained through calibration, convert the length limit of 50 mm into a pixel threshold of 100 pixels. Second, based on the calibration results of the instrument and probe, convert the amplitude limit of 80% DAC into an image grayscale threshold. For example, if the grayscale value corresponding to the DAC baseline in the image is 200 (using 8-bit grayscale, ranging from 0 to 255), then 80% DAC corresponds to a grayscale threshold of 160. These converted values, including the length threshold of 100 pixels and the amplitude threshold of 160, are the preset evaluation thresholds.

[0116] Automated evaluation is performed on the residual image data obtained from the foregoing steps. First, an adaptive threshold segmentation algorithm is applied to the residual image data to separate potential defect signals from background noise. Specifically, the Otsu method can be used to automatically calculate a gray value that maximizes the between-class variance between the foreground (i.e., defect signals) and background (i.e., noise) of the image as the segmentation threshold. After binarizing the residual image using this threshold, all connected pixel regions in the image with gray values higher than this threshold are automatically identified as independent connected domains of potential defect signals.

[0117] For each identified connected domain, its quantified geometric features and amplitude features are extracted. The calculation method of geometric features is as follows: Calculate the minimum and maximum coordinates of all pixels within this connected domain on the horizontal axis of the image (corresponding to the transverse direction of the weld), and the difference between them is the projection length of this defect along the transverse direction of the weld, in pixels. Calculate the minimum and maximum coordinates of this connected domain on the vertical axis of the image (corresponding to the depth direction of the ultrasonic beam), and the difference between them is the projection height along the depth direction. The extraction of amplitude features is achieved by scanning all pixels within this connected domain and finding the maximum gray value among them as the maximum wave amplitude of this defect region.

[0118] The characteristic parameters extracted from each defect connected domain are compared one by one with the preset evaluation thresholds. The evaluation logic strictly follows the acceptance criteria in the preset standard: As long as any key characteristic parameter of the defect exceeds its corresponding allowable limit value, it is determined as unqualified. Therefore, during the automatic determination process, if it is detected that the transverse projection length of a certain connected domain is 120 pixels, which is greater than the threshold of 100 pixels, or its maximum wave amplitude is 180, which is greater than the threshold of 160, then it is immediately determined that this defect is unqualified, and further determined that the weld area at the pipe end of this welded pipe is unqualified.

[0119] Only when the projection lengths of all detected connected domains do not exceed 100 pixels and all maximum wave amplitudes do not exceed 160, is it finally determined that this weld area is qualified. The evaluation conclusion, together with the associated defect characteristic data and image position information, is automatically recorded and a structured inspection report is generated, thus completing the entire automated evaluation process.

[0120] Although the various steps are described in the above sequential order in the above embodiments, those skilled in the art can understand that for the purpose of achieving the effects of this embodiment, different steps do not have to be executed in such an order, and they can be executed simultaneously (in parallel) or in a reversed order, and these simple changes are all within the protection scope of the present invention.

[0121] An automatic evaluation system for transverse ultrasonic phased array of the weld at the pipe end of a welded pipe according to the second embodiment of the present invention is based on an automatic evaluation method for transverse ultrasonic phased array of the weld at the pipe end of a welded pipe. This system includes:

[0122] The data acquisition module is configured to perform ultrasonic phased array scanning on the weld seam within a set range at the end of the welded pipe to acquire ultrasonic phased array spectrum data of the weld seam in the transverse direction.

[0123] The image processing module is configured to select target image data containing the weld and its adjacent area from the acquired ultrasonic phased array atlas data; and to preprocess the target image data to suppress noise.

[0124] A partition recognition module is configured to recognize multiple partitions with different surface states distributed along the longitudinal direction of the weld in the target image data. The partitions include at least the original weld area that has not been ground, the weld transition area between the original weld and the ground area, and the ground weld area. For each partition, the module distinguishes between the weld body area and the weld toe area of ​​the weld edge.

[0125] The structural echo processing module is configured to, for each determined weld toe region, identify the inherent structural echo signal generated by the welded pipe geometry from the target image data based on the difference characteristics of sound path and amplitude; and filter out all identified inherent structural echo signals from the target image data to obtain residual image data.

[0126] The defect assessment module is configured to set an assessment threshold based on a preset defect acceptance standard; and to assess whether the transverse defects of the welded pipe end weld are qualified or unqualified based on the comparison result between the signal features in the residual image data and the assessment threshold.

[0127] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process and related explanations of the methods described above can be found in the corresponding processes in the foregoing system embodiments, and will not be repeated here.

[0128] It should be noted that the above-described embodiment of the automatic ultrasonic phased array evaluation system for transverse weld seams at pipe ends is merely an example illustrating the division of the functional modules. In practical applications, the functions can be assigned to different functional modules as needed, i.e., the modules or steps in the embodiments of the present invention can be further decomposed or combined. For example, the modules in the above embodiments can be merged into one module, or further divided into multiple sub-modules to complete all or part of the functions described above. The names of the modules and steps involved in the embodiments of the present invention are merely for distinguishing the various modules or steps and are not considered as an improper limitation of the present invention.

[0129] A device according to a third embodiment of the present invention includes:

[0130] At least one processor;

[0131] and a memory communicatively connected to at least one of the processors;

[0132] The memory stores instructions that can be executed by the processor to implement the above-described method for automatic evaluation of transverse ultrasonic phased array weld seams at the pipe end of a welded pipe.

[0133] A computer-readable storage medium according to a fourth embodiment of the present invention stores computer instructions, which are executed by the computer to implement the above-described method for automatic evaluation of transverse ultrasonic phased array weld seams at the pipe end of a welded pipe.

[0134] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process and related descriptions of the storage device and processing device described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0135] The terms “first”, “second”, etc., are used to distinguish similar objects, not to describe or indicate a specific order or sequence.

[0136] The term "comprising" or any other similar term is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus / device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent in such process, method, article, or apparatus / device.

[0137] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. An automatic evaluation method for transverse ultrasonic phased array weld seams at the ends of welded pipes, characterized in that, include: Ultrasonic phased array scanning was performed on the weld seam within a set range at the end of the welded pipe to obtain transverse ultrasonic phased array spectrum data of the weld seam; From the acquired ultrasonic phased array atlas data, target image data containing the weld and its adjacent area are selected; the target image data is preprocessed to suppress noise; Identify multiple partitions with different surface states distributed along the longitudinal direction of the weld in the target image data. The partitions include at least the original weld area that has not been ground, the weld transition area between the original weld and the ground area, and the ground weld area. For each partition, distinguish its weld body area from the weld toe area at the weld edge. For each identified weld toe region, based on the difference in sound path and amplitude, the inherent structural echo signal generated by the welded pipe geometry is identified from the target image data; all identified inherent structural echo signals are filtered out from the target image data to obtain residual image data. A judgment threshold is set according to the preset defect acceptance standard; based on the comparison result between the signal features in the residual image data and the judgment threshold, the transverse defects of the welded pipe end are judged as qualified or unqualified.

2. The method according to claim 1, characterized in that, Target image data containing the weld and its adjacent area is selected from the acquired ultrasonic phased array atlas data. Specifically, based on a preset position range of the weld at the pipe end, image data corresponding to the preset position range is extracted from the ultrasonic phased array atlas data and used as the target image data.

3. The method according to claim 1, characterized in that, Identifying multiple partitions distributed along the longitudinal direction of the weld in the target image data, including: In the target image data, a grayscale profile curve is extracted along a preset direction parallel to the weld seam, penetrating the region. Calculate the first derivative of the grayscale profile curve and detect the extreme points where the absolute value of the first derivative exceeds a preset first threshold. The location of the extreme points is initially marked as a potential partition boundary indicated by the abrupt change in weld surface height. Within the adjacent intervals defined by the potential partition boundary, the gray-level co-occurrence matrix of the target image data is calculated respectively, and the contrast feature value of the gray-level co-occurrence matrix is ​​extracted. Based on the numerical relationship and trend of the contrast feature value in adjacent intervals, the specific boundaries of the original weld area, the weld transition area, and the weld grinding area are confirmed and finally defined; wherein, the contrast feature value corresponding to the weld grinding area is expected to be smaller than that of the original weld area.

4. The method according to claim 1, characterized in that, For each identified weld toe region, the inherent structural echo signal is identified based on the differences in sound path and amplitude, including: A reference acoustic path correction model corresponding to the weld toe region in different zones was established based on standard samples; Calculate the degree of matching between the actual acoustic path of the echo signal in the corresponding region of the target image data and the theoretical acoustic path given by the reference acoustic path correction model; Calculate the difference between the echo signal amplitude of the corresponding region in the target image data and the background noise amplitude of the adjacent region; The echo signal is generated as a confidence score of the inherent structure echo signal based on a weighted combination of the matching degree and the difference degree. Echo signals with a confidence score higher than a preset threshold are identified as inherent structure echo signals.

5. The method according to claim 1, characterized in that, Filtering out all identified inherent structure echo signals from the target image data specifically involves locating the region in the target image data that is determined to be an inherent structure echo signal, and setting the signal amplitude within that region to a value consistent with the image background.

6. The method according to claim 1, characterized in that, Preprocessing the target image data to suppress noise includes: performing image scaling on the target image data to unify the analysis scale, and performing Gaussian blur filtering to eliminate random noise.

7. The method according to claim 1, characterized in that, The evaluation is based on a comparison between the signal features in the residual image data and the evaluation threshold, including: Adaptive threshold segmentation is applied to the residual image data to identify all potential defective signal connected components; For each defect signal connected region, its geometric features and amplitude features are extracted. The geometric features include at least the projected length along the transverse direction of the weld and the projected height along the depth direction of the ultrasonic beam. The amplitude features include at least the maximum wave amplitude value within the region. The geometric or amplitude features of each defective signal connected component are compared with a preset evaluation threshold. If the geometric or amplitude features do not exceed their corresponding threshold, they are judged as qualified; otherwise, they are judged as unqualified.

8. The method according to claim 4, characterized in that, The matching degree between the actual sound path of the echo signal in the corresponding region of the target image data and the theoretical sound path given by the reference sound path correction model is calculated as follows: For the original weld area, the weld transition area, and the weld toe area of ​​the weld grinding area, respectively, calculate the absolute error between the actual sound path and the theoretical sound path; The absolute error is input into a preset error function to calculate the matching degree; The error function for the weld transition region is configured to output a higher matching degree value for the same absolute error input than the error functions for the original weld region and the weld grinding region.

9. An automatic evaluation system for transverse ultrasonic phased array weld seams at the ends of welded pipes, based on the automatic evaluation method for transverse ultrasonic phased array weld seams at the ends of welded pipes according to any one of claims 1-8, characterized in that, The system includes: The data acquisition module is configured to perform ultrasonic phased array scanning on the weld seam within a set range at the end of the welded pipe to acquire ultrasonic phased array spectrum data of the weld seam in the transverse direction. The image processing module is configured to select target image data containing the weld and its adjacent area from the acquired ultrasonic phased array atlas data; and to preprocess the target image data to suppress noise. A partition recognition module is configured to recognize multiple partitions with different surface states distributed along the longitudinal direction of the weld in the target image data. The partitions include at least the original weld area that has not been ground, the weld transition area between the original weld and the ground area, and the ground weld area. For each partition, the module distinguishes between the weld body area and the weld toe area of ​​the weld edge. The structural echo processing module is configured to, for each determined weld toe region, identify the inherent structural echo signal generated by the welded pipe geometry from the target image data based on the difference characteristics of sound path and amplitude; and filter out all identified inherent structural echo signals from the target image data to obtain residual image data. The defect assessment module is configured to set an assessment threshold based on a preset defect acceptance standard; and to assess whether the transverse defects of the welded pipe end weld are qualified or unqualified based on the comparison result between the signal features in the residual image data and the assessment threshold.

10. A device, characterized in that, include: At least one processor; and a memory communicatively connected to at least one of the processors; The memory stores instructions that can be executed by the processor to implement the automatic evaluation method of transverse ultrasonic phased array for welded pipe end welds as described in any one of claims 1-8.

Images