Automatic Detection Method and System for Zinc Coating Thickness of Galvanized Steel Pipes

Abstract

This invention discloses an automatic detection method and system for zinc coating thickness of galvanized steel pipes, belonging to the field of coating thickness detection technology. The automatic detection system for zinc coating thickness of galvanized steel pipes includes a galvanized steel pipe partitioning module and a zinc coating thickness detection module. This invention effectively solves the systematic interference caused by uneven weld permeability and microstructure differences on magnetic induction thickness measurement results by identifying and partitioning the weld texture on the steel pipe surface, distinguishing the weld area from the substrate area at the detection level. This prevents the weld from being misjudged as abrupt changes in zinc coating thickness, thus significantly improving the authenticity and consistency of online measurement results. This invention also enhances adaptability to steel pipe vibration, runout, and surface roughness changes, reducing dependence on mechanical stability and installation accuracy.

Description

Technical Field

[0001] This invention relates to the field of coating thickness detection technology, and in particular to an automatic method and system for detecting the zinc coating thickness of galvanized steel pipes. Background Technology

[0002] Hot-dip galvanized steel pipes are widely used in fire protection, gas supply, and water supply. The uniformity of the zinc layer thickness directly affects their corrosion resistance and service life. Among existing technologies, the differential gravity method is accurate, but it is an offline average detection and cannot reflect local thickness differences. X-ray fluorescence method is expensive and has radiation protection issues. Laser measurement method is easily affected by steel pipe vibration, ellipticity, and dimensional tolerances, making it difficult to achieve stable online detection. Although the traditional magnetic induction thickness measurement method is suitable for online applications, it is sensitive to lift-off distance and vibration height, and the uneven magnetic permeability in the weld area can be easily misjudged as a sudden change in thickness, leading to distorted measurement results.

[0003] Therefore, there is an urgent need for a method for detecting the zinc layer thickness of galvanized steel pipes that is non-contact, vibration-resistant, weld interference-resistant, and suitable for online applications. Summary of the Invention

[0004] The present invention aims to provide an automatic detection method and system for zinc coating thickness of galvanized steel pipes, enabling online, continuous and automatic detection of zinc coating thickness of steel pipes, and significantly improving detection coverage and data representativeness.

[0005] An automatic method for detecting the zinc coating thickness of galvanized steel pipes includes the following steps:

[0006] A zinc coating thickness detection framework is constructed above the perimeter of the galvanized steel pipe conveying path; the galvanized steel pipe to be inspected is conveyed along the conveying path for automatic zinc coating thickness detection; the surface image of the galvanized steel pipe to be inspected is obtained based on the zinc coating thickness detection framework; the weld texture features of the steel pipe are extracted from the surface image; the steel pipe thickness measurement zones are divided based on the weld texture features; the steel pipe thickness measurement zones include special zones of the steel pipe weld and ordinary zones of the steel pipe substrate.

[0007] When the galvanized steel pipe to be inspected passes through the galvanized steel pipe conveying path, the feature vector of the galvanized steel pipe belonging to different steel pipe thickness measurement zones is obtained; based on the galvanized steel pipe feature vector, the thickness difference of the galvanized steel pipe to be inspected is retrieved to obtain the difference zinc layer thickness of different steel pipe thickness measurement zones; among which, the difference zinc layer thickness includes the base zinc layer thickness and the weld zinc layer thickness.

[0008] Based on the framework for detecting zinc coating thickness in steel pipes, the longitudinal position of the galvanized steel pipe to be tested is obtained in different steel pipe thickness measurement zones; the average zinc coating thickness of the galvanized steel pipe to be tested is detected by combining the longitudinal position of the steel pipe and the difference in zinc coating thickness.

[0009] As a preferred embodiment of the present invention, the specific steps for constructing a steel pipe zinc coating thickness detection framework include:

[0010] A circular inspection bracket for steel pipes is set up along the circumference of the galvanized steel pipe conveying path; several non-contact inspection units for steel pipes are evenly distributed along the circumference on the circular inspection bracket for steel pipes; each non-contact inspection unit for steel pipes integrates a magnetic induction sensor and a laser displacement sensor; at the same time, a steel pipe visual imaging unit and a steel pipe length measurement coding unit are set up at the front end of the circular inspection bracket for steel pipes.

[0011] By combining the pre-set annular steel pipe testing brackets, a steel pipe zinc layer thickness testing frame is obtained.

[0012] As a preferred embodiment of the present invention, the specific steps for extracting the weld texture features of a steel pipe from a surface image include:

[0013] The steel pipe surface image is acquired based on the steel pipe visual imaging unit; the horizontal and vertical gradients of each pixel in the steel pipe surface image are calculated; the image structure tensor matrix of each pixel is constructed based on the horizontal and vertical gradients; the image structure tensor matrix is ​​decomposed into eigenvalues ​​to obtain the steel pipe weld texture features; the steel pipe weld texture features include the main feature values ​​and secondary feature values ​​of the steel pipe image.

[0014] The texture coherence index of the steel pipe is calculated based on the principal and secondary feature values ​​of the steel pipe image; the texture coherence index of the steel pipe is projected along the lateral coordinate of the steel pipe surface image to obtain the texture coherence distribution curve of the steel pipe; the center phase of the steel pipe weld is identified by searching in the texture coherence distribution curve; the topological boundary of the steel pipe weld is determined according to the center phase of the steel pipe weld.

[0015] As a preferred embodiment of the present invention, the specific steps for dividing the steel pipe thickness measurement zones based on the weld texture characteristics include:

[0016] The steel pipe weld coverage sector is constructed based on the center phase and topological boundary of the steel pipe weld; the overlap rate between the sampling center line of each non-contact detection unit and the steel pipe weld coverage sector is traversed to obtain the steel pipe weld coverage rate;

[0017] If the coverage rate of the steel pipe weld is greater than the preset interference threshold, the steel pipe thickness measurement zone corresponding to the non-contact detection unit is classified as a special zone for steel pipe welds; if the coverage rate of the steel pipe weld is less than or equal to the preset interference threshold, the steel pipe thickness measurement zone corresponding to the non-contact detection unit is classified as a normal zone for steel pipe substrate.

[0018] As a preferred technical solution of the present invention, the specific steps for retrieving the thickness difference of the galvanized steel pipe to be detected based on the feature vector of the galvanized steel pipe include:

[0019] Obtain a pre-defined three-dimensional response model of a galvanized steel pipe;

[0020] For the feature vector of galvanized steel pipe in the ordinary area of ​​steel pipe substrate, the probe lift-off distance value is extracted from the feature vector of galvanized steel pipe as the X-axis coordinate, and the magnetic induction voltage response value is extracted as the Y-axis coordinate. The X-axis coordinate and Y-axis coordinate are bilinearly interpolated in the preset three-dimensional response model of galvanized steel pipe to obtain the value corresponding to the Z-axis of the surface as the base zinc layer thickness corresponding to the ordinary area of ​​steel pipe substrate.

[0021] For the feature vector of galvanized steel pipe in the special zone of steel pipe weld, the calculation is performed in steps: Step S1, assume that the feature vector of galvanized steel pipe in the special zone of steel pipe weld is the feature vector of ordinary galvanized steel pipe; obtain the ordinary probe lift-off distance value and ordinary magnetic induction voltage response value based on the feature vector of ordinary galvanized steel pipe;

[0022] Step S2: Based on the feature vector of the galvanized steel pipe in the special zone of the steel pipe weld, feature extraction is performed to obtain the special probe lift-off distance value and the special magnetic induction voltage response value; the simplified difference between the special probe lift-off distance value and the special magnetic induction voltage response value and the ordinary probe lift-off distance value and the ordinary magnetic induction voltage response value is calculated to obtain the magnetic disturbance residual.

[0023] Step S3: Obtain steel pipe inspection time series data based on laser displacement sensor; perform differential processing on steel pipe inspection time series data to obtain weld surface curvature characteristics; invert the magnetic disturbance residual, weld surface curvature characteristics, ordinary probe lift-off distance value and ordinary magnetic induction voltage response value in the preset galvanized steel pipe three-dimensional response model to obtain the weld zinc layer thickness corresponding to the special area of ​​steel pipe weld.

[0024] As a preferred technical solution of the present invention, the specific steps for detecting the average zinc layer thickness of the galvanized steel pipe to be tested, in conjunction with the longitudinal position of the steel pipe and the differential zinc layer thickness, include:

[0025] The steel pipe is divided into several continuous longitudinal virtual slices according to its longitudinal position; the thickness of the base zinc layer and the thickness of the weld zinc layer in different steel pipe thickness measurement zones are mapped to the corresponding longitudinal virtual slices according to the steel pipe inspection time series data to obtain the steel pipe slice thickness data;

[0026] For each longitudinal virtual slice, calculate the average value of the base zinc layer thickness and the average value of the weld zinc layer thickness in that longitudinal virtual slice; if the base zinc layer thickness and weld zinc layer thickness are not present in a certain longitudinal virtual slice, then use the steel pipe slice thickness data of the adjacent longitudinal virtual slice to complete the data.

[0027] The physical curvature ratio of the galvanized steel pipe to be inspected is determined based on the topological boundary of the steel pipe weld; the equivalent average thickness of the zinc layer of each longitudinal virtual slice is obtained by weighted fusion of the average thickness of the base zinc layer and the average thickness of the weld zinc layer.

[0028] The longitudinal virtual slices at both ends of the galvanized steel pipe to be tested are removed. The equivalent average thickness of the zinc layer of all remaining longitudinal virtual slices is longitudinally integrated and averaged to obtain the final average zinc layer thickness of the galvanized steel pipe to be tested.

[0029] An automatic zinc coating thickness detection system for galvanized steel pipes includes:

[0030] The galvanized steel pipe partitioning module includes a steel pipe partitioning unit. This unit is used to construct a zinc layer thickness detection framework around the perimeter of the galvanized steel pipe conveying path. The module then transports the galvanized steel pipe to be inspected along the conveying path for automatic zinc layer thickness detection. Based on the zinc layer thickness detection framework, it acquires an image of the galvanized steel pipe surface. It extracts weld texture features from the surface image and divides the pipe thickness measurement into partitions based on these features. Each partition includes a special weld area and a general substrate area.

[0031] The zinc coating thickness detection module includes a zone detection unit and an average detection unit. The zone detection unit is used to acquire the feature vectors of the galvanized steel pipes belonging to different steel pipe thickness measurement zones when the galvanized steel pipe to be inspected passes through the galvanized steel pipe conveying path. Based on the galvanized steel pipe feature vectors, the thickness difference of the galvanized steel pipe to be inspected is retrieved to obtain the differential zinc coating thickness of different steel pipe thickness measurement zones. The differential zinc coating thickness includes the base zinc coating thickness and the weld zinc coating thickness. The average detection unit is used to acquire the longitudinal position of the galvanized steel pipe to be inspected in different steel pipe thickness measurement zones based on the steel pipe zinc coating thickness detection framework. The average zinc coating thickness of the galvanized steel pipe to be inspected is detected by combining the longitudinal position of the steel pipe and the differential zinc coating thickness.

[0032] The present invention has the following advantages:

[0033] 1. This invention identifies and partitions the weld texture on the steel pipe surface, distinguishing the weld area from the substrate area at the detection level. This effectively solves the systematic interference caused by uneven weld magnetic permeability and microstructure differences on magnetic induction thickness measurement results, preventing the weld from being misjudged as a sudden change in zinc layer thickness, thus significantly improving the authenticity and consistency of online measurement results. By constructing a three-dimensional response model that integrates the relationship between probe lift-off distance, magnetic induction voltage response, and zinc layer thickness, and combining interpolation inversion and residual compensation algorithms, this invention can still achieve micron-level thickness resolution under non-contact lift-off conditions of several millimeters. This significantly enhances the system's adaptability to steel pipe vibration, runout, and surface roughness changes, and reduces dependence on mechanical stability and installation accuracy.

[0034] 2. This invention separates and compensates for magnetic and geometric anomalies in the weld area, enabling the zinc layer thickness in the weld area to be output under the same evaluation system as the base area. This avoids the information loss caused by the simple rejection of the weld area in traditional methods. Through a comprehensive processing strategy of longitudinal virtual slicing, circumferential arc weighted fusion, and end rejection, a physically meaningful equivalent average zinc layer thickness is obtained. This ensures that the final result reflects the true galvanization level of the entire steel pipe and has good resistance to local anomalies and noise, meeting the dual requirements of quality assessment and process control. Attached Figure Description

[0035] Figure 1 This is a schematic diagram of the automatic zinc coating thickness detection system for galvanized steel pipes used in an embodiment of the present invention. Detailed Implementation

[0036] To enable those skilled in the art to better understand the technical solutions of this invention, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of this invention.

[0037] Example 1: An automatic detection method for zinc coating thickness of galvanized steel pipes, comprising the following steps:

[0038] A zinc coating thickness detection framework is constructed above the perimeter of the galvanized steel pipe conveying path; the galvanized steel pipe to be inspected is conveyed along the conveying path for automatic zinc coating thickness detection; the surface image of the galvanized steel pipe to be inspected is obtained based on the zinc coating thickness detection framework; the weld texture features of the steel pipe are extracted from the surface image; the steel pipe thickness measurement zones are divided based on the weld texture features; the steel pipe thickness measurement zones include special zones of the steel pipe weld and ordinary zones of the steel pipe substrate.

[0039] The specific steps for constructing a framework for detecting the zinc coating thickness of steel pipes include:

[0040] A circular inspection bracket for steel pipes is set up along the circumference of the galvanized steel pipe conveying path; several non-contact inspection units for steel pipes are evenly distributed along the circumference on the circular inspection bracket for steel pipes; each non-contact inspection unit for steel pipes integrates a magnetic induction sensor and a laser displacement sensor; at the same time, a steel pipe visual imaging unit and a steel pipe length measurement coding unit are set up at the front end of the circular inspection bracket for steel pipes.

[0041] The steel pipe zinc layer thickness detection frame is obtained by combining the pre-set steel pipe ring detection bracket.

[0042] In practical applications, a ring-shaped inspection bracket is set along the outer circumference of the galvanized steel pipe on the conveying path. The ring-shaped inspection bracket forms a stable ring structure around the steel pipe to support various inspection devices and ensure their relative position to the steel pipe. The steel pipe is always at the center of this ring structure as it moves continuously in the production line, thus providing the basic conditions for realizing synchronous inspection at multiple positions around the circumference.

[0043] Several non-contact steel pipe testing units are evenly arranged along the circumference on the steel pipe ring testing bracket. During operation, the non-contact steel pipe testing units maintain a certain gap with the surface of the steel pipe and do not make physical contact, thereby avoiding wear or impact on the testing device caused by steel pipe vibration, surface roughness or weld excess, while ensuring the continuity and stability of the testing process.

[0044] Each non-contact steel pipe inspection unit integrates a magnetic induction sensor and a laser displacement sensor. The magnetic induction sensor is used to acquire the magnetic response signal formed by the steel pipe substrate and the galvanized layer, and this magnetic response corresponds to the thickness of the zinc layer. The laser displacement sensor is used to measure the actual distance change between the sensor and the outer surface of the steel pipe in real time, providing an accurate displacement reference for distance compensation and thickness inversion of the magnetic induction signal, thereby reducing the impact of steel pipe runout and vibration on measurement accuracy.

[0045] A steel pipe visual imaging unit is set at the front end of the steel pipe ring inspection bracket. The steel pipe visual imaging unit is used to acquire continuous image information of the outer surface of the steel pipe. By analyzing the surface texture features, the weld area and non-weld area are identified, providing a basis for the area division and anomaly identification of subsequent inspection data. At the same time, a steel pipe length measurement and coding unit is set at the front end. The steel pipe length measurement and coding unit works synchronously with the steel pipe conveying mechanism and outputs a position signal corresponding to the movement distance of the steel pipe, which is used to establish the correspondence between the inspection data and the longitudinal position of the steel pipe.

[0046] By assembling and coordinating the steel pipe ring detection bracket, non-contact detection unit, visual imaging unit, and length measurement coding unit, a steel pipe zinc coating thickness detection frame is formed. This enables the collection of zinc coating thickness information at multiple points around the circumference and in a continuous longitudinal direction when the steel pipe passes through the steel pipe zinc coating thickness detection frame, providing a complete hardware foundation for online automatic detection and subsequent data processing.

[0047] The specific steps for extracting weld texture features from steel pipe surface images include:

[0048] The steel pipe surface image is acquired based on the steel pipe visual imaging unit; the horizontal and vertical gradients of each pixel in the steel pipe surface image are calculated; the image structure tensor matrix of each pixel is constructed based on the horizontal and vertical gradients; the image structure tensor matrix is ​​decomposed into eigenvalues ​​to obtain the steel pipe weld texture features; the steel pipe weld texture features include the main feature values ​​and secondary feature values ​​of the steel pipe image.

[0049] When the galvanized steel pipes to be inspected are conveyed on the galvanized steel pipe conveying path for automatic zinc layer thickness detection, the galvanized steel pipes to be inspected move forward continuously or intermittently under the drive of the production line conveying mechanism and pass through the effective detection area of ​​the steel pipe zinc layer thickness detection frame. The galvanized steel pipe conveying path refers to the movement trajectory and posture constraint range of the steel pipe from the inlet to the outlet on the production line. The detection frame forms a fixed detection benchmark on this path, so that the subsequently acquired images, positions and thickness signals can correspond in the same spatial coordinates and time series, meeting the requirements of online, continuous and automated detection.

[0050] When acquiring images of the galvanized steel pipe surface based on the steel pipe zinc coating thickness detection framework, the steel pipe vision imaging unit performs image acquisition on the outer surface of the steel pipe. The steel pipe surface image refers to two-dimensional image data containing information on the brightness and texture changes of the outer surface of the steel pipe, which can reflect visual features such as welds, zinc flowers, scratches, oxide spots, and surface roughness differences. The imaging unit usually acquires continuous image frames through stable illumination and a fixed field of view, and keeps them consistent with the movement direction of the steel pipe, so as to establish a traceable correspondence between the image information and the longitudinal position of the steel pipe.

[0051] When extracting the texture features of steel pipe welds from images of the steel pipe surface, the welds are distinguished from the substrate surface by utilizing the differences in texture directionality and abrupt changes in grayscale in the weld area during imaging. The texture features of steel pipe welds refer to the feature quantities that can characterize the texture structure and directional consistency of the weld area. Unlike simple grayscale threshold segmentation, it emphasizes that the weld has a stronger directional structure and anisotropy in local areas, thus maintaining good recognition stability even when the zinc flower is complex, the lighting changes, or the surface is reflective.

[0052] The texture coherence index of the steel pipe is calculated based on the principal and secondary feature values ​​of the steel pipe image; the texture coherence index of the steel pipe is projected along the lateral coordinate of the steel pipe surface image to obtain the texture coherence distribution curve of the steel pipe; the center phase of the steel pipe weld is identified by searching within the texture coherence distribution curve; the topological boundary of the steel pipe weld is determined based on the center phase of the steel pipe weld.

[0053] When dividing the steel pipe thickness measurement zones based on the texture characteristics of the steel pipe weld, the identified weld area is calibrated separately as a special interference area, while the remaining areas are used as regular measurement areas for thickness data acquisition and fusion. The steel pipe thickness measurement zone refers to the functional division of the measurement position within the circumference and longitudinal range of the steel pipe. Its purpose is to avoid misjudgment of the magnetic induction thickness measurement signal caused by changes in magnetic permeability, differences in microstructure, or geometric protrusions in the weld area, while ensuring the representativeness and consistency of the thickness assessment of the base area.

[0054] When the steel pipe thickness measurement partitioning includes a special zone for the steel pipe weld and a general zone for the steel pipe matrix, the special zone for the steel pipe weld refers to a topological range around the weld centerline. Within this topological range, due to the heat-affected zone of welding and the treatment of scraping, the magnetic parameters and surface morphology may be significantly different, requiring strategies such as shielding, elimination, or separate modeling and compensation. The general zone for the steel pipe matrix refers to a stable region far from the influence of the weld, whose magnetic and geometric characteristics are closer to the state of a uniform matrix, making it suitable for conventional thickness inversion and statistical evaluation. By partitioning the steel pipe thickness measurement, zinc layer thickness detection can significantly reduce the systematic interference of the weld on the thickness results while ensuring online efficiency.

[0055] When acquiring images of the steel pipe surface using a steel pipe vision imaging unit, the unit continuously captures images of the outer surface of the steel pipe before it enters the detection area or at the initial stage of entry, forming a digital image that can be used for calculation. The image typically contains an array of pixels, with each pixel corresponding to the brightness information of a tiny region on the steel pipe surface. Through reasonable exposure and lighting control, the texture differences between the weld area and the substrate area are made analyzable in the image, providing stable input for subsequent gradient and tensor calculations. When calculating the horizontal and vertical gradients of each pixel in the steel pipe surface image, the image differentiation concept is used to calculate the grayscale change rate in the horizontal and vertical directions respectively. The horizontal gradient represents the strength and direction of grayscale change in the horizontal neighborhood of a pixel, while the vertical gradient represents the strength and direction of grayscale change in the vertical neighborhood of a pixel. These two types of gradients together characterize local edges, texture direction, and the degree of structural abrupt changes, forming the fundamental step in extracting structural information from the original brightness information.

[0056] When constructing the image structure tensor matrix for each pixel based on the horizontal and vertical gradients, the local gradients are statistically fused in the neighborhood to form a second-order matrix describing the directionality and anisotropy of the texture. The image structure tensor matrix is ​​a mathematical object used to characterize the local image structure. It integrates the magnitude and direction distribution of the gradient and can distinguish flat areas without obvious direction, striped texture areas with a single main direction, and corner areas with intersecting structures, thereby providing stronger noise resistance and resistance to illumination disturbances for robust recognition of weld textures.

[0057] When performing eigenvalue decomposition on the image structure tensor matrix to obtain the texture features of the steel pipe weld, the principal eigenvalues ​​and secondary eigenvalues ​​of the matrix are obtained through decomposition, and these are used to quantify the directional consistency and intensity difference of the local structure. Eigenvalue decomposition refers to the process of decomposing the matrix into feature quantities that reflect its principal direction and energy distribution. The principal eigenvalue represents the structural intensity along the main texture direction, and the secondary eigenvalue represents the structural intensity in the direction orthogonal to it. When the weld area exhibits a strong directional strip texture, the principal eigenvalues ​​and secondary eigenvalues ​​will show a more significant difference, thus forming a set of texture features that can be used to identify the weld.

[0058] When calculating the texture coherence index of steel pipes based on principal and secondary eigenvalues, the relative difference between the two is converted into a normalized index to describe whether the local texture has a consistent principal direction. The texture coherence index is an indicator that measures the degree of concentration of texture direction in a local area. High coherence means that the texture exhibits a clear single-directional structure, while low coherence means that the texture direction is dispersed or approximately isotropic. Since welds often appear as continuous bands with prominent directionality, this index can more clearly distinguish welds from random zinc flower textures and noise textures.

[0059] When projecting the coherence index of the steel pipe texture along the horizontal coordinate of the steel pipe surface image to obtain the coherence distribution curve of the steel pipe texture, the coherence index at the same horizontal position is accumulated or averaged in the vertical range to form a one-dimensional curve that varies with the horizontal coordinate. The horizontal coordinate refers to the coordinate axis in the image that is perpendicular to the steel pipe axis or the direction of steel pipe movement. The purpose of this projection operation is to transform the two-dimensional texture discrimination problem into a one-dimensional peak localization problem, so that the weld seam appears as a stable high coherence response band in the curve, thereby improving the robustness and computational efficiency of center localization.

[0060] When searching for and identifying the weld center phase in the coherence distribution curve of steel pipe texture, the center position of the weld in the horizontal coordinate is determined by detecting the peak position or significant protrusion area of ​​the distribution curve. The weld center phase refers to the position parameter of the weld centerline in the horizontal coordinate of the image, which is used to characterize the degree of lateral offset of the weld relative to the imaging field of view. By finding the center point with the maximum response or that meets the threshold and shape constraints in the curve, the stable positioning of the weld center can be maintained even under slight rotation of the steel pipe, illumination fluctuations, or complex surface textures.

[0061] When determining the topological boundary of a steel pipe weld based on the center phase, the center phase is used as a reference to expand to both sides. Combined with the decreasing trend of the coherence index, the edge transition width, or the preset weld influence width, the left and right boundary positions of the weld area in the image are determined. The weld topological boundary refers to the boundary of the connected range occupied by the special area of ​​the weld in the image space. It is used to assign subsequent thickness measurement points to the special area of ​​the weld or the ordinary area of ​​the substrate, so as to avoid or compensate for weld interference during magnetic induction thickness measurement, and finally realize a consistent, well-defined, and repeatable online zinc layer thickness measurement process.

[0062] When the galvanized steel pipe to be inspected passes through the galvanized steel pipe conveying path, the feature vector of the galvanized steel pipe belonging to different steel pipe thickness measurement zones is obtained; based on the galvanized steel pipe feature vector, the thickness difference of the galvanized steel pipe to be inspected is retrieved to obtain the difference zinc layer thickness of different steel pipe thickness measurement zones; among which, the difference zinc layer thickness includes the base zinc layer thickness and the weld zinc layer thickness.

[0063] The specific steps for dividing steel pipe thickness measurement zones based on the texture characteristics of steel pipe welds include:

[0064] The steel pipe weld coverage sector is constructed based on the center phase and topological boundary of the steel pipe weld. The center phase of the weld is understood as the center position parameter of the weld in the circumferential direction of the steel pipe, and the topological boundary of the weld is understood as the left and right boundary positions of the weld's influence range in the circumferential direction. Then, with the center of the steel pipe as the vertex and the circumferential angle corresponding to the weld boundary as two boundary rays, a fan-shaped area with a clear angle range is formed on the outer surface of the steel pipe. This fan-shaped area is the steel pipe weld coverage sector. The steel pipe weld coverage sector represents the spatial range on the circumference of the steel pipe that may be affected by the non-uniform magnetic interference of the weld. Its function is to convert the original weld position relationship in the image coordinates into a circumferential angle range that the detection frame can understand, thereby providing a unified geometric basis for subsequent judgment on whether each detection unit has "scanned the weld".

[0065] The overlap rate between the sampling centerline of each non-contact inspection unit and the weld coverage sector of the steel pipe is traversed to obtain the weld coverage rate. The sampling centerline is understood as the representative measurement direction or center position of the measurement trajectory of a certain non-contact inspection unit during measurement, which reflects the measurement direction and measurement area center of the unit on the circumference of the steel pipe. The geometric relationship between the sampling centerline of each inspection unit and the weld coverage sector is calculated to obtain the degree of overlap between the two in the circumferential angle. The degree of overlap is characterized by the overlap rate, and thus the weld coverage rate of the steel pipe is obtained. The weld coverage rate of the steel pipe refers to the quantitative result of what proportion of the effective measurement area of ​​the inspection unit falls into the weld interference sector. It can transform the potential influence of the weld on the measurement from qualitative judgment to a calculable and thresholdable criterion, which facilitates consistent zoning rules under different steel pipe specifications and different unit installation angles.

[0066] If the coverage rate of the steel pipe weld is greater than the preset interference threshold, the steel pipe thickness measurement zone corresponding to the non-contact detection unit is classified as a special zone for steel pipe welds; if the coverage rate of the steel pipe weld is less than or equal to the preset interference threshold, the steel pipe thickness measurement zone corresponding to the non-contact detection unit is classified as a normal zone for steel pipe substrate.

[0067] When the coverage of the steel pipe weld exceeds the preset interference threshold, the steel pipe thickness measurement zone corresponding to the non-contact detection unit is designated as a special zone for the steel pipe weld. This is because a coverage exceeding the threshold means that a significant proportion of the measurement area of ​​the detection unit is affected by changes in the magnetic permeability, microstructure differences, or abnormal surface morphology of the weld area. The response of the magnetic induction sensor is more likely to reflect material magnetic differences rather than actual thickness changes. Therefore, strategies such as shielding, rejection, or separate modeling compensation need to be adopted during the data processing stage. The preset interference threshold is a discrimination boundary set based on experimental calibration or process experience. It is used to define the extent to which weld interference must be treated as a special object to ensure the reliability and traceability of thickness calculation.

[0068] When the weld coverage of the steel pipe is less than or equal to the preset interference threshold, the steel pipe thickness measurement zone corresponding to the non-contact detection unit is divided into the ordinary zone of the steel pipe substrate. This is because the coverage not exceeding the threshold means that the measurement area of ​​the detection unit is mainly located in the stable region of the substrate, and the influence of the weld on the magnetic response is weak or negligible. The magnetic induction signal is more representative of the zinc layer thickness change itself, so conventional distance compensation and thickness inversion models can be used for calculation and participation in circumferential statistics. The setting of the ordinary zone of the steel pipe substrate enables the system to maximize the use of effective measurement data under the premise of ensuring controllable weld interference, improve the coverage and measurement efficiency of online detection, and maintain the consistency of the partitioning results under different working conditions.

[0069] The specific steps for retrieving thickness differences in galvanized steel pipes based on their feature vectors include:

[0070] Obtain a pre-defined three-dimensional response model of the galvanized steel pipe. First, call or load a pre-established three-dimensional response relationship dataset. This model is used to describe the voltage response law of the magnetic induction sensor under different probe lift-off distances and different zinc layer thicknesses. The three-dimensional response model can be understood as a curved surface or lookup table data. Its horizontal dimension corresponds to the actual distance between the probe and the surface of the steel pipe, that is, the lift-off distance. The vertical dimension corresponds to the magnetic induction voltage response value. The height dimension of the curved surface corresponds to the zinc layer thickness or a parameter equivalent to the thickness. This model is usually established through standard sample calibration or experimental data acquisition and is matched with the specific sensor model, installation structure, excitation parameters and steel pipe substrate material state, so as to ensure that the zinc layer thickness can be reliably deduced from the measured distance and voltage.

[0071] For the feature vector of galvanized steel pipe in the ordinary area of ​​the steel pipe substrate, the probe lift-off distance value is extracted as the X-axis coordinate, and the magnetic induction voltage response value is extracted as the Y-axis coordinate. The X-axis and Y-axis coordinates are then bilinearly interpolated in a pre-defined galvanized steel pipe three-dimensional response model to obtain the value corresponding to the Z-axis of the surface, which is taken as the zinc layer thickness of the substrate in the ordinary area of ​​the steel pipe substrate. First, the probe lift-off distance value is extracted as the horizontal axis coordinate, and then the magnetic induction voltage response value is extracted as the vertical axis coordinate. The galvanized steel pipe feature vector refers to a feature vector formed by multiple source sensors at the same sampling time or spatial location. The set of characterization parameters includes at least the lift-off distance information obtained from laser displacement and the voltage response information output by the magnetic induction sensor. Since the preset three-dimensional response model is often stored in the form of a discrete grid, the actual measured horizontal and vertical coordinates may fall between grid points. Therefore, bilinear interpolation addressing needs to be performed on adjacent grid points, that is, interpolation estimation is performed simultaneously in both the horizontal and vertical directions to obtain the value corresponding to the surface height direction. This value is the thickness of the zinc layer of the steel pipe substrate in the ordinary area. In this way, the thickness of the ordinary area of ​​the steel pipe substrate can be stably inverted directly by using the combination relationship between distance and voltage.

[0072] For the feature vector of the galvanized steel pipe in the special zone of the steel pipe weld, a step-by-step calculation is performed: Step S1, assuming the feature vector of the galvanized steel pipe in the special zone of the steel pipe weld is the same as the feature vector of the ordinary galvanized steel pipe; obtain the ordinary probe lift-off distance value and the ordinary magnetic induction voltage response value based on the feature vector of the ordinary galvanized steel pipe; firstly, the feature vector of the special zone of the steel pipe weld is temporarily regarded as the feature vector of the ordinary galvanized steel pipe in the calculation logic. The purpose of this step is to construct a baseline state without considering weld magnetic anomalies and surface morphology anomalies; then, according to the same feature extraction rules as the ordinary zone, the ordinary probe lift-off distance value and the ordinary magnetic induction voltage response value are obtained from this baseline state, which are used to characterize the distance and voltage combination that the system should present if there is no weld disturbance at this location. This baseline value is not equivalent to the actual measured value, but provides a reference for subsequent identification of additional deviations caused by the weld.

[0073] Step S2: Feature extraction is performed based on the feature vector of the galvanized steel pipe in the special area of ​​the steel pipe weld to obtain the special probe lift-off distance value and the special magnetic induction voltage response value. The simplified difference between the special probe lift-off distance value and the special magnetic induction voltage response value and the ordinary probe lift-off distance value and the ordinary magnetic induction voltage response value is calculated to obtain the magnetic disturbance residual. The special probe lift-off distance value and the special magnetic induction voltage response value are extracted to reflect the actual measurement state of the probe under the influence of the weld. Then, the difference between the special lift-off distance value and the ordinary lift-off distance value, and the difference between the special voltage response value and the ordinary voltage response value are merged and simplified to obtain the magnetic disturbance residual. The magnetic disturbance residual can be understood as the set of additional deviations introduced by the weld area relative to the reference state. It comprehensively reflects the degree of disturbance caused by factors such as changes in magnetic permeability, differences in microstructure, weld reinforcement height, and local geometric abrupt changes in the weld area to the magnetic induction measurement link. It provides a key input for the next step of separating the weld disturbance from the thickness signal and compensating for it.

[0074] Step S3: Acquire time-series data of steel pipe inspection based on laser displacement sensor; perform differential processing on the time-series data of steel pipe inspection to obtain the surface curvature characteristics of weld; invert the magnetic disturbance residual, weld surface curvature characteristics, ordinary probe lift-off distance value, and ordinary magnetic induction voltage response value in the preset three-dimensional response model of galvanized steel pipe to obtain the zinc layer thickness of weld corresponding to the special area of ​​steel pipe weld; continuously collect data sequence of displacement changing with time during the process of steel pipe passing through the inspection area, so that each time point corresponds to a distance or displacement measurement value of a surface position; the time-series data of steel pipe inspection refers to continuous measurement data arranged in chronological order, which can reflect the geometric change trend of weld area caused by excess height or surface transition; the time-series data... Differential processing refers to calculating the rate of change of the displacement sequence to extract the rate of change and the steepness of the surface morphology, and further obtaining the surface curvature characteristics of the weld. The curvature characteristics are used to characterize the degree of bending or protrusion of the surface and are strongly correlated with the geometric shape of the weld. Finally, the magnetic disturbance residual, the surface curvature characteristics of the weld, and the ordinary lift-off distance value and ordinary magnetic induction voltage response value obtained in step S1 are input into the preset three-dimensional response model of the galvanized steel pipe for inversion calculation. Inversion refers to solving for the thickness result that best fits the model constraint under the condition of knowing multiple observations and disturbance characteristics, so as to obtain the weld zinc layer thickness corresponding to the special area of ​​the steel pipe weld, so that the thickness of the weld area is no longer misjudged by magnetic anomalies, and the consistent thickness output can be compared with the ordinary area of ​​the substrate.

[0075] Based on the framework for detecting zinc coating thickness in steel pipes, the longitudinal position of the galvanized steel pipe to be tested is obtained in different steel pipe thickness measurement zones; the average zinc coating thickness of the galvanized steel pipe to be tested is detected by combining the longitudinal position of the steel pipe and the difference in zinc coating thickness.

[0076] The specific steps for determining the average zinc coating thickness of the galvanized steel pipe under test, based on the longitudinal position of the steel pipe and the differential zinc coating thickness, include:

[0077] The steel pipe is divided into several continuous longitudinal virtual slices based on its longitudinal position. The thickness of the base zinc layer and the weld zinc layer in different steel pipe thickness measurement zones are mapped to the corresponding longitudinal virtual slices based on the steel pipe inspection time series data to obtain the steel pipe slice thickness data. The position signal output by the steel pipe length measurement encoding unit is used to establish a correspondence with the inspection time series, mapping the time of each thickness sampling to the specific position of the steel pipe along the axial direction. The longitudinal position of the steel pipe refers to the coordinate along the length of the steel pipe, used to identify which segment of the steel pipe the thickness data comes from. Different steel pipe thickness measurement zones refer to the inspection areas divided in the circumferential direction according to the special area of ​​the weld and the ordinary area of ​​the base. Therefore, the result of this step is to assign a traceable axial position label to each thickness data from different zones, so that subsequent calculations can not only compare the thickness differences between different zones, but also describe the variation law of thickness along the length of the steel pipe.

[0078] The entire length range of the galvanized steel pipe to be inspected is discretized at equal intervals according to a preset slice length, forming a set of interval units arranged continuously along the axial direction. Each interval unit is called a longitudinal virtual slice. The longitudinal virtual slice is a segmentation method in the sense of data processing. It does not change the physical structure of the steel pipe, but is used to classify the thickness data obtained by continuous sampling according to position, so as to summarize the circumferential multi-point measurement results within the same axial interval. Then, the base zinc layer thickness and weld zinc layer thickness obtained in different thickness measurement zones are mapped to their respective longitudinal virtual slices according to the longitudinal position relationship corresponding to the steel pipe inspection time series data, so as to obtain the steel pipe slice thickness data in each slice. The steel pipe slice thickness data refers to the set of thicknesses falling within the same slice interval and coming from multiple inspection zones or multiple samplings, providing a data basis for subsequent statistical averaging and completion.

[0079] For each longitudinal virtual slice, calculate the average value of the base zinc layer thickness and the average value of the weld zinc layer thickness in that longitudinal virtual slice; if the base zinc layer thickness and weld zinc layer thickness are not present in a certain longitudinal virtual slice, then use the steel pipe slice thickness data of the adjacent longitudinal virtual slice to complete the data.

[0080] Within the slice, the data belonging to the ordinary area of ​​the substrate are statistically averaged, and the data belonging to the special area of ​​the weld are also statistically averaged to obtain two representative thicknesses within the same axial position segment. The average value is used to suppress single-point noise and instantaneous fluctuations, making the thickness characterization of the slice layer more stable. When there is no substrate zinc layer thickness or weld zinc layer thickness data in a certain longitudinal virtual slice, the reasons usually include sampling loss, image occlusion causing the partition to be unusable, or local data being removed. In this case, the steel pipe slice thickness data of the adjacent longitudinal virtual slice is used to complete the data. That is, the statistical results of the previous and next slices are used for interpolation or approximation to ensure the continuity of the longitudinal thickness sequence and avoid the deviation caused by gaps in the subsequent weighted fusion and integral averaging.

[0081] The physical curvature ratio of the galvanized steel pipe to be inspected is determined based on the topological boundary of the weld seam. The equivalent average zinc layer thickness for each longitudinal virtual slice is obtained by weighted fusion of the average thickness of the base zinc layer and the average thickness of the weld zinc layer. The angular range corresponding to the special weld area on the circumference of the steel pipe is converted into a circumferential proportion to describe the proportion of the weld area on the entire circumferential surface. The physical curvature ratio of the steel pipe refers to the ratio of the circumferential angle of the weld-covered sector to the angle of the entire circle, reflecting the actual area weight of the weld-affected area on the circumference. Subsequently, the average base zinc layer thickness and the average weld zinc layer thickness obtained in each longitudinal virtual slice are weighted and fused. Weighted fusion refers to assigning weights to the two types of thicknesses according to the circumferential proportion and summing them to obtain the equivalent average zinc layer thickness of the slice. The equivalent average zinc layer thickness represents the equivalent average thickness of the entire circumference surface after considering the proportions of the weld area and the base area on the circumference, ensuring that the slice thickness reflects both the dominant area of ​​the base and the actual contribution of the weld area.

[0082] The longitudinal virtual slices at both ends of the galvanized steel pipe to be tested are removed. The equivalent average thickness of the zinc layer of all remaining longitudinal virtual slices is longitudinally integrated and averaged to obtain the final average zinc layer thickness of the galvanized steel pipe to be tested.

[0083] Several slices near the beginning and end of the steel pipe are excluded from the calculation because the ends of the steel pipe are often affected by process factors such as cutting, clamping, plating, and zinc dripping, resulting in greater thickness fluctuations and poor representativeness. Directly including them in the averaging would amplify the impact of end anomalies on the evaluation of the entire steel pipe. The equivalent average zinc layer thickness of the remaining longitudinal virtual slices is longitudinally integrated and averaged. This means summing the slice thickness sequence along the length of the steel pipe and normalizing it according to the effective length. Essentially, this is equivalent to averaging the equivalent thickness over the effective length of the entire steel pipe in an area sense, thus obtaining the final average zinc layer thickness of the galvanized steel pipe to be tested. This makes the result reflect the overall level of the entire length and has stability against local anomalies and sampling fluctuations.

[0084] In this embodiment, the overall computational framework for automatic detection of zinc coating thickness in steel pipes is based on a combination of sensor physical calibration model and data-driven interpolation inversion algorithm. The core foundation is a pre-set three-dimensional response model of galvanized steel pipe. This three-dimensional response model uses probe lift-off distance and magnetic induction voltage response as independent variables and zinc coating thickness as dependent variable. It is represented by discrete grid surface or dense lookup table data, and a calibration surface is established by collecting magnetic responses at different lift-off distances using standard samples or samples with known thickness. The implementation method is to pre-store the surface data and corresponding coordinate axis scales in the system. During online detection, the lift-off distance and voltage response obtained from each sampling are mapped onto the surface, and the thickness result is obtained through interpolation addressing, thereby unifying the magnetic induction signal and geometric gap change into the same computable model.

[0085] In the machine vision section, the computational model used for weld seam recognition and partitioning is based on image gradient field and structural tensor analysis algorithms. Specifically, the gray-level change rate of the steel pipe surface image is calculated in the horizontal and vertical directions to form the gradient information of each pixel. Then, the gradients are statistically fused in the local neighborhood to construct the image structure tensor matrix. The matrix is ​​decomposed into principal and secondary eigenvalues, and the relative difference between the two is used to calculate the texture coherence index. The coherence index is used to quantify the directional consistency of local texture. The weld seam area usually exhibits a stronger directional structure, thus showing a significant peak on the coherence distribution curve. In implementation, the coherence index is projected along the horizontal coordinate of the image to form a one-dimensional distribution curve. Then, the weld seam center phase is located through peak search and morphological constraints, and the weld seam topological boundary is determined by truncation based on the coherence attenuation range or threshold, thereby completing the robust recognition of the weld seam position and the determination of repeatable boundaries.

[0086] In the steel pipe thickness measurement zoning section, a geometric mapping and overlap rate discrimination model is used. Specifically, the weld center phase and weld topological boundary are transformed from image coordinates to steel pipe circumferential angle coordinates, and a weld coverage sector is constructed based on the steel pipe center. This sector represents the circumferential range where the weld may cause magnetic non-uniform interference. Then, the geometric overlap between the sampling center line of each non-contact detection unit and the sector is traversed to calculate the weld coverage rate as an interference quantification index, and it is compared with a preset interference threshold to complete the binary classification judgment. When the coverage rate is higher than the threshold, the corresponding measurement zone is marked as a weld special zone to enter the compensation inversion process. When the coverage rate is not higher than the threshold, it is marked as a matrix ordinary zone to enter the conventional interpolation thickness calculation process, thereby transforming the weld interference problem into a calculable and configurable zoning rule.

[0087] In the thickness calculation of the ordinary region of the substrate, a bivariate interpolation addressing algorithm based on a three-dimensional response model is used to achieve thickness inversion. Specifically, the probe lift-off distance is extracted from the feature vector of the galvanized steel pipe as the lateral coordinate, and the magnetic induction voltage response is extracted as the longitudinal coordinate. Bilinear interpolation is then performed on four adjacent grid points corresponding to the three-dimensional response surface to obtain the thickness estimate in the height direction of the surface. The advantages of this method are low computational cost, real-time operation, and the ability to compensate for lift-off distance fluctuations in the thickness result using a model. This makes the thickness estimate of the ordinary region less sensitive to transmitted vibrations and vibrations, meeting the engineering requirements for online rapid measurement.

[0088] In the thickness calculation of the special zone of the weld, an algorithm combining residual separation and constraint inversion is used to compensate for the magnetic disturbance of the weld. Specifically, a normal reference state is first constructed to obtain the normal lift-off distance value and the normal magnetic induction voltage response value as a reference when there is no weld disturbance. Then, the special lift-off distance value and the special magnetic induction voltage response value are extracted from the actual feature vector of the special zone of the weld, and the difference between them and the normal reference value is used to obtain the magnetic disturbance residual, which is used to characterize the additional magnetic response offset caused by the weld. At the same time, the detection time series data of the laser displacement sensor is used for differentiation to obtain the surface curvature feature of the weld. The curvature feature describes the geometric changes caused by the weld reinforcement or transition zone. Finally, the magnetic disturbance residual, the surface curvature feature of the weld, and the normal reference value are input into the inversion solution process of the three-dimensional response model. The most consistent thickness result with the observation is obtained under the model constraint through the minimum error criterion, so that the thickness output of the weld area reflects the real zinc layer thickness as much as possible rather than the artifact caused by magnetic inhomogeneity.

[0089] In the longitudinal fusion and average thickness calculation of the steel pipe, a spatiotemporal mapping model based on coded positioning and a slice statistical integration algorithm are adopted. Specifically, the length measurement coding unit is used to establish the correspondence between the detection time and the longitudinal position of the steel pipe. The steel pipe is discretized into several continuous longitudinal virtual slices according to the longitudinal position. Then, the thickness results of the ordinary area of ​​the substrate and the special area of ​​the weld are mapped to their respective slices according to the time series position to form slice thickness data. The average thickness of the substrate and the average thickness of the weld are calculated for each slice. When a certain type of data is missing, the data of the adjacent slices are interpolated to complete the sequence to ensure the continuity of the sequence. Then, the proportion of the physical curvature of the weld area on the circumference is determined according to the weld topological boundary. The two types of average thickness are weighted and fused using this proportion to obtain the equivalent average thickness of the slice. Finally, after removing the slices at the beginning and end, the equivalent average thickness of the remaining slices is longitudinally integrated and averaged to obtain the final average zinc layer thickness of the entire steel pipe, thereby realizing the unified evaluation of the circumferential zoning results and the longitudinal positioning results.

[0090] When data units differ, this embodiment employs a unified dimensional system and unit conversion calibration. The principle is that all model inputs and outputs must be converted to the same consistent unit system before computation, and the unit definitions are solidified during the model building phase. Specifically, this involves establishing clear physical quantity mapping relationships for each sensor channel; for example, laser displacement is expressed in millimeters or micrometers, magnetic induction voltage in volts, encoder position in millimeters, and image coordinates in pixels. Before online computation, calibration coefficients are used to convert pixel coordinates to circumferential angles or actual lengths, coded pulses to longitudinal displacements, and to perform proportional conversions between micrometers and millimeters, and between millivolts and volts, ensuring that the input units for the horizontal and vertical axes of the 3D response model are completely consistent with those used in calibration modeling. For steps requiring inversion optimization or error comparison, data with different dimensions are normalized to ensure that the numerical amplitudes of each input quantity are within a similar range, preventing any single quantity from dominating the solution process due to excessively large numerical magnitudes. This ensures that interpolation addressing, residual calculation, and inversion solving operate reliably under the premise of consistent units and numerical stability.

[0091] Example 2: Automatic zinc coating thickness detection system for galvanized steel pipes (see [link]). Figure 1 As shown, it includes:

[0092] The galvanized steel pipe partitioning module includes a steel pipe partitioning unit. This unit is used to construct a zinc layer thickness detection framework around the perimeter of the galvanized steel pipe conveying path. The module then transports the galvanized steel pipe to be inspected along the conveying path for automatic zinc layer thickness detection. Based on the zinc layer thickness detection framework, it acquires an image of the galvanized steel pipe surface. It extracts weld texture features from the surface image and divides the pipe thickness measurement into partitions based on these features. Each partition includes a special weld area and a general substrate area.

[0093] The zinc coating thickness detection module includes a zone detection unit and an average detection unit. The zone detection unit is used to acquire the feature vectors of the galvanized steel pipes belonging to different steel pipe thickness measurement zones when the galvanized steel pipe to be inspected passes through the galvanized steel pipe conveying path. Based on the galvanized steel pipe feature vectors, the thickness difference of the galvanized steel pipe to be inspected is retrieved to obtain the differential zinc coating thickness of different steel pipe thickness measurement zones. The differential zinc coating thickness includes the base zinc coating thickness and the weld zinc coating thickness. The average detection unit is used to acquire the longitudinal position of the galvanized steel pipe to be inspected in different steel pipe thickness measurement zones based on the steel pipe zinc coating thickness detection framework. The average zinc coating thickness of the galvanized steel pipe to be inspected is detected by combining the longitudinal position of the steel pipe and the differential zinc coating thickness.

[0094] It should be understood that those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims. Parts not described in detail in this specification are prior art known to those skilled in the art.

Claims

1. An automatic detection method for zinc coating thickness of galvanized steel pipes, characterized in that, Includes the following steps: A zinc coating thickness detection framework is constructed above the perimeter of the galvanized steel pipe conveying path; the galvanized steel pipe to be inspected is conveyed along the conveying path for automatic zinc coating thickness detection; the surface image of the galvanized steel pipe to be inspected is obtained based on the zinc coating thickness detection framework; the weld texture features of the steel pipe are extracted from the surface image; the steel pipe thickness measurement zones are divided based on the weld texture features; the steel pipe thickness measurement zones include special zones of the steel pipe weld and ordinary zones of the steel pipe substrate. When the galvanized steel pipe to be inspected passes through the galvanized steel pipe conveying path, the feature vector of the galvanized steel pipe belonging to different steel pipe thickness measurement zones is obtained; based on the galvanized steel pipe feature vector, the thickness difference of the galvanized steel pipe to be inspected is retrieved to obtain the difference zinc layer thickness of different steel pipe thickness measurement zones; among which, the difference zinc layer thickness includes the base zinc layer thickness and the weld zinc layer thickness. Based on the framework for detecting zinc coating thickness in steel pipes, the longitudinal position of the galvanized steel pipe to be tested is obtained in different steel pipe thickness measurement zones; the average zinc coating thickness of the galvanized steel pipe to be tested is detected by combining the longitudinal position of the steel pipe and the difference in zinc coating thickness.

2. The automatic detection method for zinc layer thickness of galvanized steel pipe according to claim 1, characterized in that, The specific steps for constructing a framework for detecting the zinc coating thickness of steel pipes include: A circular inspection bracket for steel pipes is set up along the circumference of the galvanized steel pipe conveying path; several non-contact inspection units for steel pipes are evenly distributed along the circumference on the circular inspection bracket for steel pipes; each non-contact inspection unit for steel pipes integrates a magnetic induction sensor and a laser displacement sensor; at the same time, a steel pipe visual imaging unit and a steel pipe length measurement coding unit are set up at the front end of the circular inspection bracket for steel pipes. By combining the pre-set annular steel pipe testing brackets, a steel pipe zinc layer thickness testing frame is obtained.

3. The automatic detection method for zinc layer thickness of galvanized steel pipe according to claim 2, characterized in that, The specific steps for extracting weld texture features from steel pipe surface images include: The steel pipe surface image is acquired based on the steel pipe visual imaging unit; the horizontal and vertical gradients of each pixel in the steel pipe surface image are calculated; the image structure tensor matrix of each pixel is constructed based on the horizontal and vertical gradients; the image structure tensor matrix is ​​decomposed into eigenvalues ​​to obtain the steel pipe weld texture features; the steel pipe weld texture features include the main feature values ​​and secondary feature values ​​of the steel pipe image. The texture coherence index of the steel pipe is calculated based on the principal and secondary feature values ​​of the steel pipe image; the texture coherence index of the steel pipe is projected along the lateral coordinate of the steel pipe surface image to obtain the texture coherence distribution curve of the steel pipe; the center phase of the steel pipe weld is identified by searching in the texture coherence distribution curve; the topological boundary of the steel pipe weld is determined according to the center phase of the steel pipe weld.

4. The automatic detection method for zinc layer thickness of galvanized steel pipe according to claim 3, characterized in that, The specific steps for dividing steel pipe thickness measurement zones based on the texture characteristics of steel pipe welds include: The steel pipe weld coverage sector is constructed based on the center phase and topological boundary of the steel pipe weld; the overlap rate between the sampling center line of each non-contact detection unit and the steel pipe weld coverage sector is traversed to obtain the steel pipe weld coverage rate; If the coverage rate of the steel pipe weld is greater than the preset interference threshold, the steel pipe thickness measurement zone corresponding to the non-contact detection unit is classified as a special zone for steel pipe welds; if the coverage rate of the steel pipe weld is less than or equal to the preset interference threshold, the steel pipe thickness measurement zone corresponding to the non-contact detection unit is classified as a normal zone for steel pipe substrate.

5. The automatic detection method for zinc layer thickness of galvanized steel pipe according to claim 4, characterized in that, The specific steps for retrieving thickness differences in galvanized steel pipes based on their feature vectors include: Obtain a pre-defined three-dimensional response model of a galvanized steel pipe; For the feature vector of galvanized steel pipe in the ordinary area of ​​steel pipe substrate, the probe lift-off distance value is extracted from the feature vector of galvanized steel pipe as the X-axis coordinate, and the magnetic induction voltage response value is extracted as the Y-axis coordinate. The X-axis coordinate and Y-axis coordinate are bilinearly interpolated in the preset three-dimensional response model of galvanized steel pipe to obtain the value corresponding to the Z-axis of the surface as the base zinc layer thickness corresponding to the ordinary area of ​​steel pipe substrate. For the feature vector of galvanized steel pipe in the special zone of steel pipe weld, the calculation is carried out in steps: Step S1, assume that the feature vector of galvanized steel pipe in the special zone of steel pipe weld is the feature vector of ordinary galvanized steel pipe; obtain the ordinary probe lift-off distance value and ordinary magnetic induction voltage response value based on the feature vector of ordinary galvanized steel pipe; Step S2: Based on the feature vector of the galvanized steel pipe in the special zone of the steel pipe weld, feature extraction is performed to obtain the special probe lift-off distance value and the special magnetic induction voltage response value; the difference between the special probe lift-off distance value and the special magnetic induction voltage response value and the ordinary probe lift-off distance value and the ordinary magnetic induction voltage response value is calculated to obtain the magnetic disturbance residual. Step S3: Obtain steel pipe inspection time series data based on laser displacement sensor; perform differential processing on steel pipe inspection time series data to obtain weld surface curvature characteristics; invert the magnetic disturbance residual, weld surface curvature characteristics, ordinary probe lift-off distance value and ordinary magnetic induction voltage response value in the preset galvanized steel pipe three-dimensional response model to obtain the weld zinc layer thickness corresponding to the special area of ​​steel pipe weld.

6. The automatic detection method for zinc coating thickness of galvanized steel pipes according to claim 5, characterized in that, The specific steps for determining the average zinc coating thickness of the galvanized steel pipe under test, based on the longitudinal position of the steel pipe and the differential zinc coating thickness, include: The steel pipe is divided into several continuous longitudinal virtual slices according to its longitudinal position; the base zinc layer thickness and weld zinc layer thickness in different steel pipe thickness measurement zones are mapped to the corresponding longitudinal virtual slices according to the steel pipe inspection time series data to obtain the steel pipe slice thickness data; For each longitudinal virtual slice, calculate the average value of the base zinc layer thickness and the average value of the weld zinc layer thickness in that longitudinal virtual slice; if the base zinc layer thickness and weld zinc layer thickness are not present in a certain longitudinal virtual slice, then use the steel pipe slice thickness data of the adjacent longitudinal virtual slice to complete the data. The physical curvature ratio of the galvanized steel pipe to be inspected is determined based on the topological boundary of the steel pipe weld; the equivalent average thickness of the zinc layer of each longitudinal virtual slice is obtained by weighted fusion of the average thickness of the base zinc layer and the average thickness of the weld zinc layer. The longitudinal virtual slices at both ends of the galvanized steel pipe to be tested are removed. The equivalent average thickness of the zinc layer of all remaining longitudinal virtual slices is longitudinally integrated and averaged to obtain the final average zinc layer thickness of the galvanized steel pipe to be tested.

7. An automatic detection system for zinc coating thickness of galvanized steel pipes, characterized in that, The system employs the automatic zinc coating thickness detection method for galvanized steel pipes according to any one of claims 1-6, including: The galvanized steel pipe partitioning module includes a steel pipe partitioning unit. This unit is used to construct a zinc layer thickness detection framework around the perimeter of the galvanized steel pipe conveying path. The module then transports the galvanized steel pipe to be inspected along the conveying path for automatic zinc layer thickness detection. Based on the zinc layer thickness detection framework, it acquires an image of the galvanized steel pipe surface. It extracts weld texture features from the surface image and divides the pipe thickness measurement into partitions based on these features. Each partition includes a special weld area and a general substrate area. The zinc coating thickness detection module includes a zone detection unit and an average detection unit. The zone detection unit is used to acquire the feature vectors of the galvanized steel pipes belonging to different steel pipe thickness measurement zones when the galvanized steel pipe to be inspected passes through the galvanized steel pipe conveying path. Based on the galvanized steel pipe feature vectors, the thickness difference of the galvanized steel pipe to be inspected is retrieved to obtain the differential zinc coating thickness of different steel pipe thickness measurement zones. The differential zinc coating thickness includes the base zinc coating thickness and the weld zinc coating thickness. The average detection unit is used to acquire the longitudinal position of the galvanized steel pipe to be inspected in different steel pipe thickness measurement zones based on the steel pipe zinc coating thickness detection framework. The average zinc coating thickness of the galvanized steel pipe to be inspected is detected by combining the longitudinal position of the steel pipe and the differential zinc coating thickness.

Images