A method and system for detecting the thickness of a zinc spray layer

Abstract

The application discloses a kind of zinc spraying layer thickness detection method and system, wherein method includes: obtaining the surface optical image of target zinc spraying component, and the edge detection of the surface optical image;Based on the edge detection result, the target zinc spraying component is dynamically sampled, and the region to be detected is obtained;Based on double-wavelength laser, the compound ranging detection is carried out to the region to be detected, and the first thickness data is obtained, and the three-dimensional topography detection is carried out to the region to be detected, and the surface roughness parameter and surface undulation size parameter are obtained;At least the surface roughness parameter and surface undulation size parameter are input into the pre-trained thickness correction model to correct the deviation of the first thickness data, and the second thickness data is obtained, and the second thickness data is corrected based on oxidation reaction information, and the third thickness data is obtained;The third thickness data is regionally adapted and processed, and the zinc spraying layer thickness detection result is obtained.The method of the application improves the accuracy of zinc spraying layer thickness detection.

Description

Technical Field

[0001] This invention relates to the field of metal surface treatment technology, and in particular to a method and system for detecting the thickness of a zinc spray layer. Background Technology

[0002] As an important anti-corrosion protective layer for steel components and other substrates, the uniformity and integrity of the zinc spray layer directly affect the service life and anti-corrosion performance of the components. Therefore, accurate detection of the zinc spray layer thickness is a key link in quality control and safety assessment.

[0003] Traditional zinc spray coating thickness testing uses a uniform sampling mode, for example, at 1 dm 2 The method of taking the average of 10 measurements within the reference plane is simple to operate, but it can only sample at a fixed density and cannot adapt to the differences in the surface structure of components. Specifically, when inspecting the corners of bridge steel components and the weld transition sections of iron tower angle steel, these areas often have uneven coating thickness due to the limited spraying angle. However, uniform sampling points may be concentrated in relatively flat areas, just avoiding the thin coating areas at the corners, making it difficult to reflect the actual protective ability of the coating on the core parts of the component. Summary of the Invention

[0004] This invention provides a method for detecting the thickness of a zinc sprayed layer, thereby addressing the technical problem of how to improve existing methods for detecting the thickness of a zinc sprayed layer and achieving the effect of improving the accuracy of zinc sprayed layer thickness detection.

[0005] To address the aforementioned technical problems, this invention provides a method for detecting the thickness of a zinc sprayed layer, comprising:

[0006] In response to a dynamic sampling signal, an optical image of the surface of the target zinc-sprayed component is acquired, and edge detection is performed on the surface optical image to obtain the edge detection result;

[0007] Based on the edge detection results, the target zinc-sprayed component is dynamically sampled to obtain the area to be detected. The dynamic sampling is designed to identify the degree of change of the surface information of the target zinc-sprayed component in the edge detection results, and to determine the corresponding sampling density based on the degree of change.

[0008] In response to the thickness measurement signal, the detection area is subjected to composite ranging detection based on dual-wavelength laser to obtain the first thickness data, and the three-dimensional morphology of the detection area is detected to obtain the surface roughness parameters and surface undulation size parameters.

[0009] The surface roughness parameter and the surface undulation size parameter are input into the pre-trained thickness correction model, and the first thickness data is corrected based on the thickness deviation value output by the thickness correction model to obtain the second thickness data.

[0010] identify oxidation reaction information of the target zinc spraying component, and correct the second thickness data based on the oxidation reaction information to obtain third thickness data;

[0011] perform regional adaptation processing on the third thickness data to obtain a zinc spraying layer thickness detection result of the target zinc spraying component.

[0012] As one of the preferred solutions, edge detection is performed on the surface optical image to obtain an edge detection result, including:

[0013] The surface optical image is sequentially subjected to grayscale conversion and smoothing denoising processing to obtain grayscale image data;

[0014] Gradient calculation is performed on the grayscale image data based on an improved Sobel operator, and valid edge pixels of the gradient calculation result are screened based on a double-threshold method, wherein the valid edge pixels include strong edge pixels and weak edge pixels;

[0015] Morphological optimization is performed on the strong edge pixels and the weak edge pixels to obtain a continuous edge contour;

[0016] Boundary region division is performed on the surface optical image based on the curvature value of the continuous edge contour to obtain the edge detection result.

[0017] As one of the preferred solutions, the edge detection result includes a fluctuation-prone region, a planar stable region, and a pore-dense region;

[0018] Boundary region division is performed on the surface optical image based on the curvature value of the continuous edge contour to obtain the edge detection result, including:

[0019] The curvature value of each pixel point on the continuous edge contour is calculated, wherein the curvature value is obtained based on fitting a quadratic curve with the spatial coordinates of adjacent pixel points;

[0020] The region corresponding to the contour pixel point with a curvature value greater than a first curvature threshold value is marked as a fluctuation-prone region, and the region corresponding to the contour pixel point with a curvature value less than a second curvature threshold value is marked as a potential planar region;

[0021] Texture feature extraction is performed on the grayscale image data of the potential planar region, a pore-dense region of the potential planar region is screened based on the obtained geometric features, and the part of the potential planar region that is not marked as a pore-dense region is divided into a planar stable region.

[0022] As one of the preferred solutions, the target zinc spraying component is dynamically sampled based on the edge detection result to obtain a to-be-detected region, including:

[0023] According to the boundary region division in the edge detection result, a fluctuation-prone area, a planar stable area and a pore-dense area of the target zinc spraying component are determined;

[0024] Based on the differentiated sampling density, the fluctuation-prone area, the planar stable area and the pore-dense area are sampled respectively to obtain a set of sampling points;

[0025] The set of sampling points is subjected to coordinate verification to obtain a to-be-detected area covering the key area of the target zinc spraying component, wherein the coordinate verification is designed to eliminate invalid sampling points beyond the boundary of the target zinc spraying component.

[0026] As one of the preferred solutions, the dual-wavelength laser includes a zinc characteristic wavelength and a base material characteristic wavelength of the target zinc spraying component;

[0027] Based on the dual-wavelength laser, a composite ranging detection is performed on the to-be-detected area to obtain first thickness data, including:

[0028] The dual-wavelength laser ranging module is positioned at each sampling point of the to-be-detected area, and the dual-wavelength laser is controlled to be synchronously emitted to the surface of the sampling point to obtain the time of flight of the dual-wavelength laser according to the laser receiver;

[0029] Based on the time of flight of the dual-wavelength laser, a first distance value from the surface of the zinc spraying layer to the dual-wavelength laser ranging module and a second distance value from the surface of the base material to the dual-wavelength laser ranging module are calculated respectively;

[0030] Based on the first distance value and the second distance value, a dual-wavelength distance difference value of each sampling point is calculated to obtain the first thickness data of the to-be-detected area.

[0031] As one of the preferred solutions, the to-be-detected area is subjected to three-dimensional topography detection to obtain surface roughness parameters and surface fluctuation size parameters, including:

[0032] The to-be-detected area is subjected to regional three-dimensional topography data acquisition to obtain original point cloud data of the to-be-detected area;

[0033] The original point cloud data is subjected to noise filtering, baseline correction and data smoothing processing in sequence to obtain three-dimensional profile data;

[0034] Based on the three-dimensional profile data, surface roughness parameters of the to-be-detected area are calculated, and the three-dimensional profile data is subjected to frequency spectrum analysis according to Fourier transform to obtain surface fluctuation size parameters of the to-be-detected area.

[0035] As one of the preferred solutions, the at least surface roughness parameter and the surface fluctuation size parameter are input into a pre-trained thickness correction model, and the first thickness data is corrected based on the thickness deviation value output by the thickness correction model to obtain second thickness data, including:

[0036] Obtaining roughness parameter sets, surface fluctuation size parameter sets, first thickness measurement data sets and corresponding real thickness data sets of a plurality of zinc spraying component samples, and calculating a deviation data set between the first thickness measurement data set and the real thickness data set;

[0037] Constructing a thickness correction model based on an improved random forest regression algorithm, wherein the thickness correction model takes the roughness parameter set and the surface fluctuation size parameter set as input variables, and takes the deviation data set as output variable;

[0038] Inputting the surface roughness parameter and the surface fluctuation size parameter of the region to be detected into the thickness correction model to obtain the corresponding target deviation amount, and correcting the first thickness data based on the target deviation amount to obtain second thickness data.

[0039] As one of the preferred solutions, the oxidation reaction information of the target zinc spraying component is identified, and the second thickness data is corrected based on the oxidation reaction information to obtain third thickness data, including:

[0040] Obtaining feature spectrum data of the region to be detected based on laser-induced breakdown spectroscopy detection method, and extracting zinc element characteristic peak intensity value and oxygen element characteristic peak intensity value in the feature spectrum data;

[0041] Constructing a calibration curve of zinc-oxygen element characteristic peak intensity ratio and oxide layer thickness, and calculating the oxide layer thickness value of each sampling point in the region to be detected according to the zinc element characteristic peak intensity value, the oxygen element characteristic peak intensity value and the calibration curve;

[0042] Correcting the thickness value of each sampling point in the second thickness data based on the oxide layer thickness value to obtain third thickness data after stripping the oxide layer.

[0043] As one of the preferred solutions, the third thickness data is regionally adapted to obtain the zinc spraying layer thickness detection result of the target zinc spraying component, including:

[0044] According to the region division of the edge detection result, the easy fluctuation area, the plane stable area and the pore dense area are determined, and each region is assigned a corresponding weight coefficient, wherein the weight coefficient is set based on the influence degree of the region on the overall performance of the zinc spraying layer;

[0045] Count the number of sampling points of the third thickness data in each region, calculate the arithmetic mean of the third thickness data of all sampling points in each region, and obtain a regional average thickness value;

[0046] Multiply the regional average thickness value of each region by the weight coefficient of the region to obtain a regional weighted thickness value, and give the target zinc spraying member a zinc spraying layer thickness value obtained by the regional weighted thickness value.

[0047] Another aspect of the present application provides a zinc spraying layer thickness detection system, comprising:

[0048] The acquisition module is configured to acquire a surface optical image of the target zinc spraying member in response to a dynamic sampling signal, and perform edge detection on the surface optical image to obtain an edge detection result.

[0049] The sampling module is configured to perform dynamic sampling on the target zinc spraying member based on the edge detection result to obtain a region to be detected, wherein the dynamic sampling is designed to identify the degree of change of the surface information of the target zinc spraying member in the edge detection result, and determine the corresponding sampling density based on the degree of change.

[0050] The measurement module is configured to perform compound ranging detection on the region to be detected based on dual-wavelength laser in response to a thickness measurement signal to obtain first thickness data, and perform three-dimensional topography detection on the region to be detected to obtain a surface roughness parameter and a surface undulation size parameter.

[0051] The first correction module is configured to input at least the roughness parameter and the surface undulation size parameter into a pre-trained thickness correction model, and perform bias correction on the first thickness data based on a thickness bias value output by the thickness correction model to obtain second thickness data.

[0052] The second correction module is configured to identify oxidation reaction information of the target zinc spraying member, and correct the second thickness data based on the oxidation reaction information to obtain third thickness data.

[0053] The output module is configured to perform regional adaptation processing on the third thickness data to obtain a zinc spraying layer thickness detection result of the target zinc spraying member.

[0054] Compared with the prior art, the present application has at least one of the following advantages:

[0055] 1) The dynamic sampling strategy based on edge detection of the present application can intelligently identify feature regions such as fluctuation-prone regions and flat stable regions of the surface of the member, automatically increase the sampling points at key parts such as corners and welds where the coating is prone to uneven thickness, avoid quality misjudgment caused by missing key regions due to traditional uniform sampling, and optimize the sampling density in flat regions to reduce redundant detection.

[0056] 2) At the same time, the present application combines the roughness and surface relief parameters obtained by three-dimensional topography detection, and corrects the deviation of the laser ranging data through the pre-trained thickness correction model, effectively eliminates the interference of surface zinc grain accumulation, micro concave and other micro topography on thickness measurement, controls the single measurement error in a smaller range, and solves the data distortion problem caused by surface topography interference in the traditional method. BRIEF DESCRIPTION OF DRAWINGS

[0057] Figure 1 is a flowchart of the zinc spraying layer thickness detection method in one of the embodiments of the present application;

[0058] Figure 2 is a component surface roughness / surface relief schematic diagram in one of the embodiments of the present application;

[0059] Figure 3 is a schematic diagram of the zinc spraying layer thickness detection system in one of the embodiments of the present application;

[0060] REFERENCE NUMERALS:

[0061] Among them, 11, acquisition module; 12, sampling module; 13, measurement module; 14, first correction module; 15, second correction module; 16, output module. DETAILED DESCRIPTION

[0062] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. The purpose of providing these embodiments is to make the disclosure of the present application more thorough and comprehensive. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of the present application.

[0063] In the description of the present application, the terms "first", "second", "third" and the like are only used for description purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first", "second", "third" and the like can explicitly or implicitly include one or more features. In the description of the present application, unless otherwise specified, the meaning of "multiple" is two or more.

[0064] In the description of the present application, it is necessary to explain that, unless otherwise defined, all technical and scientific terms used in the present application have the same meanings as those commonly understood by the person skilled in the art. The terms used in the specification of the present application are only for the purpose of describing the specific embodiments and are not intended to limit the present application. For the person skilled in the art, the specific meanings of the above terms in the present application can be understood according to the specific circumstances.

[0065] An embodiment of the present application provides a zinc spraying layer thickness detection method, and specifically, please refer to Figure 1 , Figure 1 A flowchart of the zinc spraying layer thickness detection method in one embodiment of the present application is shown, which comprises steps S1-S6:

[0066] S1: In response to a dynamic sampling signal, a surface optical image of a target zinc spraying component is acquired, and edge detection is performed on the surface optical image to obtain an edge detection result;

[0067] The embodiment extracts the structural features of the surface of the component accurately through a series of image processing techniques, and provides accurate area division basis for subsequent dynamic sampling and thickness detection.

[0068] Preferably, in one embodiment of the present application, the edge detection is performed on the surface optical image to obtain the edge detection result, comprising:

[0069] The surface optical image is sequentially subjected to gray scale conversion and smoothing denoising processing to obtain gray scale image data;

[0070] The gray scale image data is subjected to gradient calculation based on an improved Sobel operator, and valid edge pixels of the gradient calculation result are screened based on a double-threshold method, wherein the valid edge pixels comprise strong edge pixels and weak edge pixels;

[0071] The strong edge pixels and the weak edge pixels are subjected to morphological optimization to obtain a continuous edge contour;

[0072] The surface optical image is subjected to boundary region division based on the curvature value of the continuous edge contour to obtain the edge detection result.

[0073] Specifically, the first step of processing the surface optical image is gray scale conversion. The color optical image contains color information of three channels of red, green and blue (RGB), and the data amount is large and the color difference may interfere with edge detection, such as the reflection of the zinc spraying layer surface and the color difference which may be misjudged as an edge. The color image is converted into a single-channel gray scale image through gray scale conversion, which not only simplifies the data dimension, but also retains the light and dark contrast features of the pixels.

[0074] Further, the gray scale converted image is subjected to smoothing and denoising processing. The surface of the zinc spraying member can have noise or fine impurities during shooting, which can cause abnormal fluctuations in the gray scale value and be easily misjudged as a "pseudo-edge". In this embodiment, the noise is suppressed by a filtering algorithm. The Gaussian filter smoothes the high-frequency noise by weighted average of the pixels in the surrounding area, while retaining the overall trend of the edge. The median filter takes the median value after sorting the gray scale values in the pixel neighborhood, and has better effect on suppressing salt and pepper noise, and is suitable for the granular impurities that can exist on the surface of the zinc spraying layer. After preprocessing, the obtained gray scale image data not only removes redundant information, but also retains the gray scale change characteristics of the true edge.

[0075] It can be understood that the essence of the edge is the area where the gray scale value in the image changes sharply, and the purpose of gradient calculation is to quantify the "intensity" and "direction" of the change. The traditional Sobel operator calculates the gradient amplitude and direction of each pixel by convolving two 3x3 convolution kernels with the gray scale image, but has limitations in the zinc spraying layer detection scene: the surface of the zinc spraying layer can have fine concave-convex and pores, and the traditional Sobel operator has weak response to low-contrast edges, which can easily lead to missed detection.

[0076] In this embodiment, an improved Sobel operator is used, and the core optimization is reflected in:

[0077] The size of the convolution kernel is enlarged (such as 5x5), and the ability to capture fine edges is enhanced. The small pore edges on the surface of the zinc spraying layer usually have small span, and a large size kernel can improve the gradient response;

[0078] An adaptive weight is introduced, which dynamically adjusts the convolution kernel weight according to the gray scale distribution of the pixel neighborhood, reduces the weight for the area with gentle gray scale change, and increases the weight for the suspected edge area, so as to enhance the edge response.

[0079] The gradient result calculated by the improved Sobel operator can accurately reflect the intensity and direction of the obvious edge on the surface of the zinc spraying layer, and can also capture the gray scale change of the fine structure, providing more comprehensive basic data for subsequent edge pixel screening.

[0080] The gradient calculation result contains a large number of edge pixels, but some pixels can be noise or "pseudo-edges" in the area with gentle gray scale change, such as the gray scale gradient caused by slight reflection on the surface of the zinc spraying layer. The double-threshold method sets two thresholds: a high threshold Th and a low threshold Tl, usually Th=2xTl, and classifies and screens the edge pixels.

[0081] The strong edge pixel is a pixel with a gradient amplitude greater than Th, which corresponds to an area with a sharp change in gray scale (such as a clear component boundary or a deep pore edge), and is determined as a "definite edge" and directly retained. The weak edge pixel is a pixel with a gradient amplitude between Tl and Th, which may be an extension of a real edge or a false edge and needs to be further verified. The invalid pixel is a pixel with a gradient amplitude less than Tl, which is directly determined as a non-edge and removed.

[0082] The verification logic for the weak edge pixel is: if the weak edge pixel is directly connected with the strong edge pixel, it is determined as part of a real edge and retained; if the weak edge pixel exists in isolation, it is determined as a false edge and removed.

[0083] The edge pixels screened by the double-threshold value may have breaks or burrs, which affect the accuracy of subsequent contour analysis. Morphological optimization processes the edge pixels through erosion and dilation operations. Specifically, the erosion operation scans the edge image with a structure element, only retaining the edge pixels completely covered by the structure element, which can remove small burrs and isolated points on the edge; the dilation operation also scans with a structure element, retaining the edge pixels as long as there is an overlap with the structure element, which can connect broken edge segments and fill small gaps.

[0084] In actual operation, an opening operation of "erosion followed by dilation" is usually used, or the order is adjusted according to the degree of edge breakage, to finally obtain a continuous and smooth edge contour. For example, the pore edge on the surface of a zinc sprayed component may be locally broken due to the shooting angle, and the broken part can be connected by the dilation operation to form a complete pore contour; the noise burrs on the edge are removed by the erosion operation to ensure the purity of the contour.

[0085] The continuous edge contour reflects the geometric shape of the surface of the zinc sprayed component, and the curvature value of the contour (a parameter describing the bending degree of the curve) can quantify the structural characteristics of different regions.

[0086] Preferably, in an embodiment of the present application, the edge detection result includes a volatile region, a planar stable region, and a pore dense region.

[0087] Based on the curvature value of the continuous edge contour, the surface optical image is divided into boundary regions to obtain an edge detection result, including:

[0088] The curvature value of each pixel point on the continuous edge contour is calculated, wherein the curvature value is obtained by fitting a quadratic curve based on the spatial coordinates of adjacent pixel points.

[0089] The region corresponding to the contour pixel point with a curvature value greater than a first curvature threshold is marked as a volatile region, and the region corresponding to the contour pixel point with a curvature value less than a second curvature threshold is marked as a potential planar region.

[0090] The gray image data of the potential planar region is subjected to texture feature extraction, a pore dense region of the potential planar region is screened based on the obtained geometric features, and a part of the potential planar region which is not marked as the pore dense region is divided into a planar stable region.

[0091] It should be noted that the curvature value is a key parameter for describing the bending degree of the contour, and the greater the curvature, the more severe the bending of the contour; the smaller the curvature, the flatter the contour, and the calculation is based on the spatial coordinates of the continuous edge contour, and is realized by local curve fitting:

[0092] Specifically, for the continuous edge contour after morphological optimization, the two-dimensional spatial coordinates of each pixel in the image coordinate system are extracted point by point to form a complete contour point set, ensuring that the position information of each point is traceable;

[0093] For the target pixel point on the contour, 11 points are selected as the local neighborhood, including the front and back 5 adjacent pixel points. The number of neighborhood points can balance the local details and overall trend, avoid fitting deviation caused by single point noise or distant contour interference, and adapt to the slow change characteristics of the zinc spraying component surface contour;

[0094] Based on the coordinate data of the neighborhood points, a quadratic curve is fitted by the least square method. The quadratic curve can effectively describe the bending state of the contour locally, whether it is a slight arc or a significant bend, and can accurately reflect the curve shape, providing a basic model for curvature value calculation;

[0095] The bending degree of the target pixel point is determined by the fitted quadratic curve shape - the more severe the curve bending, the greater the corresponding curvature value; the flatter the curve, the smaller the curvature value. The size of the curvature value directly reflects the complexity of the regional surface structure.

[0096] Further, based on the distribution characteristics of the curvature value, the preliminary classification of the region is realized through a pre-set threshold value. The first curvature threshold and the second curvature threshold are calibrated by sample statistics - a large number of edge contour curvature data of the same type of zinc spraying component are collected, and the distribution rule is analyzed. The first curvature threshold takes the critical value of the high curvature value interval, such as covering 90% of the severe bending region, and the second curvature threshold takes the critical value of the low curvature value interval, such as covering 90% of the flat region.

[0097] If the curvature value of the contour pixel point is greater than the first curvature threshold, it indicates that the contour of the region where the point is located is severely bent, and the thickness of the zinc spraying layer in such a region is prone to unevenness due to changes in the spraying angle and flow rate, so it is marked as an easy fluctuation region. If the curvature value of the contour pixel point is less than the second curvature threshold, it indicates that the contour of the region where the point is located is flat, but there may be pores, fine depressions and other microstructures, so it is temporarily marked as a potential planar region and needs to be further refined and analyzed.

[0098] It should be noted that although the potential flat area is overall flat, the microstructure difference is significant, and needs to be subdivided through texture feature extraction. Specifically, the gray image of the potential flat area is divided into sub-windows of a fixed size, such as 10x10 pixels, and the sub-window size is adapted to the typical size of the zinc spraying layer pores, such as 5-50 pixels, to ensure that the microstructure features can be effectively captured. For each sub-window, the geometric features related to the pores are extracted, including:

[0099] Local entropy, which reflects the degree of disorder of gray distribution, the entropy value of the pore area is high due to the light and dark jump; pore density, which distinguishes pore and non-pore pixels through threshold segmentation, calculates the proportion of pore pixels, and the higher the proportion, the more dense the pores; average pore area, which performs area statistics on the segmented pore connected domain, reflects the overall size of the pores.

[0100] Based on the sample training, the feature threshold is set, such as local entropy > 3.5, pore density > 10%, and average pore area > 8 pixels 2 The sub-window that meets the threshold condition is determined as a pore dense area. Such areas are prone to be affected by the effective thickness of the zinc spraying layer due to the presence of many pores, and need to be detected. In the potential flat area, the set of sub-windows that are not marked as pore dense areas have flat texture, few and small pores, and good uniformity of the zinc spraying layer thickness, and are divided into flat stable areas.

[0101] S2: based on the edge detection result, dynamically sampling the target zinc spraying member to obtain a to-be-detected area, wherein the dynamic sampling is designed to identify the degree of change of the surface information of the target zinc spraying member in the edge detection result, and the corresponding sampling density is determined based on the degree of change;

[0102] This step realizes the optimized allocation of detection resources through differentiated sampling strategies according to the area characteristics divided by edge detection - deploying high-density sampling points in key areas and reducing redundant sampling in uniform areas, and finally forming a to-be-detected area covering the key features of the member.

[0103] Preferably, in an embodiment of the present application, the dynamic sampling of the target zinc spraying member based on the edge detection result to obtain a to-be-detected area comprises:

[0104] According to the boundary area division in the edge detection result, the easy fluctuation area, the flat stable area and the pore dense area of the target zinc spraying member are determined;

[0105] Based on the differentiated sampling density, the easy fluctuation area, the flat stable area and the pore dense area are sampled to obtain a set of sampling points;

[0106] The coordinate verification is performed on the sampling point set to obtain a to-be-detected area covering a key area of the target zinc spraying component, wherein the coordinate verification is designed to eliminate invalid sampling points beyond the boundary of the target zinc spraying component.

[0107] In this embodiment, according to the boundary area division in the edge detection result, firstly, the easily fluctuating area, the planar stable area and the aperture dense area in the image coordinate system need to be mapped with the physical space coordinates of the target zinc spraying component, to ensure accurate correspondence between the image area and the actual component surface position. Specifically, by using the intrinsic parameters (focal length, pixel size) and extrinsic parameters (relative position and attitude with the component) of the image acquisition device, the pixel coordinates of the area in the image are converted into three-dimensional physical coordinates of the component surface, to realize one-to-one correspondence between the image area and the physical area. Taking the overall contour of the component obtained by edge detection as a reference, the boundary range of each area in the physical space is determined, to provide a coordinate reference for the spatial distribution of the subsequent sampling points.

[0108] According to the structural complexity and thickness detection sensitivity of different areas, different sampling densities are set based on the influence weight of the area on the overall quality of the zinc spraying layer and the thickness fluctuation risk, specifically including:

[0109] Easily fluctuating area: due to severe contour bending, the zinc spraying process is most affected by the spraying angle and flow rate, and the thickness fluctuation range can reach ±30% or more, so the highest sampling density is needed to capture local thickness differences;

[0110] Aperture dense area: although the overall contour is smooth, apertures can cause effective thickness to decrease, and the randomness of aperture distribution is strong, so a medium sampling density is needed to cover potential defect areas;

[0111] Planar stable area: the surface is smooth and has few apertures, and the zinc spraying thickness is uniform, so the lowest sampling density is adopted to reduce redundant calculation.

[0112] Specific sampling parameters are determined by experiment calibration and engineering requirements to determine the density threshold. In an embodiment of the present application, the sampling density of the easily fluctuating area can be set to 50-100 points / cm 2 , to ensure that each small protrusion or depression is covered by a sampling point; the sampling density of the aperture dense area can be set to 20-50 points / cm 2 , to balance the aperture capture rate and detection efficiency; and the sampling density of the planar stable area can be set to 5-20 points / cm 2 , to ensure overall trend representation by uniform grid distribution.

[0113] The generated sampling points can have invalid points beyond the physical boundary of the target zinc spraying component due to region boundary fitting errors or mapping deviations, which need to be removed through coordinate checking. Specifically, the overall contour of the component obtained by edge detection is taken as the reference boundary, including the outer contour boundary of the component and the boundary of the zinc spraying layer coverage. The three-dimensional physical coordinates (x, y, z) of each sampling point are checked twice, including:

[0114] Space boundary checking: judging whether the sampling point (x, y) is within the two-dimensional projection range of the component outer contour, if it is beyond, it is marked as an invalid point;

[0115] Zinc spraying layer coverage checking: combined with the zinc spraying area recognition result of the previous image, the sampling points in the unsprayed area are removed.

[0116] The effective sampling points after checking are subjected to local density optimization. After coordinate checking and optimization, the remaining effective sampling point set constitutes the detection area. Through the above dynamic sampling process, both the local thickness characteristics of the fluctuation area and the pore dense area can be captured through high-density sampling, and the overall state of the flat stable area can be efficiently represented through low-density sampling, which not only ensures the detection accuracy, but also improves the sampling efficiency by 40%-60% compared with uniform sampling, providing a precise and efficient detection point set basis for subsequent thickness detection.

[0117] S3: in response to the thickness measurement signal, performing compound ranging detection on the detection area based on dual-wavelength laser to obtain first thickness data, and performing three-dimensional topography detection on the detection area to obtain surface roughness parameters and surface fluctuation size parameters;

[0118] This step obtains the initial thickness data of the zinc spraying layer through dual-wavelength laser compound ranging, and captures the surface micro features combined with three-dimensional topography detection, providing basic data for subsequent thickness correction.

[0119] Preferably, in an embodiment of the present application, the dual-wavelength laser includes a zinc characteristic wavelength and a base material characteristic wavelength of the target zinc spraying component;

[0120] Performing compound ranging detection on the detection area based on dual-wavelength laser to obtain first thickness data, including:

[0121] Positioning the dual-wavelength laser ranging module to each sampling point of the detection area, and controlling the dual-wavelength laser to emit to the surface of the sampling point synchronously to obtain the dual-wavelength laser flight time according to the laser receiver;

[0122] Based on the dual-wavelength laser flight time, the first distance value from the zinc spraying layer surface to the dual-wavelength laser ranging module and the second distance value from the base material surface to the dual-wavelength laser ranging module are calculated respectively;

[0123] A dual-wavelength distance difference value of each sampling point is calculated based on the first distance value and the second distance value, to obtain first thickness data of the region to be detected.

[0124] It should be noted that the core of the dual-wavelength laser composite ranging is to utilize the differential response (reflection / transmission characteristics) of zinc and the base material to specific wavelength laser to achieve accurate ranging of the surface of the zinc spraying layer and the surface of the base material, so as to directly calculate the physical thickness.

[0125] In the embodiment, the dual-wavelength laser ranging module (integrating a laser emitter, a receiver, an optical lens and a positioning mechanism) is driven by a precision guide rail or a mechanical arm to be positioned in turn according to the sampling point coordinates of the region to be detected. The positioning accuracy is controlled within ±0.01 mm, so as to ensure that the deviation between the center of the laser spot and the sampling point coordinates is less than or equal to 0.05 mm, and the measurement error caused by sampling deviation is avoided.

[0126] The zinc characteristic wavelength is selected to be matched with the laser of the absorption peak of zinc element, such as 213.9 nm, that is, the resonance absorption wavelength of zinc atom. The wavelength laser is strongly reflected on the surface of the zinc spraying layer and is difficult to penetrate to the base material. The base material characteristic wavelength is selected to be a wavelength that can penetrate the zinc spraying layer, such as 1064 nm, which has strong penetration to the zinc layer and is highly reflected by the steel base material (if the base material is steel), so as to ensure that it can reach the surface of the base material and return. Table 1 is a table of base materials and characteristic wavelengths.

[0127] Table 1

[0128]

[0129] The dual-wavelength laser is triggered to be synchronously emitted by a high-precision timing controller, and the emission interval is less than or equal to 5 ns, so as to ensure that the measurement time difference of the two lasers at the same sampling point is negligible, and the distance deviation caused by the slight vibration of the component is avoided.

[0130] The laser receiver adopts a photodiode to detect the reflected light signal with high sensitivity, and a time-to-digital converter is used to record the flight time of the laser from emission to reception. Based on the propagation speed of the laser in the air, the first distance value, the distance from the surface of the zinc spraying layer to the module, and the second distance value, the distance from the surface of the base material to the module, are calculated respectively. The physical thickness of the zinc spraying layer at each sampling point is the dual-wavelength distance difference value.

[0131] After all the sampling points are traversed, a first thickness data set of the region to be detected is formed, which directly reflects the physical accumulation thickness of the zinc spraying layer, but does not consider the interference factors such as surface topography and oxidation.

[0132] Preferably, in one embodiment of the present application, the three-dimensional topography detection of the region to be detected is performed to obtain surface roughness parameters and surface fluctuation size parameters, which comprises:

[0133] Collecting regionally three-dimensional topography data of the to-be-detected region to obtain original point cloud data of the to-be-detected region;

[0134] Performing noise filtering, baseline correction and data smoothing processing on the original point cloud data in sequence to obtain three-dimensional profile data;

[0135] Calculating surface roughness parameters of the to-be-detected region based on the three-dimensional profile data, and performing frequency spectrum analysis on the three-dimensional profile data according to Fourier transform to obtain surface fluctuation size parameters of the to-be-detected region.

[0136] It should be noted that three-dimensional topography detection captures the microscopic concave-convex features of the zinc spraying layer surface, quantifies the roughness and macro fluctuation size, and provides parameter support for subsequent correction of the interference of surface topography on thickness measurement. Among them, the surface roughness refers to the microscopic geometric shape features composed of smaller pitch peaks and valleys on the surface of the component, and the surface fluctuation refers to the microscopic geometric shape features composed of larger pitch peaks and valleys with a certain macro trend on the surface of the component. The pitch of the peaks and valleys is usually larger than that of the surface roughness, and the fluctuation amplitude is also relatively larger, which can reflect the high-low change trend of the surface at a larger scale. As shown in Figure 2 , Figure 2 is a schematic diagram of the surface roughness / surface fluctuation of a component in the embodiment of the present application.

[0137] In this embodiment, a laser microscope or a white light interferometer is used to adapt to the micron-level roughness and millimeter-level fluctuation features of the zinc spraying layer surface. The edge detection divided easy fluctuation area, pore dense area and plane stable area are used to collect the original point cloud data of each area. Table 2 is a scanning parameter comparison table.

[0138] Table 2

[0139]

[0140] Further, the original point cloud data is preprocessed, and the statistical filtering method is used to eliminate outliers, which specifically includes calculating the average distance of each point and the neighborhood points, and if the distance is greater than 3 times the standard deviation, it is determined as noise, and the missing values are completed by interpolation. The overall tilt is eliminated by plane fitting, which specifically includes least square plane fitting of the preprocessed point cloud, calculation of the height deviation of each point relative to the fitting plane, and elimination of systematic errors caused by component placement tilt. The corrected point cloud is smoothed by Gaussian filtering to suppress high-frequency noise and retain effective topographic features above microns to obtain three-dimensional profile data.

[0141] Based on the three-dimensional profile data, surface roughness parameters are calculated, including: Ra (arithmetic mean deviation), which is the arithmetic mean of the absolute values of the heights of all points within the sampling length relative to the reference surface; Rz (maximum height), which is the height difference between the highest peak and the lowest valley within the sampling length; Rq (root mean square deviation), which is the root mean square value of the height deviation, which is more sensitive to the degree of surface fluctuation.

[0142] Further, a one-dimensional Fourier transform is performed on the z-direction height signal of the profile data to obtain a frequency spectrum, with the horizontal axis being the spatial frequency and the vertical axis being the amplitude. The wavelength corresponding to the peak value in the frequency spectrum is the main fluctuation size; the peak amplitude reflects the strength of the fluctuation of this size. Finally, surface fluctuation size parameters such as the dominant fluctuation wavelength and the average fluctuation amplitude are obtained.

[0143] S4: inputting at least the surface roughness parameters and the surface fluctuation size parameters into a pre-trained thickness correction model, and performing deviation correction on the first thickness data based on the thickness deviation value output by the thickness correction model to obtain second thickness data;

[0144] This step eliminates the influence of surface topography and other interference factors on the first thickness data through a data-driven thickness correction model, and obtains second thickness data closer to the true value.

[0145] Preferably, in one embodiment of the present application, the at least inputting of the surface roughness parameters and the surface fluctuation size parameters into the pre-trained thickness correction model, and the performing of deviation correction on the first thickness data based on the thickness deviation value output by the thickness correction model to obtain second thickness data, comprises:

[0146] Obtaining roughness parameter sets, surface fluctuation size parameter sets, first thickness measurement data sets and corresponding true thickness data sets of a plurality of zinc spraying component samples, and calculating a deviation data set between the first thickness measurement data sets and the true thickness data sets;

[0147] Constructing a thickness correction model based on an improved random forest regression algorithm, wherein the thickness correction model takes the roughness parameter sets and the surface fluctuation size parameter sets as input variables, and takes the deviation data set as an output variable;

[0148] Inputting the surface roughness parameters and the surface fluctuation size parameters of the region to be detected into the thickness correction model to obtain a corresponding target deviation amount, and performing correction calculation on the first thickness data based on the target deviation amount to obtain second thickness data.

[0149] In this embodiment, at least 500 zinc spraying component samples are collected, and four types of data are synchronously obtained for each sample to construct a training data set, i.e., an input feature set, first thickness measurement data, true thickness data, and a deviation data set. The input feature set can include surface roughness parameters (arithmetic mean deviation, maximum height, root mean square deviation), surface relief size parameters (dominant relief wavelength, average relief amplitude), environmental parameters (temperature and humidity at the time of detection), and pore density data (pore number per unit area). The true thickness data is obtained by a destructive detection method and serves as a reference standard for deviation calculation. The deviation data set is the difference between the first thickness and the true thickness.

[0150] In the data preprocessing stage, abnormal values in the deviation and roughness features are removed by a preprocessing method, and Z-score normalization is used to eliminate the dimension difference of different features. The training set and the validation set are divided in a ratio of 7:3 to ensure data distribution consistency.

[0151] In this embodiment, an improved random forest regression algorithm is selected as the basic model because it can handle high-dimensional nonlinear features and has strong robustness to noise data, and is suitable for complex mapping relationships of zinc spraying layer thickness deviation. The improved random forest regression algorithm of this embodiment is optimized for the non-linear features of the zinc spraying layer thickness deviation in three ways: the contribution of each input feature is calculated by recursive feature elimination, and high importance features are given higher weights when the decision tree is split; the Bayesian optimization algorithm is used to search for the optimal combination of hyperparameters, including the number of decision trees (200-500), the maximum depth (10-30 layers), and the minimum number of split samples (5-10), to avoid overfitting or underfitting of the model; the gradient boosting mechanism is introduced to further reduce the deviation by stacking the XGBoost and random forest.

[0152] During the model training process, the input features (such as roughness, relief size, and pore density) of the training set are used as independent variables, and the deviation data set is used as the dependent variable. The mapping relationship between the features and the deviation is fitted by the improved random forest algorithm to generate a thickness correction model.

[0153] Further, the roughness parameters, surface relief size parameters, and other data of each sampling point obtained in step S3 are input into the thickness correction model, and the model outputs the target deviation amount of each sampling point. The second thickness data is equal to the first thickness data minus the target deviation amount predicted by the model. After correction, the second thickness data eliminates the interference of surface topography, environment, and other factors, and is closer to the actual physical thickness of the zinc spraying layer.

[0154] S5: identifying the oxidation reaction information of the target zinc spraying component, and correcting the second thickness data based on the oxidation reaction information to obtain third thickness data;

[0155] Preferably, in one embodiment of the present application, the oxidation reaction information of the target zinc spraying component is identified, and the second thickness data is corrected based on the oxidation reaction information to obtain third thickness data, which comprises:

[0156] The characteristic spectral data of the to-be-detected region is obtained based on the laser-induced breakdown spectroscopy detection method, and the zinc element characteristic peak intensity value and the oxygen element characteristic peak intensity value in the characteristic spectral data are extracted;

[0157] A calibration curve of the zinc-oxygen element characteristic peak intensity ratio and the oxide layer thickness is constructed, and the oxide layer thickness value of each sampling point in the to-be-detected region is calculated according to the zinc element characteristic peak intensity value, the oxygen element characteristic peak intensity value, and the calibration curve;

[0158] The thickness value of each sampling point in the second thickness data is corrected based on the oxide layer thickness value to obtain the third thickness data after the oxide layer is stripped.

[0159] It should be noted that the laser-induced breakdown spectroscopy detection method is to obtain the element characteristic spectrum generated by the interaction of laser and matter, so as to realize the qualitative and quantitative analysis of the oxidation reaction information. The principle is to focus high-energy laser pulses on the surface of the to-be-detected region of the target zinc spraying component, so that the laser energy is instantaneously absorbed by the surface matter of the component, so that the local region is rapidly heated to a plasma state. During the cooling process of the plasma, atoms or ions will emit characteristic spectra of specific wavelengths, and these spectra carry the element composition information of the matter.

[0160] Specifically, by focusing high-energy laser pulses on the surface of the zinc spraying layer, the local material is instantaneously gasified and ionized to form a high-temperature plasma. When the excited state atoms / ions in the plasma transition, electromagnetic radiation of specific wavelengths is released. The spectrometer collects the plasma radiation light through an optical fiber, and after grating dispersion, the CCD detector converts it into an electrical signal to generate spectral data containing the characteristic peaks of each element.

[0161] Further, the original characteristic spectral data contains characteristic peak signals of multiple elements, and the characteristic peak intensity values of zinc and oxygen elements need to be accurately extracted from them. The characteristic peak intensity of zinc (Zn) and oxygen (O) is accurately extracted from the spectral data. Based on the known element characteristic wavelength, the target peak is located, for example, zinc can be selected as 213.9 nm or 334.5 nm, and oxygen can be selected as 777.2 nm or 844.6 nm. Then, the adaptive threshold algorithm is used to identify the characteristic peak vertex, calculate the peak intensity value, and normalize and calibrate the intensity value through the laser energy meter signal to ensure that the data of different sampling points are comparable.

[0162] Zinc element mainly exists in the zinc spraying layer, and oxygen element mainly comes from the oxide layer (such as ZnO and other oxides) generated by the oxidation reaction, and the characteristic peak intensity ratio of the two has a specific correlation with the thickness of the oxide layer. Building a calibration curve is a key link to realize the quantitative calculation of the thickness of the oxide layer. Specifically, a series of standard samples with known oxide layer thickness need to be prepared first. The preparation of the standard sample can be realized by controlling the oxidation conditions (such as temperature, humidity, oxidation time, etc.) of the zinc spraying component, or by using physical methods to prepare oxide layers of different thicknesses on pure zinc layers. For each standard sample, the characteristic peak intensity values of zinc and oxygen elements are obtained by using the above LIBS detection method, and the zinc-oxygen element characteristic peak intensity ratio is calculated.

[0163] Further, the actual thickness of the oxide layer of the standard sample is taken as the ordinate, and the corresponding zinc-oxygen element characteristic peak intensity ratio is taken as the abscissa, and a mathematical relationship model between the two, i.e. the calibration curve, is established by data fitting. The fitting method can be linear fitting, polynomial fitting or nonlinear fitting according to the data distribution characteristics. In the fitting process, the goodness of fit of the calibration curve needs to be tested, and the determination coefficient (R 2 ) is usually used for evaluation, and R 2 closer to 1, the better the fitting effect of the calibration curve. At the same time, the accuracy of the calibration curve needs to be verified by a verification sample to ensure its good applicability within a certain range of oxide layer thickness.

[0164] For the target zinc spraying component to be detected, multiple point sampling is carried out according to the preset sampling interval, and the characteristic peak intensity values of zinc and oxygen elements are obtained by LIBS detection at each sampling point, and the zinc-oxygen element characteristic peak intensity ratio is calculated.

[0165] The zinc-oxygen element characteristic peak intensity ratio of each sampling point is substituted into the calibration curve that has been constructed, and the thickness value of the oxide layer at the sampling point is obtained by reverse calculation through the mathematical model of the calibration curve. In the calculation process, abnormal values need to be identified and processed, for example, when the intensity ratio of a certain sampling point exceeds the effective range of the calibration curve, it needs to be re-detected or reasonably estimated by using the interpolation method to ensure the reliability of the thickness value of the oxide layer at each sampling point.

[0166] The second thickness data is the total thickness detection value of the zinc spraying component containing the oxide layer, and since the oxide layer is not an effective zinc spraying protective layer, it needs to be separated from the total thickness to obtain the third thickness data reflecting the actual effective zinc spraying layer thickness.

[0167] The correction process is based on the oxide layer thickness value of each sampling point, and the formula "third thickness data = second thickness data - oxide layer thickness value" is used for calculation. For the detection area with multiple sampling points, the correction calculation needs to be completed point by point, and finally the third thickness data of the whole detection area after stripping the oxide layer is obtained. After the correction is completed, the third thickness data needs to be analyzed for uniformity, and the actual distribution quality of the zinc spraying layer is evaluated, which provides accurate data support for the performance evaluation and quality control of the target zinc spraying component.

[0168] S6: performing regional adaptation processing on the third thickness data to obtain the zinc spraying layer thickness detection result of the target zinc spraying component.

[0169] This step converts the third thickness data, i.e. the pure zinc spraying layer thickness after stripping the oxide layer, into a comprehensive detection result that can reflect the overall quality of the component through regional weighted fusion strategy.

[0170] Preferably, in an embodiment of the present application, the regional adaptation processing on the third thickness data to obtain the zinc spraying layer thickness detection result of the target zinc spraying component comprises:

[0171] According to the regional division of the edge detection result, the weight coefficient corresponding to each region is assigned, wherein the weight coefficient is set based on the influence degree of the region on the overall performance of the zinc spraying layer;

[0172] The number of sampling points of the third thickness data in each region is counted, and the arithmetic mean of the third thickness data of all sampling points in each region is calculated to obtain the regional average thickness value;

[0173] The regional average thickness value of each region is multiplied by the weight coefficient of the region to obtain the regional weighted thickness value, and the zinc spraying layer thickness value of the target zinc spraying component is obtained by the regional weighted thickness value.

[0174] In this embodiment, the setting of the weight coefficient needs to be based on the influence degree of each region on the overall function of the zinc spraying layer, and is calibrated through engineering experience and experimental data to ensure that the thickness contribution of the key region is fully reflected.

[0175] It can be understood that the core function of the zinc spraying layer is to achieve corrosion protection by uniformly covering the base material, and the failure risk of different areas is significantly different. The easy fluctuation area is mostly the area with large curvature such as the corner of the component, which is prone to uneven thickness due to angle deviation during spraying, and is more prone to coating damage due to stress concentration during service, which is a high-risk area of corrosion failure and has the greatest impact on overall performance; the surface pores of the pore-intensive area are prone to reduce the compactness of the coating, leading to accelerated water vapor penetration and substrate corrosion, although the overall profile is smooth, but the micro defects have a significant impact on the corrosion performance; the surface of the flat stable area is smooth and the thickness is uniform, and the thickness fluctuation is small when the spraying process is stable, and the failure risk is the lowest, and the impact on the overall performance is relatively secondary.

[0176] In an embodiment of the present application, the typical weight coefficient can be set as:

[0177] Easy fluctuation area: weight coefficient = 0.4; pore-intensive area: weight coefficient = 0.35; flat stable area: weight coefficient = 0.25. The total weight sum needs to satisfy + + = 1, which can be adjusted according to the use scene of the component.

[0178] The third thickness data of each area is statistically averaged to eliminate local random errors and obtain a characteristic value representing the overall thickness level of the area. Specifically, the third thickness data set is traversed, the number of valid sampling points in each area is counted according to the area division rule of edge detection, and the arithmetic mean value of the third thickness data in each area is calculated, which is expressed as:

[0179]

[0180] wherein, is the sampling point in the area, is the sampling point index in the area, is the total number of sampling points in the area, and the comprehensive thickness value reflecting the overall zinc spraying layer quality of the component is obtained by summing the product of the area average thickness and the weight coefficient.

[0181] Another embodiment of the present application provides a zinc spraying layer thickness detection system, specifically, please refer to Figure 3 , Figure 3 showing the flowchart of the zinc spraying layer thickness detection system in one embodiment of the present application, which includes:

[0182] The acquisition module 11 is used to acquire the surface optical image of the target zinc spraying component in response to the dynamic sampling signal, and perform edge detection on the surface optical image to obtain an edge detection result.

[0183] The sampling module 12 is configured to sample the target zinc spraying component based on the edge detection result to obtain a detection area, wherein the dynamic sampling is designed to identify a variation degree of surface information of the target zinc spraying component in the edge detection result, and determine a corresponding sampling density based on the variation degree;

[0184] The measuring module 13 is configured to perform compound ranging detection on the detection area based on the dual-wavelength laser in response to the thickness measurement signal to obtain first thickness data, and perform three-dimensional topography detection on the detection area to obtain a surface roughness parameter and a surface fluctuation size parameter.

[0185] The first correction module 14 is configured to input at least the roughness parameter and the surface fluctuation size parameter into a pre-trained thickness correction model, and perform bias correction on the first thickness data based on a thickness bias value output by the thickness correction model to obtain second thickness data.

[0186] The second correction module 15 is configured to identify oxidation reaction information of the target zinc spraying component, and correct the second thickness data based on the oxidation reaction information to obtain third thickness data.

[0187] The output module 16 is configured to perform area adaptation processing on the third thickness data to obtain a zinc spraying layer thickness detection result of the target zinc spraying component.

[0188] The above embodiments only express several embodiments of the present application, and the description is relatively specific and detailed, but it cannot be understood as a limitation on the patent scope of the present application. It should be noted that for ordinary skilled persons in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application should be subject to the appended claims.

Claims

1. A method for detecting a zinc layer thickness, characterized by, The method comprises the following steps: in response to a dynamic sampling signal, acquiring a surface optical image of a target zinc spraying component, and performing edge detection on the surface optical image to obtain an edge detection result; based on the edge detection result, performing dynamic sampling on the target zinc spraying component to obtain a detection area, wherein the dynamic sampling is designed to identify the degree of change of the surface information of the target zinc spraying component in the edge detection result, and to determine the corresponding sampling density based on the degree of change; in response to a thickness measurement signal, performing compound ranging detection on the detection area based on dual-wavelength laser to obtain first thickness data, and performing three-dimensional topography detection on the detection area to obtain surface roughness parameters and surface fluctuation size parameters; at least inputting the surface roughness parameters and the surface fluctuation size parameters into a pre-trained thickness correction model, and correcting the first thickness data based on the thickness deviation value output by the thickness correction model to obtain second thickness data; identifying oxidation reaction information of the target zinc spraying component, and correcting the second thickness data based on the oxidation reaction information to obtain third thickness data; performing regional adaptation processing on the third thickness data to obtain zinc spraying layer thickness detection results of the target zinc spraying component; wherein the dual-wavelength laser comprises a zinc characteristic wavelength and a base material characteristic wavelength of the target zinc spraying component; performing compound ranging detection on the detection area based on dual-wavelength laser to obtain first thickness data, comprising: positioning the dual-wavelength laser ranging module to each sampling point of the detection area, and controlling the dual-wavelength laser to be emitted synchronously to the surface of the sampling point to obtain the dual-wavelength laser time of flight according to the laser receiver; based on the dual-wavelength laser time of flight, calculating a first distance value from the zinc spraying layer surface to the dual-wavelength laser ranging module and a second distance value from the base material surface to the dual-wavelength laser ranging module, respectively; based on the first distance value and the second distance value, calculating a dual-wavelength distance difference value of each sampling point to obtain the first thickness data of the detection area.

2. The method of claim 1, wherein the zinc layer thickness is detected by the method of claim 1. performing edge detection on the surface optical image to obtain an edge detection result, comprising: performing gray scale conversion and smoothing denoising processing on the surface optical image in sequence to obtain gray scale image data; based on an improved Sobel operator, performing gradient calculation on the gray scale image data, and based on a double-threshold method, screening effective edge pixels of the gradient calculation result, wherein the effective edge pixels include strong edge pixels and weak edge pixels; performing morphological optimization on the strong edge pixels and the weak edge pixels to obtain a continuous edge contour; based on the curvature value of the continuous edge contour, dividing the boundary region of the surface optical image to obtain the edge detection result.

3. The method of claim 2, wherein the step of detecting the thickness of the zinc layer is performed by using a laser. The edge detection result includes a fluctuation-prone area, a planar stable area, and a pore-dense area; based on the curvature value of the continuous edge contour, dividing the boundary region of the surface optical image to obtain the edge detection result, comprising: calculating the curvature value of each pixel point on the continuous edge contour, wherein the curvature value is obtained by fitting a quadratic curve based on the spatial coordinates of adjacent pixel points; Mark the region corresponding to the contour pixel point with the curvature value greater than the first curvature threshold as a fluctuation-prone region, and mark the region corresponding to the contour pixel point with the curvature value less than the second curvature threshold as a potential planar region; Perform texture feature extraction on the gray image data of the potential planar region, filter the pore-dense region of the potential planar region based on the obtained geometric features, and divide the part of the potential planar region that is not marked as the pore-dense region into a planar stable region.

4. The method of claim 3, wherein the step of detecting the thickness of the zinc layer is performed by using a laser. The dynamic sampling of the target zinc spraying component based on the edge detection result to obtain the to-be-detected region comprises: According to the boundary region division in the edge detection result, determine the fluctuation-prone region, the planar stable region and the pore-dense region of the target zinc spraying component; Sample the fluctuation-prone region, the planar stable region and the pore-dense region based on differentiated sampling densities to obtain a set of sampling points; Perform coordinate verification on the set of sampling points to obtain a to-be-detected region covering the key region of the target zinc spraying component, wherein the coordinate verification is designed to eliminate invalid sampling points that exceed the boundary of the target zinc spraying component.

5. The method of claim 1, wherein the thickness of the zinc layer is detected by the method of claim 1. The three-dimensional topography detection of the to-be-detected region to obtain the surface roughness parameter and the surface fluctuation size parameter comprises: Perform regional three-dimensional topography data collection on the to-be-detected region to obtain original point cloud data of the to-be-detected region; Perform noise filtering, baseline correction and data smoothing processing on the original point cloud data in sequence to obtain three-dimensional contour data; Calculate the surface roughness parameter of the to-be-detected region based on the three-dimensional contour data, and perform frequency spectrum analysis on the three-dimensional contour data according to Fourier transform to obtain the surface fluctuation size parameter of the to-be-detected region.

6. The method of detecting the thickness of a zinc layer according to claim 1, wherein The at least inputting the surface roughness parameter and the surface fluctuation size parameter into the pre-trained thickness correction model, and performing deviation correction on the first thickness data based on the thickness deviation value output by the thickness correction model to obtain second thickness data comprises: Obtain roughness parameter sets, surface fluctuation size parameter sets, first thickness measurement data sets and corresponding real thickness data sets of a plurality of zinc spraying component samples, and calculate the deviation data set between the first thickness measurement data set and the real thickness data set; Construct a thickness correction model based on an improved random forest regression algorithm, wherein the thickness correction model takes the roughness parameter set and the surface fluctuation size parameter set as input variables, and takes the deviation data set as an output variable; Input the surface roughness parameter and the surface fluctuation size parameter of the to-be-detected region into the thickness correction model to obtain the corresponding target deviation amount, and perform correction calculation on the first thickness data based on the target deviation amount to obtain second thickness data.

7. The zinc layer thickness detection method of claim 1, wherein, The identification of the oxidation reaction information of the target zinc spraying component, and the correction of the second thickness data based on the oxidation reaction information to obtain third thickness data comprises: Obtain feature spectrum data of the to-be-detected region based on laser-induced breakdown spectroscopy detection, and extract zinc element characteristic peak intensity values and oxygen element characteristic peak intensity values in the feature spectrum data; constructing a calibration curve of the intensity ratio of the zinc element characteristic peak and the oxide layer thickness, and calculating the oxide layer thickness value of each sampling point in the to-be-detected region according to the zinc element characteristic peak intensity value, the oxygen element characteristic peak intensity value and the calibration curve; correcting the thickness value of each sampling point in the second thickness data based on the oxide layer thickness value to obtain third thickness data after the oxide layer is stripped.

8. The method of detecting the thickness of a zinc layer according to claim 1, wherein The region-adaptive processing of the third thickness data to obtain the zinc spraying layer thickness detection result of the target zinc spraying component includes: According to the region division of the easily fluctuating area, the planar stable area and the pore dense area determined according to the edge detection result, a corresponding weight coefficient is assigned to each region, wherein the weight coefficient is set based on the influence degree of the region on the overall performance of the zinc spraying layer; The number of sampling points of the third thickness data in each region is counted, and the arithmetic mean of the third thickness data of all sampling points in each region is calculated to obtain a regional average thickness value; The regional average thickness value of each region is multiplied by the weight coefficient of the region to obtain a regionally weighted thickness value, and the zinc spraying layer thickness value of the target zinc spraying component is obtained based on the regionally weighted thickness value.

9. A zinc layer thickness detection system, characterized by, It includes: An acquisition module is configured to acquire a surface optical image of a target zinc spraying component in response to a dynamic sampling signal, and perform edge detection on the surface optical image to obtain an edge detection result; A sampling module is configured to perform dynamic sampling on the target zinc spraying component based on the edge detection result to obtain a to-be-detected region, wherein the dynamic sampling is designed to identify the degree of change of the surface information of the target zinc spraying component in the edge detection result, and determine the corresponding sampling density based on the degree of change; A measurement module is configured to perform compound ranging detection on the to-be-detected region based on dual-wavelength laser in response to a thickness measurement signal to obtain first thickness data, and perform three-dimensional topography detection on the to-be-detected region to obtain surface roughness parameters and surface undulation size parameters; A first correction module is configured to input at least the roughness parameters and the surface undulation size parameters into a pre-trained thickness correction model, and perform deviation correction on the first thickness data based on a thickness deviation value output by the thickness correction model to obtain second thickness data; A second correction module is configured to identify oxidation reaction information of the target zinc spraying component, and correct the second thickness data based on the oxidation reaction information to obtain third thickness data; An output module is configured to perform region-adaptive processing on the third thickness data to obtain a zinc spraying layer thickness detection result of the target zinc spraying component; The dual-wavelength laser includes a zinc characteristic wavelength and a base material characteristic wavelength of the target zinc spraying component; The compound ranging detection on the to-be-detected region based on the dual-wavelength laser to obtain the first thickness data includes: Positioning the dual-wavelength laser ranging module to each sampling point of the to-be-detected region, and controlling the dual-wavelength laser to be synchronously emitted to the surface of the sampling point to obtain the time of flight of the dual-wavelength laser according to the laser receiver; The first distance value from the surface of the zinc spraying layer to the dual-wavelength laser ranging module and the second distance value from the surface of the base material to the dual-wavelength laser ranging module are respectively calculated based on the dual-wavelength laser time of flight; The dual-wavelength distance difference value of each sampling point is calculated based on the first distance value and the second distance value, so as to obtain the first thickness data of the to-be-detected area.

Images