A crack detection method and system for duplex steel

Abstract

The present application provides a crack detection method and system for duplex steel, which relates to the field of crack detection technology. The method comprises: building a detection platform for duplex steel crack detection; traversingly scanning duplex steel through the detection platform to obtain an initial steel data set, performing wavelet packet denoising on the initial steel data set to obtain an optimized steel data set; obtaining a steel image according to the optimized steel data set; processing the steel image using an image processing algorithm to construct a steel crack information set; establishing a scoring system based on the steel crack information set to construct a steel quality evaluation model; obtaining a quality assessment grade for the inspected steel and generating a duplex steel quality assessment report. The method solves the technical problems in the prior art such as the inability to clearly detect internal cracks, cumbersome operation process, slow detection speed, poor detection accuracy, and inability to integrate evaluation and analysis.

Description

Technical Field

[0001] The present application relates to the technical field of crack detection, and in particular to a crack detection method and system for duplex steel. Background Art

[0002] With the rapid development of modern industry, duplex steel, thanks to its excellent combination of high strength, toughness, and corrosion resistance, is increasingly being used in numerous fields. For example, in industries such as automotive manufacturing, petrochemicals, and marine engineering, duplex steel is widely used in the manufacture of critical components. Due to the critical importance of duplex steel components in practical applications, quality assurance during their production is crucial, and crack detection is a crucial component among these quality inspection criteria. Cracks can severely impact the strength, toughness, and service life of duplex steel components, and may even cause sudden failure during use, leading to safety incidents. Therefore, to ensure the quality and safety of duplex steel products, effective crack detection is essential during the production process.

[0003] Currently, existing crack detection technologies for duplex steel have numerous limitations. For example, while traditional magnetic particle testing can detect surface cracks, its ability to detect internal cracks is limited. Magnetic particle testing also requires surface pretreatment, making the process cumbersome and slow. Its accuracy is also susceptible to interference from external magnetic fields. Furthermore, this testing method lacks integrated analytical tools and cannot directly assess steel quality, relying solely on manual estimation, which is inaccurate.

[0004] In summary, existing crack detection technologies for duplex steel suffer from technical shortcomings, including an inability to clearly detect internal cracks, cumbersome operation, slow detection speed, poor detection accuracy, and an inability to integrate assessment and analysis. Therefore, a method is needed to address these issues. Summary of the Invention

[0005] The present disclosure provides a crack detection method and system for duplex steel, which are used to solve the technical problems in the prior art, such as the inability to clearly detect internal cracks, cumbersome operation process, slow detection speed, poor detection accuracy, and inability to integrate evaluation and analysis.

[0006] According to a first aspect of the present disclosure, a method for detecting cracks in dual-phase steel is provided, comprising:

[0007] A detection platform is built for duplex steel crack detection. The detection platform includes a trigger device, a receiving device, and a moving device. The trigger device is used to emit laser vertically to the surface of the duplex steel to stimulate broadband ultrasound. The receiving device is used to obtain ultrasound data. The moving device is used to drive the trigger device and the receiving device to move.

[0008] traversally scanning the duplex steel using the detection platform to obtain an initial steel data set, performing wavelet packet noise reduction processing on the initial steel data set to obtain an optimized steel data set, wherein the optimized steel data set includes ultrasonic data and time-space labels, and the ultrasonic data includes direct wave data and reflected wave data;

[0009] Drawing a B-scan image according to the steel optimization data set, converting the B-scan image into an imaging image, and obtaining a steel imaging image;

[0010] The steel image is processed using an image processing algorithm to extract the crack edge contour and determine the number of cracks. The crack position and length are determined by analyzing the coordinate information of the crack edge contour. Based on the crack position, the crack position influencing factor and the crack depth are calculated. A steel crack information set is constructed based on the crack number, position influencing factor, length, and depth information. The steel crack information set constructed based on the crack number, position influencing factor, length, and depth information includes:

[0011] Use morphological operations and edge detection algorithms to optimize crack edges, extract crack edge contours based on contour tracking algorithms, and determine the number of cracks;

[0012] By calculating the center coordinates of the crack edge contour (x c ,y c ), determine the crack location, the specific formula is: where x i 、y i They represent the horizontal and vertical coordinates of the edge contour points, respectively, and n is the number of contour points;

[0013] The total length of the crack is calculated by superimposing the distance between the coordinates of the crack edge contour points. The specific formula is: Where i is the serial number of the point;

[0014] Calculate the crack position influence factor P according to the specific position of the crack in the duplex steel C , construct the following formula:

[0015] P C =αS R ;

[0016] Among them, P C is the crack position influencing factor, with a value range of 0 <P C <1, S R is the stress concentration coefficient at the crack location, and α is the normalization coefficient used to adjust P C The value range is between 0 and 1;

[0017] According to the crack location and the spatiotemporal labels in the steel optimization dataset, the crack depth is further calculated. The specific formula is: Among them, Δt is the reception time difference between the direct wave and the reflected wave, and v is the propagation speed of ultrasonic waves in duplex steel;

[0018] Construct a steel crack information set based on the crack quantity, position influence factor, length, and depth information. The steel crack information set is sorted according to the crack position, and the detailed information of each crack is collected and integrated;

[0019] Combine the crack quantity, crack position influence factor, crack length, and crack depth information in the steel crack information set to establish a scoring system and construct a steel quality evaluation model. Among them, constructing the steel quality evaluation model includes:

[0020] Construct a steel quality evaluation model based on the crack quantity, crack position influence factor, crack length, and crack depth. The formula is as follows:

[0021]

[0022] Among them, Q is the steel quality evaluation index, and the range is 0 < Q < 1. n is the total number of cracks, n max is the maximum allowable number of cracks, i is the crack serial number, L Ci is the crack length of the i-th crack, L max is the maximum allowable crack length, D ci is the crack depth of the i-th crack, D max is the maximum allowable crack depth, P Ci is the crack position influence factor of the i-th crack. Cracks in different positions have different influences. w1, w2, w3, and w4 are weight coefficients;

[0023] When Q < 0.5, it is judged as a severe defect and the quality is unqualified;

[0024] When 0.5 ≤ Q < 0.7, it is judged as a moderate defect and the quality is unqualified;

[0025] When 0.7 ≤ Q < 1, it is judged as a minor defect and the quality is qualified;

[0026] When Q is not in the range of 0 - 1, it means that there is a standard exceeding the allowable maximum value, and the steel quality is directly judged as unqualified;

[0027] Import the steel crack information set into the steel quality evaluation model to obtain the quality assessment grade of the inspected steel and generate a duplex steel quality evaluation report.

[0028] According to the second aspect of the present disclosure, a crack detection system for duplex steel is provided, including:

[0029] A detection platform construction module is used to build a detection platform for dual-phase steel crack detection. The detection platform includes a trigger device, a receiving device, and a moving device. The trigger device is used to emit laser vertically to the surface of the dual-phase steel to stimulate broadband ultrasound. The receiving device is used to obtain ultrasonic data. The moving device is used to drive the trigger device and the receiving device to move.

[0030] A data acquisition and optimization module is used to traversally scan the duplex steel through the detection platform to obtain an initial steel data set, perform wavelet packet noise reduction on the initial steel data set, and obtain an optimized steel data set, wherein the optimized steel data set includes ultrasonic data and time and space labels, and the ultrasonic data includes direct wave data and reflected wave data;

[0031] An imaging image acquisition module is used to draw a B-scan image based on the steel optimization data set, convert the B-scan image into an imaging image, and obtain a steel imaging image;

[0032] The crack information extraction module is used to process the steel imaging image using an image processing algorithm, extract the crack edge contour, determine the number of cracks, determine the crack position and length by analyzing the coordinate information of the crack edge contour, calculate the crack position influencing factor and crack depth based on the crack position, and construct a steel crack information set based on the crack number, position influencing factor, length, and depth information. The steel crack information set constructed based on the crack number, position influencing factor, length, and depth information includes:

[0033] Use morphological operations and edge detection algorithms to optimize crack edges, extract crack edge contours based on contour tracking algorithms, and determine the number of cracks;

[0034] By calculating the center coordinates of the crack edge contour (x c ,y c ), determine the crack location, the specific formula is: where x i 、y i They represent the horizontal and vertical coordinates of the edge contour points, respectively, and n is the number of contour points;

[0035] The total length of the crack is calculated by superimposing the distance between the coordinates of the crack edge contour points. The specific formula is: Where i is the serial number of the point;

[0036] Calculate the crack position influence factor P according to the specific position of the crack in the duplex steel C , construct the following formula:

[0037] P C =αS R ;

[0038] Among them, P C is the crack position influence factor, and the numerical range is 0 < P C < 1, S R is the stress concentration coefficient at the position where the crack is located, and α is the normalization coefficient used to adjust the P C value range to make it between 0 and 1;

[0039] According to the crack position and the spatio-temporal tags in the steel optimization dataset, the crack depth is further calculated. The specific formula is: Among them, Δt is the reception time difference between the direct wave and the reflected wave, and v is the propagation speed of ultrasonic waves in duplex steel;

[0040] Based on the crack number, position influence factor, length and depth information, a steel crack information set is constructed. The steel crack information set is sorted according to the crack position and汇集整合每条裂纹的详细信息;

[0041] Quality evaluation model construction module. The quality evaluation model construction module is used to establish a scoring system and construct a steel quality evaluation model by combining the crack number, crack position influence factor, crack length and crack depth information in the steel crack information set. Among them, constructing the steel quality evaluation model includes:

[0042] Based on the crack number, crack position influence factor, crack length, and crack depth, a steel quality evaluation model is constructed. The formula is as follows:

[0043]

[0044] Among them, Q is the steel quality evaluation index, and the range is 0 < Q < 1. n is the total number of cracks, n max is the maximum allowable number of cracks, i is the crack serial number, L Ci is the crack length of the i-th crack, L max is the maximum allowable crack length, D ci is the crack depth of the i-th crack, D max is the maximum allowable crack depth, P Ci is the crack position influence factor of the i-th crack. Cracks in different positions have different influences. w1, w2, w3, and w4 are weight coefficients;

[0045] When Q < 0.5, it is judged as a severe defect and the quality is unqualified;

[0046] When 0.5 <= Q < 0.7, it is judged as a moderate defect and the quality is unqualified;

[0047] When 0.7 <= Q < 1, it is judged as a minor defect and the quality is qualified;

[0048] It should be noted that the expression "汇集整合每条裂纹的详细信息" in the original text seems a bit unclear in its exact meaning. The above translation tries to convey the general idea as accurately as possible. If there are more specific requirements or corrections for this part, it can be adjusted accordingly.When Q is not in the range of 0-1, it means that the standard exceeds the maximum allowable value, and the steel quality is directly judged to be unqualified;

[0049] The quality assessment and reporting module is used to import the steel crack information set into the steel quality evaluation model, obtain the quality assessment grade of the inspected steel, and generate a duplex steel quality assessment report.

[0050] One or more technical solutions provided in this disclosure have at least the following technical effects or advantages:

[0051] A detection platform is built for duplex steel crack detection. The detection platform includes a trigger device, a receiving device, and a moving device. The trigger device is used to emit laser vertically to the surface of the duplex steel to stimulate broadband ultrasound. The receiving device is used to obtain ultrasound data. The moving device is used to drive the trigger device and the receiving device to move.

[0052] traversally scanning the duplex steel using the detection platform to obtain an initial steel data set, performing wavelet packet noise reduction processing on the initial steel data set to obtain an optimized steel data set, wherein the optimized steel data set includes ultrasonic data and time-space labels, and the ultrasonic data includes direct wave data and reflected wave data;

[0053] Drawing a B-scan image according to the steel optimization data set, converting the B-scan image into an imaging image, and obtaining a steel imaging image;

[0054] The steel image is processed using an image processing algorithm to extract the crack edge contour and determine the number of cracks. The crack position and length are determined by analyzing the coordinate information of the crack edge contour. Based on the crack position, the crack position influencing factor and the crack depth are calculated. A steel crack information set is constructed based on the crack number, position influencing factor, length, and depth information. The steel crack information set constructed based on the crack number, position influencing factor, length, and depth information includes:

[0055] Use morphological operations and edge detection algorithms to optimize crack edges, extract crack edge contours based on contour tracking algorithms, and determine the number of cracks;

[0056] By calculating the center coordinates of the crack edge contour (x c ,y c ), determine the crack location, the specific formula is: where x i 、y i They represent the horizontal and vertical coordinates of the edge contour points, respectively, and n is the number of contour points;

[0057] The total length of the crack is calculated by superimposing the distance between the coordinates of the crack edge contour points. The specific formula is: Where i is the serial number of the point;

[0058] Calculate the crack position influence factor P according to the specific position of the crack in the duplex steel C , and construct the following formula:

[0059] P C = αS R ;

[0060] Where, P C is the crack position influence factor, and the value range is 0 < P C < 1, S R is the stress concentration coefficient at the position where the crack is located, and α is the normalization coefficient, which is used to adjust the value range of P C to make it between 0 and 1;

[0061] Further calculate the crack depth based on the crack position and the spatio-temporal tags in the steel optimization dataset. The specific formula is: Where, Δt is the reception time difference between the direct wave and the reflected wave, and v is the propagation speed of ultrasonic waves in duplex steel;

[0062] Construct a steel crack information set based on the crack number, position influence factor, length and depth information. The steel crack information set is sorted according to the crack position, and the detailed information of each crack is collected and integrated;

[0063] Combine the crack number, crack position influence factor, crack length and crack depth information in the steel crack information set to establish a scoring system and construct a steel quality evaluation model. Among them, constructing a steel quality evaluation model includes:

[0064] Construct a steel quality evaluation model based on the crack number, crack position influence factor, crack length, and crack depth. The formula is as follows:

[0065]

[0066] Where, Q is the steel quality evaluation index, and the range is 0 < Q < 1. n is the total number of cracks, n max is the maximum allowable number of cracks, i is the crack serial number, L Ci is the crack length of the i-th crack, L max is the maximum allowable crack length, D ci is the crack depth of the i-th crack, D max is the maximum allowable crack depth, P Ci is the crack position influence factor of the i-th crack. Cracks in different positions have different influences. w1, w2, w3, and w4 are weight coefficients;

[0067] When Q < 0.5, it is judged as a severe defect and the quality is unqualified;

[0068] When 0.5 ≤ Q < 0.7, it is judged as a moderate defect and the quality is unqualified;

[0069] When 0.7 ≤ Q < 1, it is judged as a minor defect and the quality is qualified;

[0070] When Q is not within the range of 0 - 1, it means that there is a standard exceeding the allowable maximum value, and the steel quality is directly judged as unqualified;

[0071] Import the steel crack information set into the steel quality evaluation model, obtain the quality assessment grade of the inspected steel, and generate a dual - phase steel quality evaluation report.

[0072] It solves the technical problems in the prior art, such as the inability to clearly detect internal cracks, the cumbersome operation process, the slow detection speed, the poor detection accuracy, and the inability to perform integrated evaluation and analysis.

[0073] The above description is only an overview of the technical solution of this application. In order to be able to understand the technical means of this application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features, and advantages of this application more obvious and understandable, the following specifically gives the specific implementation manners of this application. Brief Description of the Drawings

[0074] In order to more clearly illustrate the technical solutions in the present disclosure or the prior art, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only exemplary. For those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0075] Figure 1 It is a schematic flow chart of a method for detecting cracks in dual - phase steel provided by an embodiment of this application;

[0076] [[ID=二十八]] Figure 2 It is a schematic structural diagram of a crack detection system for dual - phase steel provided by an embodiment of this application.

[0077] Explanation of the reference numerals in the drawings: Detection platform building module 11, data acquisition and optimization module 12, imaging map acquisition module 13, crack information extraction module 14, quality evaluation model construction module 15, quality assessment and report module 16. Detailed Embodiments [[ID=三十五]]

[0078] The following description of exemplary embodiments of the present disclosure is provided in conjunction with the accompanying drawings, which include various details of the embodiments of the present disclosure to facilitate understanding. These details should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0079] Example 1: A crack detection method for dual-phase steel provided in the present disclosure is described with reference to Figure 1 For illustration, the methods include:

[0080] S1: Build a detection platform for duplex steel crack detection, the detection platform includes a trigger device, a receiving device and a moving device, the trigger device is used to emit laser vertically to the surface of the duplex steel to stimulate broadband ultrasound, the receiving device is used to obtain ultrasound data, and the moving device is used to drive the trigger device and the receiving device to move;

[0081] Specifically, the inspection platform is a planar structure designed to support and inspect duplex steel. Built using laser ultrasonic technology and a contact-based inspection principle, it consists of a triggering device, a receiving device, and a moving device, supplemented by a synchronization control unit. The platform employs a modular design, with each device working collaboratively via a high-speed communication protocol. This ensures precise timing synchronization between laser emission, data acquisition, and motion positioning, meeting the requirements for high-precision inspection of the complex microstructure of duplex steel. The triggering device utilizes an Nd:YAG pulsed laser and is secured above the inspection platform via a fixture, ensuring vertical laser illumination of the sample surface. The receiving device, an ultrasonic angle probe, is mounted below the inspection platform, aligned with the triggering device and maintaining good contact with the sample surface. Ultrasonic coupling agent is applied between the probe and the sample surface to reduce ultrasonic reflection and attenuation at the interface, thereby improving signal transmission efficiency. The ultrasonic signal received by the probe is transmitted via a cable to a signal acquisition device, such as an oscilloscope, for recording and analysis. The moving device includes a displacement controller and a stepper motor to assist in the horizontal movement of the triggering and receiving devices.

[0082] S2: traversally scanning the duplex steel using the detection platform to obtain an initial steel data set, performing wavelet packet denoising on the initial steel data set to obtain an optimized steel data set, wherein the optimized steel data set includes ultrasonic data and time-space labels, and the ultrasonic data includes direct wave data and reflected wave data;

[0083] Specifically, since the duplex steel to be inspected is a uniform rectangular shape, a Cartesian coordinate system is established with the center of the steel as the coordinate origin, the length of the steel as the x-axis, the width as the y-axis, and the coordinate axis scale unit as 1mm. The raster scanning path, scanning beam diameter, and speed are set according to the coordinate system plan. For critical areas such as edges, an adaptive dense scanning strategy is adopted to reduce the local scanning spacing. An Nd:YAG pulsed laser is activated and incident perpendicularly on the surface of the steel to be inspected, ensuring that the thermoelastic effect dominates the ultrasonic wave generation. An ultrasonic oblique probe and oscilloscope are used to receive and store the ultrasonic signal, obtaining ultrasonic data at that location. The acquisition location and time of the ultrasonic signal are also recorded as the spatiotemporal label of the data. A mobile device drives the trigger device and receiving device to perform a traversal scan along a preset path on the steel surface, obtaining ultrasonic data for all locations to be inspected on the duplex steel. The acquired data is integrated into an initial steel dataset, which includes the inspection location, direct wave data and acquisition time, and reflected wave data and acquisition time.

[0084] The initial steel dataset is processed based on wavelet transform theory. The symlet wavelet basis function is preferentially used to perform wavelet transform on the initial ultrasonic signal. The Wpdec function is then used to perform a full binary tree decomposition of the wavelet-transformed signal to generate a wavelet packet coefficient tree. For each terminal node, its energy entropy and power spectral density are calculated, and a node feature matrix is ​​constructed. Adaptive threshold screening of terminal nodes is performed based on the Stein unbiased risk estimation criterion. Finally, the optimal subband is combined with the reconstructed signal, and the reconstructed ultrasonic signal is recombined with the corresponding spatiotemporal labels to obtain the optimized steel dataset.

[0085] S3: Drawing a B-scan image according to the steel optimization data set, converting the B-scan image into an imaging image, and obtaining a steel imaging image;

[0086] Specifically, the horizontal and vertical coordinates are determined based on the spatial information in the spatiotemporal labels of the steel optimization dataset. A linear quantization algorithm is used to represent the ultrasonic signal intensity corresponding to each detection point on the coordinate plane as brightness or grayscale. Stronger ultrasonic signal intensity corresponds to higher grayscale values ​​in the graph, and vice versa. During the rendering process, a bilinear interpolation algorithm is used to ensure image accuracy. This algorithm calculates the brightness or grayscale value of the interpolated point by taking a weighted average of four adjacent pixels. In this way, the information from the steel optimization dataset is mapped onto a two-dimensional plane, resulting in the creation of a B-scan image.

[0087] A histogram equalization algorithm redistributes the image's grayscale histogram, making the grayscale distribution more uniform and improving image contrast. A median filter algorithm removes noise. Through the combined application of these algorithms and specialized techniques, the B-scan image is ultimately transformed into a steel image that intuitively displays the internal structural characteristics of the steel, providing a powerful basis for steel quality assessment and analysis.

[0088] S4: Processing the steel image using an image processing algorithm to extract crack edge contours, determine the number of cracks, determine the crack position and length by analyzing the coordinate information of the crack edge contours, calculate the crack position influencing factor and crack depth based on the crack position, and construct a steel crack information set based on the crack number, position influencing factor, length, and depth information;

[0089] Specifically, the Canny edge detection algorithm is used to process steel imaging images. The specific operations include: using a Gaussian filter to smooth the steel imaging image; using a first-order partial derivative operator to calculate the gradient amplitude and direction of each pixel in the image; and performing non-maximum suppression on the gradient amplitude. Along the gradient direction, the gradient amplitude of each pixel is compared with that of its adjacent pixels. If the gradient amplitude of the pixel is not a local maximum, it is marked as a non-edge point, thereby refining the edge; setting a high threshold and a low threshold. If the gradient amplitude of a pixel is greater than the high threshold, it is determined to be an edge point. If the gradient amplitude is less than the low threshold, it is determined to be a non-edge point. If the gradient amplitude is between the high threshold and the low threshold, it is only considered an edge point if it is connected to a determined edge point. In this way, the edge contours of all cracks in the steel imaging image are obtained, and the total number of cracks is obtained.

[0090] After obtaining the crack edge contour, its coordinate information is analyzed. The crack location can be determined by the coordinate range of the pixel points in the edge contour. The crack length is calculated by superimposing the distances between the coordinates of the crack edge contour points. Simultaneously, the stress concentration factor is calculated based on the crack location, and the crack location influencing factor is derived. The crack depth is calculated by calculating the arrival time difference between the direct wave and the crack reflection wave in the steel optimization dataset, combined with the material sound velocity. The obtained crack number, location influencing factor, length, and depth information are stored in the steel crack information set.

[0091] S5: Establish a scoring system based on the number of cracks, crack location influencing factors, crack length and crack depth information in the steel crack information set to construct a steel quality evaluation model;

[0092] Specifically, a steel quality evaluation model was constructed using a weighted comprehensive evaluation logic approach based on the crack count, crack location influencing factors, crack length, and crack depth information contained in the steel crack information set. The model's construction logic comprehensively considers the impact of various crack-related factors on the performance of duplex steel. By assigning weights to each factor and weighted summing the scores from different aspects, a comprehensive evaluation score was obtained that fully reflects the steel's quality. The weight of each factor reflects its relative importance in the steel quality evaluation, while the score for each factor is derived based on the degree of influence of crack-related factors on steel quality. In this way, the quality of duplex steel can be systematically and comprehensively evaluated.

[0093] S6: Import the steel crack information set into the steel quality evaluation model, obtain the quality assessment grade of the inspected steel, and generate a duplex steel quality assessment report.

[0094] Specifically, crack information is extracted from the steel crack data set, imported into the steel quality assessment model, and a quality score, Q, is calculated. The quality assessment grading criteria are pre-defined. For example, three grading levels can be set: severe defects, moderate defects, and minor defects. When Q is within a high score range, it is considered minor defects, indicating that the quality of the inspected steel is good in all aspects, crack-related factors have a minimal impact on the steel's quality, and the steel can well meet the requirements of its use. When Q is in the middle score range, it is considered moderate defects, indicating that the steel has some crack-related issues. While these may not immediately lead to serious structural damage, they may degrade performance or pose safety hazards in long-term use or under specific conditions, and therefore should not be used. When Q is below a certain score, it is considered severe defects, indicating that the cracks in the steel have severely affected its quality and may not meet normal use requirements, requiring further treatment, such as repair, scrapping, or remanufacturing. When Q is outside the pre-defined score range, it indicates that some factors have significantly exceeded a fixed threshold. Without considering all influencing factors, the steel can be directly deemed unqualified. The calculated Q value is then used to determine the quality assessment grade of the inspected steel. Based on the quality assessment results, a duplex steel quality assessment report is generated. This report includes basic information such as steel batch and specifications. It details the crack profile, quality score, and rating, analyzes the main influencing factors, and provides recommendations. It also summarizes the quality status and provides recommendations for subsequent use and handling. The overall presentation is concise and accurate, clearly presenting the steel quality situation.

[0095] Furthermore, step S2 of this application also includes:

[0096] Plan the scanning path for the steel surface, adopt a rasterized full-coverage scanning mode, dynamically optimize the scanning spacing based on the spot size, and enable an adaptive encryption scanning strategy for edge areas;

[0097] The trigger device injects a pulsed laser vertically into the steel surface, generating broadband ultrasonic waves through the thermoelastic effect. The receiving device uses an ultrasonic angle probe and an oscilloscope to achieve synchronous acquisition of contact ultrasonic signals. The moving device drives the trigger device and the receiving device to move at a constant speed along the planned path, completing a double-sided line scan of the steel area to be inspected.

[0098] The receiving device acquires data into an initial steel data set, where the initial steel data set includes ultrasonic data and time-space labels corresponding to the data, and the ultrasonic data includes direct wave data and reflected wave data.

[0099] Specifically, based on the established coordinate system, the Delaunay triangulation algorithm is used to construct the initial rasterized scanning network. The scanning spacing is dynamically optimized based on the laser spot diameter. In the center region of the steel, the spacing is set to 1.2 to 1.5 times the spot diameter, achieving a balance between detection efficiency and spatial resolution. In the edge regions, a gradient sensitivity factor is introduced to automatically reduce the spacing to less than 0.8 times the spot diameter, ensuring a defect capture probability greater than 99%. Furthermore, when the scanning position is within 5-10 mm of the boundary, an adaptive dense scanning mode is enabled. A radial basis function interpolation algorithm is used to predict the potential defect distribution density in the edge stress concentration area. This prediction drives the local mesh refinement of the scanning path, reducing the scanning spacing to half that of the center region. Based on the set scanning network, the scanning position coordinates are determined, and the movement paths of the trigger and receiver are planned using an S-shaped path. Furthermore, a boundary buffer mechanism is embedded in the motion control system, and velocity and acceleration feedforward control is used to ensure smooth steering of the trigger and receiver in the edge regions, avoiding signal acquisition distortion caused by mechanical vibration.

[0100] The trigger device uses a short pulse laser to vertically incident on the steel surface. The laser is set to a pulse width of 5 to 20 ns and an energy density of 1 to 5 J / cm 2 The system uses the thermoelastic effect to vertically excite broadband ultrasonic waves on the surface of duplex steel, with a spot diameter controlled to 50 to 200 μm. The receiving device integrates an ultrasonic angle probe and an oscilloscope to enable synchronous acquisition of contact ultrasonic signals. A moving device drives the trigger and receiving devices along a planned path at a constant speed, completing bilateral line scans of the steel area to be inspected. During the scanning process, the relative position and posture of the trigger and receiving devices must be stable to ensure accurate data collection.

[0101] The data acquired by the receiving device is stored in the initial steel data set. The initial steel data set includes ultrasonic data and the corresponding spatiotemporal tags. Ultrasonic data includes direct wave data and reflected wave data. Direct wave data reflects the information of ultrasonic waves propagating linearly inside the steel. Reflected wave data contains ultrasonic wave information reflected from internal interfaces of the steel, such as defect interfaces and interfaces of different tissue structures. It is key data for analyzing the internal structure and defects of steel. The two are received at different times. Generally, the direct wave is received first, and the reflected wave appears after the direct wave. The spatiotemporal tags accurately record the time and spatial position corresponding to each data point, so that the data can be accurately analyzed and processed later.

[0102] Furthermore, step S2 of this application also includes:

[0103] The symlet function is used to perform wavelet transform on the initial ultrasonic signal, and the optimal decomposition layer number is dynamically determined according to the main frequency bandwidth of the signal to ensure that the sub-band bandwidth corresponding to the highest layer node covers the noise-dominant frequency band;

[0104] The Wpdec function is used to perform full binary tree decomposition on the wavelet transformed signal to generate a wavelet packet coefficient tree. For each terminal node, its energy entropy and power spectral density are calculated, and a node feature matrix is ​​constructed. The terminal nodes are adaptively threshold screened based on the unbiased estimation principle, and the signal of the screened node coefficients is reconstructed to obtain a steel optimization dataset. The steel optimization dataset retains the spatiotemporal labels of the initial steel dataset through label index mapping.

[0105] Specifically, based on the frequency domain characteristics of the ultrasonic signal of duplex steel, the symlet wavelet basis function is preferably used to perform wavelet transform on the initial ultrasonic signal. This type of basis function has approximate symmetry and high-order vanishing moment characteristics, which can effectively match the transient mutation characteristics of the ultrasonic echo signal. The number of decomposition layers is dynamically adjusted according to the main frequency bandwidth of the signal, usually set to 5-7 layers, to ensure that the subband bandwidth corresponding to the highest-level node covers the noise-dominated frequency band while retaining the time-frequency resolution of the defect-sensitive frequency band. The Wpdec function is used to perform a full binary tree decomposition on the wavelet transformed signal to generate a wavelet packet coefficient tree. For each terminal node, its energy entropy and power spectral density are calculated to construct a node feature matrix. Among them, the energy entropy is used to quantify the energy distribution difference between the signal and the noise, and the power spectral density is used to identify the dispersion effect caused by material anisotropy. Based on the Stein unbiased risk estimation criterion, the terminal nodes are adaptively threshold-screened: high-frequency noise nodes: energy entropy is greater than the preset threshold of 0.85 and the PSD shows white noise characteristics in the 20-50MHz frequency band; power frequency interference nodes: energy entropy is concentrated in a specific narrow band, such as the 50Hz fundamental frequency, its harmonics and no overlap with the acoustic signal frequency band; sensor noise nodes: the PSD amplitude is significantly higher than the background noise floor in the unexcited state. Hard threshold processing is performed on the above-mentioned noise-dominated nodes, forcing their coefficients to zero to suppress the noise component. The optimal subband is combined with the reconstructed signal, retaining the low-frequency subband (0.5-10MHz) containing defect-sensitive information and the high-resolution intermediate frequency subband (10~15MHz), and the Wprec function is used to reconstruct the signal of the filtered node coefficients. A boundary extension compensation algorithm is introduced in the reconstruction process to eliminate the edge oscillation distortion caused by wavelet decomposition. The reconstructed signal time-domain waveform must meet the following constraints: the peak amplitude attenuation rate of the defect reflection wave is less than 15%; the time delay difference between the direct wave and the reflected wave is less than 5ns; and the energy retention rate of the signal in the effective frequency band is greater than 90%. The denoising process strictly preserves the spatiotemporal labels of the original dataset, ensuring data consistency through a label index mapping mechanism. During the signal decomposition and reconstruction process, a hash table is used to record the mapping between each waveform segment and its corresponding spatial coordinates. This ultimately results in an optimized steel dataset.

[0106] Furthermore, step S3 of this application also includes:

[0107] Obtain a steel optimization dataset, and map each measurement point in the dataset onto a two-dimensional plane based on its position coordinates in the spatiotemporal label;

[0108] Assign grayscale values ​​to each point according to the ultrasonic echo intensity, use interpolation to connect adjacent measurement points to form continuous lines or images, and construct a B-scan image reflecting the internal cross-sectional structure of the steel;

[0109] The B-scan image is preprocessed by histogram equalization and denoising to transform it into a steel imaging image.

[0110] Specifically, first, for each measurement point in the steel optimization dataset, a point is marked on the two-dimensional plane based on the position identifier in its spatiotemporal label. Ensure that the position of each measurement point on the two-dimensional plane can accurately reflect its actual positional relationship in the steel. Assign a certain grayscale value to each marked point based on the intensity of the ultrasonic echo. The stronger the ultrasonic echo intensity, the higher the assigned grayscale value, for example, closer to white. Conversely, the weaker the echo intensity, the lower the grayscale value, for example, closer to black. Connect adjacent measurement points to form a continuous line or image. Since the measurement points are obtained inside the steel according to a certain scanning path, there is physical continuity between adjacent measurement points. By connecting these points, an image that can reflect the internal structure of the steel can be constructed, namely a B-scan image. In the process of connecting the points, a suitable interpolation method is used to ensure the smoothness of the line, such as linear interpolation or spline interpolation, so that the B-scan image can more realistically reflect the structural changes inside the steel.

[0111] The B-scan image is processed using a histogram equalization algorithm. Histogram equalization is an effective image enhancement technique that aims to redistribute the image's grayscale histogram, making the grayscale distribution more uniform and thereby improving image contrast. The specific operations include: calculating the grayscale histogram of the original image; calculating the cumulative distribution function of the grayscale histogram; scaling the cumulative distribution function to obtain a new grayscale mapping value. This operation aims to map the original grayscale values ​​to a new range, making the grayscale distribution more uniform; and finally, replacing the grayscale value of each pixel in the original image according to the new grayscale mapping value. This algorithm iterates over each pixel in the original image and replaces the original grayscale value, resulting in a histogram-equalized image. After processing, the grayscale differences within that region are increased, improving image contrast. Since B-scan images may contain noise, this noise can affect subsequent image analysis and interpretation. Median filtering is used for denoising. Median filtering is a commonly used nonlinear filtering method that removes isolated noise points such as salt and pepper noise by replacing the value of each pixel with the median value of its neighboring pixels. The preprocessed B-scan image is then converted into an image for further analysis.

[0112] Furthermore, step S4 of this application also includes:

[0113] Use morphological operations and edge detection algorithms to optimize crack edges, extract crack edge contours based on contour tracking algorithms, and determine the number of cracks;

[0114] By calculating the center coordinates of the crack edge contour (x c ,y c ), determine the crack location, the specific formula is: where xi 、y i They represent the horizontal and vertical coordinates of the edge contour points, respectively, and n is the number of contour points;

[0115] The total length of the crack is calculated by superimposing the distance between the coordinates of the crack edge contour points. The specific formula is: Where i is the serial number of the point;

[0116] Calculate the crack position influence factor P according to the specific position of the crack in the duplex steel C , construct the following formula:

[0117] P C =αS R ;

[0118] Among them, P C is the crack position influencing factor, with a value range of 0 <P C <1, S R is the stress concentration coefficient at the crack location, and α is the normalization coefficient used to adjust P C The value range is between 0 and 1;

[0119] According to the crack location and the spatiotemporal labels in the steel optimization dataset, the crack depth is further calculated. The specific formula is: Wherein, Δt is the reception time difference between the direct wave and the reflected wave, and v is the propagation velocity of the ultrasonic wave in the duplex steel;

[0120] A steel crack information set is constructed based on the number of cracks, location influencing factors, length and depth information. The steel crack information set is sorted according to the crack location and collects and integrates detailed information of each crack.

[0121] Specifically, first, morphological operations are performed on the steel image containing cracks. The corrosion operation is performed first, the purpose of which is to remove some small noise points or small interfering structures in the image. The corrosion operation uses a structural element, such as a small rectangular structural element, and uses the structural element as a template for each pixel in the image. When all pixels within the structural element are foreground pixels, the pixel is retained, otherwise it is corroded. Then the expansion operation is performed. The expansion operation is the opposite of the corrosion operation. It can fill some holes caused by the corrosion operation or connect some structures that are disconnected due to corrosion. The expansion operation also uses a structural element. When there is a pixel in the structural element that is a foreground pixel, the pixel is set as a foreground pixel.

[0122] An edge detection algorithm, such as the Canny edge detection algorithm, is used to process images after morphological operations. The Canny edge detection algorithm first applies a Gaussian filter to the image to smooth it and reduce the effects of noise. It then calculates the image's gradient magnitude and direction, uses non-maximum suppression to locate possible edge locations, and finally uses double-threshold detection to determine the true edge. After Canny edge detection, a preliminary edge image of the crack is obtained. A contour tracking algorithm based on edge point search is used to extract the contour edges of all cracks in the steel. For example, starting from an edge point in the image, adjacent edge points are searched in a specific direction. During the search, the next edge point is determined based on edge continuity and directional consistency. When the search returns to the starting point or meets certain termination criteria, such as when the number of edge points found reaches a certain threshold or no more qualifying edge points are found, a contour tracking is completed, and the crack edge contour is extracted. In this way, all crack edges in the steel image are searched and extracted, and the total number of cracks is determined.

[0123] For each extracted crack edge contour, its center coordinate (x c, y c ) to determine the crack location. The specific formula is:

[0124]

[0125] where x i 、y i The horizontal and vertical coordinates of the edge contour coordinate points are represented by , respectively, and n is the number of contour points. The approximate center position of the crack in the image can be obtained through the formula, which is very important for analyzing the position relationship of cracks in duplex steel.

[0126] The total length of the crack is calculated by superimposing the distance between the coordinates of the crack edge contour points. The specific formula is:

[0127]

[0128] Where i is the index of the contour point. The formula is based on the distance formula between two points. By accumulating the distances between adjacent contour points, the total length of the crack can be obtained, which is an important indicator for evaluating crack severity.

[0129] The stress concentration factor SR is calculated based on the crack location. The stress concentration factor is a key parameter that characterizes the local stress of the material. For a crack located in the center, the stress concentration factor SR is calculated as follows: Where a is half the crack length and σ is the nominal stress. The nominal stress refers to the stress calculated on the effective cross section of the specimen without considering geometric discontinuities. This data is calculated from previous measurements and is given by the formula: Where P is the load and A0 is the original cross-sectional area of the specimen. For edge cracks, the stress concentration factor SR is calculated by the formula Calculated as follows.

[0130] The crack position influence factor P is obtained based on the calculated stress concentration factor SR C , and the formula is P C = αS R , where the normalization coefficient α is generally set as SR max which is the maximum value among the stress concentration factors of each crack.

[0131] Based on the crack position and the spatio-temporal tags in the steel optimization dataset, the crack depth is further calculated. The specific formula is: The propagation speed v of ultrasonic waves in duplex steel is approximately 5900 m / s. The direct wave is the wave that directly propagates to the receiving device, and the reflected wave is the wave that reaches the receiving device after being reflected by the crack. Δt is the reception time difference between the direct wave and the reflected wave, and this time difference reflects the propagation time of the ultrasonic wave traveling back and forth through the crack. Since the distance that the ultrasonic wave travels to the crack and then reflects back is twice the crack depth, the final result needs to be divided by two.

[0132] Based on the obtained number of cracks and the position influence factor, length, and depth information of each crack, a steel crack information set is constructed. The steel crack information set is sorted by crack position and汇集整合每条裂纹的详细信息。

[0133] Furthermore, step S5 of this application also includes:

[0134] Based on the number of cracks, crack position influence factor, crack length, and crack depth, a steel quality evaluation model is constructed. The formula is as follows:

[0135]

[0136] Where Q is the steel quality evaluation index, with a range of 0 < Q < 1, n is the total number of cracks, n max is the maximum allowable number of cracks, i is the crack serial number, L Ci is the crack length of the i-th crack, L max is the maximum allowable crack length, D ci is the crack depth of the i-th crack, D max is the maximum allowable crack depth, P Ci is the crack position influence factor of the i-th crack. Cracks in different positions have different influences. w1, w2, w3, and w4 are weight coefficients;

[0137] When Q < 0.5, it is judged as a severe defect and the quality is unqualified;

[0138] When 0.5 ≤ Q < 0.7, it is judged as a moderate defect and the quality is unqualified;

[0139] When 0.7 ≤ Q < 1, it is judged as a minor defect and the quality is qualified;

[0140] When Q is not within the range of 0 - 1, it means that there is a standard exceeding the allowable maximum value, and the steel quality is directly judged as unqualified.

[0141] Specifically, a steel quality evaluation model is constructed based on the crack number, crack position influence factor, crack length, and crack depth. The formula is:

[0142]

[0143] where L Ci is the length of the i-th crack actually detected in the steel, L max is a preset standard value, which represents the maximum allowable crack length when this kind of steel can be used normally or meet certain quality requirements. In the quality inspection of duplex steel, its set value is 10 mm; D ci is the depth of the i-th crack actually penetrating into the interior of the steel, D max is also a preset standard value, representing the maximum depth that the crack can reach when the steel can still work normally or meet the quality requirements. Its value is set to 6 mm; P Ci is the position influence factor of the i-th crack. Cracks in different positions have different influences, and the value has been obtained by the above formula; n is the total number of cracks, n max is the maximum allowable number of cracks, and the value is set to 10; w1, w2, w3, w4 are weight coefficients, which are used to measure the relative importance of crack length, crack depth, crack position influence factor, and crack number when calculating the steel quality evaluation index Q. Here, the values of the three are set as w1 = 0.2, w2 = 0.1, w3 = 0.3, w4 = 0.4; i represents the crack serial number.

[0144] Q calculated by the formula is the steel quality evaluation index, with the range of 0 < Q < 1. When the steel quality evaluation index Q calculated by the formula is less than 0.5, it is determined that the steel has severe defects and does not meet the quality requirements. This means that the crack-related parameters (length, depth, position) of the steel have a very serious impact on the steel quality. Looking at each term in the formula, it may be that the crack length is too large relative to the maximum allowable crack length, or the crack depth is close to or exceeds the maximum allowable crack depth, or the crack is in a very critical position, and under the action of the weight coefficient, the Q value finally falls within this lower range. For example, under the previously assumed values, if the crack situation in the steel changes such that Q = 0.4, according to this judgment criterion, it can be clearly determined that the steel belongs to severe defects, does not meet the quality requirements, may not be able to bear the corresponding structural or functional requirements in actual applications, and there is a relatively high safety risk. When 0.5 <= Q < 0.7, it is judged as moderate defects and the quality is unqualified. In this case, although the crack situation of the steel does not reach the level of severe defects, it still has a non-negligible impact on the steel quality. It may be that factors such as the length, depth, and position of the crack deviate from the ideal state to a certain extent, and after comprehensively considering the weight coefficient, the Q value falls within this range. For example, if Q = 0.6, it means that there are some problems with the steel. Although it may not immediately cause serious structural damage, performance degradation or potential safety hazards may occur during long-term use or under specific conditions, so it is still determined that the quality is unqualified. When 0.7 <= Q < 1, it is judged as minor defects and the quality is qualified. This indicates that the crack situation in the steel is relatively good, and factors such as the crack length, depth, and position have a relatively small impact on the overall steel quality. In this case, although the steel may not be completely perfect, it can basically meet the structural and functional requirements in actual applications, can be used normally, and the safety risk is relatively low. When Q is not within the range of 0 - 1, it means that a certain standard exceeds the allowable maximum value, and the steel quality is directly judged as unqualified.

[0145] Embodiment 2. Based on the same inventive concept as a crack detection method for a duplex steel in the foregoing embodiment, the present application further provides a crack detection system for a duplex steel. Please refer to the attached Figure 2 , the system includes:

[0146] A detection platform building module 11, which is used to build a detection platform for crack detection of duplex steel. The detection platform includes a triggering device, a receiving device, and a moving device. The triggering device is used to vertically emit laser to the surface of the duplex steel to excite broadband ultrasonic waves. The receiving device is used to obtain ultrasonic data. The moving device is used to drive the triggering device and the receiving device to move;

[0147] The data acquisition and optimization module 12 is used to traversally scan the duplex steel through the detection platform to obtain an initial steel data set, perform wavelet packet noise reduction on the initial steel data set, and obtain an optimized steel data set, wherein the optimized steel data set includes ultrasonic data and time and space labels, and the ultrasonic data includes direct wave data and reflected wave data;

[0148] An imaging image acquisition module 13 is used to draw a B-scan image based on the steel material optimization data set, convert the B-scan image into an imaging image, and obtain a steel material imaging image;

[0149] The crack information extraction module 14 is used to process the steel image using an image processing algorithm, extract the crack edge contour, determine the number of cracks, determine the crack position and length by analyzing the coordinate information of the crack edge contour, calculate the crack position influencing factor and crack depth based on the crack position, and construct a steel crack information set based on the crack number, position influencing factor, length, and depth information;

[0150] The quality evaluation model construction module 15 is used to establish a scoring system based on the crack quantity, crack location influencing factor, crack length and crack depth information in the steel crack information set to construct a steel quality evaluation model;

[0151] The quality assessment and reporting module 16 is used to import the steel crack information set into the steel quality evaluation model, obtain the quality assessment grade of the inspected steel, and generate a duplex steel quality assessment report.

[0152] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0153] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present application. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A crack detection method for duplex steel, characterized in that: The method includes: Construct a detection platform for detecting cracks in duplex steel. The detection platform includes a triggering device, a receiving device, and a moving device. The triggering device is used to vertically emit laser onto the surface of duplex steel to excite broadband ultrasonic waves. The receiving device is used to obtain ultrasonic data. The moving device is used to drive the triggering device and the receiving device to move. Traversally scan the duplex steel by the detection platform to obtain an initial dataset of the steel. Perform wavelet packet denoising on the initial dataset of the steel to obtain an optimized dataset of the steel. The optimized dataset of the steel includes ultrasonic data and spatio-temporal tags. The ultrasonic data includes direct wave data and reflected wave data. Draw a B-scan image according to the optimized dataset of the steel, convert the B-scan image into an imaging image, and obtain an imaging image of the steel. Use an image processing algorithm to process the imaging image of the steel, extract the crack edge contour, determine the number of cracks. By analyzing the coordinate information of the crack edge contour, determine the crack position and length. Based on the crack position, calculate the crack position influence factor and crack depth. Construct a crack information set of the steel according to the number of cracks, position influence factor, length, and depth information. Among them, constructing a crack information set of the steel according to the number of cracks, position influence factor, length, and depth information includes: Optimize the crack edge using morphological operations and edge detection algorithms, extract the crack edge contour based on the contour tracking algorithm, and determine the number of cracks. By calculating the center coordinates of the crack edge contour (x c ,y c ), determine the crack location, the specific formula is: where x i 、y i They represent the horizontal and vertical coordinates of the edge contour points, respectively, and n is the number of contour points; The total length of the crack is calculated by superimposing the distance between the coordinates of the crack edge contour points. The specific formula is: Where i is the serial number of the point; Calculate the crack position influence factor P according to the specific position of the crack in the duplex steel C , construct the following formula: P C =αS R ; Among them, P C is the crack position influencing factor, with a value range of 0 <P C <1, S R is the stress concentration coefficient at the crack location, and α is the normalization coefficient used to adjust P C The value range is between 0 and 1; According to the crack location and the spatiotemporal labels in the steel optimization dataset, the crack depth is further calculated. The specific formula is: Wherein, Δt is the reception time difference between the direct wave and the reflected wave, and v is the propagation velocity of the ultrasonic wave in the duplex steel; Construct a crack information set of the steel based on the number of cracks, position influence factor, length, and depth information. The crack information set of the steel is sorted according to the crack position, and the detailed information of each crack is collected and integrated. Combine the number of cracks, crack position influence factor, crack length, and crack depth information in the crack information set of the steel to establish a scoring system and construct a quality evaluation model for the steel. Among them, constructing a quality evaluation model for the steel includes: Construct a quality evaluation model for the steel based on the number of cracks, crack position influence factor, crack length, and crack depth. The formula is as follows: Among them, Q is the steel quality evaluation index, with the range of 0 < Q < 1, n is the total number of cracks, n max is the maximum allowable number of cracks, i is the crack serial number, L Ci is the crack length of the i-th crack, L max is the maximum allowable crack length, D ci is the crack depth of the i-th crack, D max is the maximum allowable crack depth, P Ci is the crack position influence factor of the i-th crack. Cracks in different positions have different influences, and w1, w2, w3, and w4 are weight coefficients; When Q < 0.5, it is judged as a severe defect and the quality is unqualified. When 0.5 <= Q < 0.7, it is judged as a moderate defect and the quality is unqualified. When 0.7 <= Q < 1, it is judged as a minor defect and the quality is qualified. When Q is not in the range of 0 - 1, it means that there is a standard exceeding the allowable maximum value, and the quality of the steel is directly judged as unqualified. Import the crack information set of the steel into the quality evaluation model of the steel, obtain the quality rating level of the inspected steel, and generate a quality assessment report for the duplex steel.

2. The crack detection method for dual-phase steel according to claim 1, characterized in that: Obtain an initial dataset of the steel, including: Plan the scanning path on the surface of the steel, adopt a rasterized full-coverage scanning mode, dynamically optimize the scanning spacing according to the spot size, and enable an adaptive encryption scanning strategy for the edge area. The triggering device vertically irradiates the surface of the steel with pulsed laser, generates broadband ultrasonic waves through the thermoelastic effect. The receiving device uses an ultrasonic oblique probe and an oscilloscope to achieve synchronous acquisition of contact ultrasonic signals. The moving device drives the triggering device and the receiving device to move uniformly along the planned path to complete the bilateral line scanning of the area to be detected of the steel. The receiving device acquires data into an initial steel data set, where the initial steel data set includes ultrasonic data and time-space labels corresponding to the data, and the ultrasonic data includes direct wave data and reflected wave data.

3. The crack detection method for dual-phase steel according to claim 1, characterized in that: Access to steel optimization datasets, including: The symlet function is used to perform wavelet transform on the initial ultrasonic signal, and the optimal decomposition layer number is dynamically determined according to the main frequency bandwidth of the signal to ensure that the sub-band bandwidth corresponding to the highest layer node covers the noise-dominant frequency band; The Wpdec function is used to perform full binary tree decomposition on the wavelet transformed signal to generate a wavelet packet coefficient tree. For each terminal node, its energy entropy and power spectral density are calculated, and a node feature matrix is ​​constructed. The terminal nodes are adaptively threshold screened based on the unbiased estimation principle, and the signal of the screened node coefficients is reconstructed to obtain a steel optimization dataset. The steel optimization dataset retains the spatiotemporal labels of the initial steel dataset through label index mapping.

4. The crack detection method for dual-phase steel according to claim 1, characterized in that: Obtain steel material imaging, including: Obtain a steel optimization dataset, and map each measurement point in the dataset onto a two-dimensional plane based on its position coordinates in the spatiotemporal label; Assign grayscale values ​​to each point according to the ultrasonic echo intensity, use interpolation to connect adjacent measurement points to form continuous lines or images, and construct a B-scan image reflecting the internal cross-sectional structure of the steel; The B-scan image is preprocessed by histogram equalization and denoising to transform it into a steel imaging image.

5. A crack detection system for dual-phase steel, characterized in that: The system is used to implement the crack detection method for dual-phase steel according to any one of claims 1 to 4, and the system comprises: A detection platform construction module is used to build a detection platform for dual-phase steel crack detection. The detection platform includes a trigger device, a receiving device, and a moving device. The trigger device is used to emit laser vertically to the surface of the dual-phase steel to stimulate broadband ultrasound. The receiving device is used to obtain ultrasonic data. The moving device is used to drive the trigger device and the receiving device to move. A data acquisition and optimization module is used to traversally scan the duplex steel through the detection platform to obtain an initial steel data set, perform wavelet packet noise reduction on the initial steel data set, and obtain an optimized steel data set, wherein the optimized steel data set includes ultrasonic data and time and space labels, and the ultrasonic data includes direct wave data and reflected wave data; An imaging image acquisition module is used to draw a B-scan image based on the steel optimization data set, convert the B-scan image into an imaging image, and obtain a steel imaging image; The crack information extraction module is used to process the steel imaging image using an image processing algorithm, extract the crack edge contour, determine the number of cracks, determine the crack position and length by analyzing the coordinate information of the crack edge contour, calculate the crack position influencing factor and crack depth based on the crack position, and construct a steel crack information set based on the crack number, position influencing factor, length, and depth information. The steel crack information set constructed based on the crack number, position influencing factor, length, and depth information includes: Optimize the crack edge using morphological operations and edge detection algorithms, extract the crack edge contour based on the contour tracking algorithm, and determine the number of cracks; By calculating the center coordinates of the crack edge contour (x c ,y c ), determine the crack location, the specific formula is: where x i 、y i They represent the horizontal and vertical coordinates of the edge contour points, respectively, and n is the number of contour points; The total length of the crack is calculated by superimposing the distance between the coordinates of the crack edge contour points. The specific formula is: Where i is the serial number of the point; Calculate the crack position influence factor P according to the specific position of the crack in the duplex steel C , construct the following formula: P C =αS R ; Among them, P C is the crack position influencing factor, with a value range of 0 <P C <1, S R is the stress concentration coefficient at the crack location, and α is the normalization coefficient used to adjust P C The value range is between 0 and 1; According to the crack location and the spatiotemporal labels in the steel optimization dataset, the crack depth is further calculated. The specific formula is: Wherein, Δt is the reception time difference between the direct wave and the reflected wave, and v is the propagation velocity of the ultrasonic wave in the duplex steel; Construct a steel crack information set based on the number of cracks, position influence factors, length, and depth information. The steel crack information set is sorted by crack position and integrates the detailed information of each crack; Quality evaluation model construction module. The quality evaluation model construction module is used to establish a scoring system and construct a steel quality evaluation model by combining the number of cracks, crack position influence factors, crack length, and crack depth information in the steel crack information set. Among them, constructing the steel quality evaluation model includes: Construct a steel quality evaluation model based on the number of cracks, crack position influence factors, crack length, and crack depth. The formula is as follows: Among them, Q is the steel quality evaluation index, with the range of 0 < Q < 1, n is the total number of cracks, n max is the maximum allowable number of cracks, i is the crack serial number, L Ci is the crack length of the i-th crack, L max is the maximum allowable crack length, D ci is the crack depth of the i-th crack, D max is the maximum allowable crack depth, P Ci is the crack position influence factor of the i-th crack. Cracks at different positions have different influences. w1, w2, w3, and w4 are weight coefficients; When Q < 0.5, it is judged as a severe defect and the quality is unqualified; When 0.5 ≤ Q < 0.7, it is judged as a moderate defect and the quality is unqualified; When 0.7 ≤ Q < 1, it is judged as a minor defect and the quality is qualified; When Q is not in the range of 0 - 1, it means that there is a standard exceeding the allowable maximum value, and the steel quality is directly judged as unqualified; Quality assessment and reporting module. The quality assessment and reporting module is used to import the steel crack information set into the steel quality evaluation model, obtain the quality assessment grade of the inspected steel, and generate a quality assessment report for duplex steel.

Images