Magnesia carbon brick automatic detection method based on laser sensor

Abstract

The invention discloses a magnesia carbon brick automatic detection method based on a laser sensor, and the method comprises the following steps: A1, carrying out the multi-angle scanning of the surface of a magnesia carbon brick through a line laser scanner, and generating high-precision three-dimensional point cloud data; a2, inhibiting graphite reflection interference on the surface of the magnesia carbon brick by adopting a polarization filtering system, and compensating a temperature drift error in real time by fusing an infrared thermometer; and step A3, preprocessing the point cloud data through GPU accelerated parallel computing. The invention relates to the technical field of magnesia carbon brick detection. According to the magnesia carbon brick automatic detection method based on the laser sensor, a polarization extinction technology and a multi-source interference compensation mechanism are fused, the problem of high-reflection surface data acquisition distortion is solved, the point cloud integrity is ensured, interference factors such as production line vibration and dust are actively inhibited, and the detection stability is ensured.

Description

Technical Field

[0001] This invention relates to the field of magnesia-carbon brick detection technology, specifically to an automatic detection method for magnesia-carbon bricks based on a laser sensor. Background Technology

[0002] In the iron and steel metallurgy industry, magnesia-carbon bricks, as key refractory materials for steelmaking furnace linings, have surface defects and dimensional accuracy that directly affect the safe service life of steel ladles. Traditional manual inspection relies on visual observation and contact measuring tools, facing severe technical bottlenecks: on the one hand, the graphite layer on the surface of the high-temperature sintered bricks produces strong specular reflection, leading to distortion of the point cloud in conventional laser scanning; on the other hand, dimensional drift caused by thermal expansion is high, while manual inspection requires cooling to below 80°C, taking more than 7 minutes per brick and resulting in a microcrack miss rate of >5.2%. Existing industry standards (YB / T 4116-2003) do not establish dynamic defect thresholds; for example, a uniform allowable value is used for the slag line area and non-working areas, causing the false positive rate for qualified products to rise to 8.3%.

[0003] Current technical improvement solutions have significant drawbacks: multi-view point cloud registration relies on traditional ICP algorithms, and processing a single brick takes over 30 seconds. Therefore, there is an urgent need to develop an automatic detection method that integrates anti-interference scanning, real-time temperature compensation, and dynamic decision-making to overcome the challenge of accurate quantitative detection of highly reflective materials under hot conditions. Summary of the Invention

[0004] To achieve the above objectives, the present invention provides the following technical solution: an automatic detection method for magnesia-carbon bricks based on a laser sensor, comprising the following steps:

[0005] Step A1: Scan the surface of the magnesia-carbon brick from multiple angles using a line laser scanner to generate high-precision three-dimensional point cloud data with a scanning resolution ≤0.1mm;

[0006] Step A2: Use a polarization filter system to suppress the interference of graphite reflection on the surface of magnesium carbon bricks, and integrate an infrared thermometer to compensate for temperature drift error in real time.

[0007] Step A3: Preprocess the point cloud data using GPU-accelerated parallel computing. The preprocessing includes outlier removal, Gaussian filtering for noise reduction, multi-view point cloud registration and stitching, and construction of a complete 3D model.

[0008] Step A4: Perform defect detection.

[0009] Surface defect identification: Input point cloud data into a pre-trained U-Net deep learning model to segment cracks, pores and spalling areas;

[0010] Dimensional accuracy calculation: By comparing the point cloud with the CAD model through an iterative nearest point algorithm, the length, width, height, flatness, and verticality deviations are dynamically calculated;

[0011] By combining a support vector machine classifier to classify defect types, the sorting mechanism is triggered in real time when the size deviation is greater than ±0.05mm or the defect area exceeds the threshold.

[0012] By integrating polarization extinction technology with a multi-source interference compensation mechanism, the problem of data acquisition distortion on highly reflective surfaces is solved, ensuring point cloud integrity, actively suppressing interference factors such as production line vibration and dust, and guaranteeing detection stability.

[0013] Preferably, the line laser scanner in step A1 is completed by a six-axis robotic arm equipped with a laser sensor, and the motion path planning of the six-axis robotic arm specifically includes the following steps:

[0014] Step S1: Generate a three-dimensional voxel mesh based on the CAD model of magnesia-carbon bricks;

[0015] Step S2: Calculate the collision-free scan trajectory using the fast extended random tree algorithm;

[0016] Step S3: The scanning axis of the laser sensor always maintains an angle of 85-90 degrees with the normal vector of the magnesia-carbon brick surface to avoid data distortion.

[0017] A parallel accelerated computing architecture and intelligent path planning algorithm are used to achieve high-speed scanning and processing of the entire surface of magnesia-carbon bricks. The detection time is reduced compared to manual operation.

[0018] Preferably, the implementation of the polarization filtering system in step A2 specifically includes the following steps:

[0019] Step D1: The laser emitting end integrates a wavelength laser and a wire grid polarizer, with the polarization direction at 0°;

[0020] Step D2: Set up an analyzer at the receiving end with a polarization direction of 60±2°, and its polarization direction is orthogonal to the principal crystal orientation of the graphite layer on the surface of the magnesia-carbon brick.

[0021] Step D3: The polarizer operates at a temperature range of -20℃ to 150℃, with a transmittance of ≥90%.

[0022] Preferably, the point cloud registration and stitching in step A3 specifically involves:

[0023] A fast point feature histogram descriptor with dimension 33 was extracted from the point cloud. A random sampling algorithm was used for initial matching, with ≥1000 iterations. The data was then fine-tuned by nonlinear optimization of the transformation matrix, and the final registration error was ≤0.02mm.

[0024] Preferably, the transfer learning optimization of the U-Net deep learning model in step A4 includes:

[0025] The base model is initialized on ImageNet pre-trained weights;

[0026] Synthetic data augmentation was applied to the real point cloud: Gaussian noise σ = 0.05 mm, random affine transformation rotation ±5°, and translation ±2 mm were applied.

[0027] The training samples contain a high-temperature point cloud dataset ranging from 200℃ to 600℃, with temperature labels annotated by an infrared thermometer.

[0028] Preferably, the threshold in step A4 is a dynamic matrix threshold, which is set differently based on the service location and defect type of the magnesia-carbon brick. The specific setting type is as follows:

[0029] Crack-type defects: maximum allowable length ≤ 3mm, and the total area of ​​branches ≤ 0.5% of the brick surface area;

[0030] Porosity or spalling defects: Individual defect area ≤ 4mm 2 The total defect area per unit area is ≤1.5%;

[0031] For brickwork working surfaces: the total defect area ratio is ≤0.8%, and for non-working surfaces it is ≤2.0%.

[0032] Preferably, the dynamic matrix threshold is calculated as follows:

[0033]

[0034] Wherein, the T 温度 The A value is the real-time temperature value from the infrared thermometer. 砖 The actual surface area of ​​the current test surface of the magnesia-carbon brick, the L 服役 The position coefficient of the brick in the ladle.

[0035] Through dynamic adaptive threshold technology, the system intelligently adjusts the judgment criteria based on the real-time temperature of the magnesia-carbon brick, its service location, and the type of defect. In high-temperature environments, it effectively suppresses false positives caused by thermal expansion, significantly reducing the false detection rate and improving the ability to detect surface cracks.

[0036] This invention provides an automatic detection method for magnesia-carbon bricks based on a laser sensor. It has the following beneficial effects:

[0037] (I) This laser sensor-based automatic detection method for magnesia-carbon bricks utilizes dynamic adaptive threshold technology. The system intelligently adjusts the judgment criteria based on the real-time temperature, service location, and defect type of the magnesia-carbon bricks. In high-temperature environments, it effectively suppresses false detections caused by thermal expansion, significantly reducing the false detection rate and improving the ability to detect surface cracks.

[0038] (II) This laser sensor-based automatic detection method for magnesia-carbon bricks achieves high-speed scanning and processing of the entire surface of magnesia-carbon bricks by employing a parallel accelerated computing architecture and intelligent path planning algorithm. The detection time is significantly reduced compared to manual operation.

[0039] (III) The automatic detection method for magnesium carbon bricks based on laser sensors solves the problem of data acquisition distortion on highly reflective surfaces by integrating polarization extinction technology and multi-source interference compensation mechanism, ensuring the integrity of point cloud, actively suppressing interference factors such as production line vibration and dust, and ensuring detection stability. Attached Figure Description

[0040] Figure 1 This is a flowchart illustrating the operation of the automatic detection method for magnesia-carbon bricks based on a laser sensor according to the present invention. Detailed Implementation

[0041] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0042] Example 1, please refer to 1, the present invention provides a technical solution:

[0043] An automatic detection method for magnesia-carbon bricks based on a laser sensor includes the following steps:

[0044] Step A1: Scan the surface of the magnesia-carbon brick from multiple angles using a line laser scanner to generate high-precision three-dimensional point cloud data with a scanning resolution ≤0.1mm;

[0045] The line laser scanner is operated by a six-axis robotic arm equipped with a laser sensor. The motion path planning of the six-axis robotic arm specifically includes the following steps:

[0046] Step S1: Generate a three-dimensional voxel mesh based on the CAD model of magnesia-carbon bricks;

[0047] Step S2: Calculate the collision-free scan trajectory using the fast extended random tree algorithm;

[0048] Step S3: The scanning axis of the laser sensor always maintains an angle of 85-90 degrees with the normal vector of the magnesia-carbon brick surface to avoid data distortion.

[0049] A parallel accelerated computing architecture and intelligent path planning algorithm are used to achieve high-speed scanning and processing of the entire surface of magnesia-carbon bricks. The detection time is reduced compared to manual operation.

[0050] Step A2: Use a polarization filter system to suppress the interference of graphite reflection on the surface of magnesium carbon bricks, and integrate an infrared thermometer to compensate for temperature drift error in real time.

[0051] The specific implementation of a polarization filtering system includes the following steps:

[0052] Step D1: The laser emitting end integrates a wavelength laser and a wire grid polarizer, with the polarization direction at 0°;

[0053] Step D2: Set up an analyzer at the receiving end with a polarization direction of 60°, and its polarization direction is orthogonal to the principal crystal orientation of the graphite layer on the surface of the magnesia-carbon brick.

[0054] Step D3: The polarizer operates at a temperature range of 75℃ and has a transmittance of 90%.

[0055] Step A3: Preprocess the point cloud data using GPU-accelerated parallel computing. The preprocessing includes outlier removal, Gaussian filtering for noise reduction, multi-view point cloud registration and stitching, and construction of a complete 3D model.

[0056] Point cloud registration and stitching are specifically as follows:

[0057] A fast point feature histogram descriptor with dimension 33 was extracted from the point cloud. A random sampling algorithm was used for initial matching, with ≥1000 iterations. The data was then fine-tuned by nonlinear optimization of the transformation matrix, and the final registration error was ≤0.02mm.

[0058] Step A4: Perform defect detection.

[0059] Surface defect identification: Input point cloud data into a pre-trained U-Net deep learning model to segment cracks, pores and spalling areas;

[0060] Transfer learning optimizations for the U-Net deep learning model include:

[0061] The base model is initialized on ImageNet pre-trained weights;

[0062] Synthetic data augmentation was applied to the real point cloud: Gaussian noise σ = 0.05 mm, random affine transformation rotation ±5°, and translation ±2 mm were applied.

[0063] The training samples include a 400℃ high-temperature point cloud dataset, labeled with temperature tags by an infrared thermometer;

[0064] Dimensional accuracy calculation: By comparing the point cloud with the CAD model through an iterative nearest point algorithm, the length, width, height, flatness, and verticality deviations are dynamically calculated;

[0065] By combining a support vector machine classifier to classify defect types, the sorting mechanism is triggered in real time when the size deviation is greater than ±0.05mm or the defect area exceeds the threshold.

[0066] The threshold is specifically a dynamic matrix threshold, which is set differently based on the service location and defect type of the magnesia-carbon brick. Its setting type is as follows:

[0067] Crack-type defects: maximum allowable length ≤ 3mm, and the total area of ​​branches ≤ 0.5% of the brick surface area;

[0068] Porosity or spalling defects: Individual defect area ≤ 4mm 2 The total defect area per unit area is ≤1.5%;

[0069] For brickwork working surfaces: the total defect area ratio is ≤0.8%, and for non-working surfaces it is ≤2.0%.

[0070] The dynamic matrix threshold is calculated as follows:

[0071]

[0072] Wherein, the T 温度 The A value is the real-time temperature value from the infrared thermometer. 砖 The actual surface area of ​​the current test surface of the magnesia-carbon brick, the L 服役 The position coefficient of the brick in the ladle.

[0073] Through dynamic adaptive threshold technology, the system intelligently adjusts the judgment criteria based on the real-time temperature of the magnesia-carbon brick, its service location, and the type of defect. In high-temperature environments, it effectively suppresses false positives caused by thermal expansion, significantly reducing the false detection rate and improving the ability to detect surface cracks.

[0074] By integrating polarization extinction technology with a multi-source interference compensation mechanism, the problem of data acquisition distortion on highly reflective surfaces is solved, ensuring point cloud integrity, actively suppressing interference factors such as production line vibration and dust, and guaranteeing detection stability.

[0075] Example 2, please refer to 1, the present invention provides a technical solution:

[0076] An automatic detection method for magnesia-carbon bricks based on a laser sensor includes the following steps:

[0077] Step A1: Scan the surface of the magnesia-carbon brick from multiple angles using a line laser scanner to generate high-precision three-dimensional point cloud data with a scanning resolution ≤0.1mm;

[0078] The line laser scanner is operated by a six-axis robotic arm equipped with a laser sensor. The motion path planning of the six-axis robotic arm specifically includes the following steps:

[0079] Step S1: Generate a three-dimensional voxel mesh based on the CAD model of magnesia-carbon bricks;

[0080] Step S2: Calculate the collision-free scan trajectory using the fast extended random tree algorithm;

[0081] Step S3: The scanning axis of the laser sensor always maintains an angle of 85-90 degrees with the normal vector of the magnesia-carbon brick surface to avoid data distortion.

[0082] A parallel accelerated computing architecture and intelligent path planning algorithm are used to achieve high-speed scanning and processing of the entire surface of magnesia-carbon bricks. The detection time is reduced compared to manual operation.

[0083] Step A2: Use a polarization filter system to suppress the interference of graphite reflection on the surface of magnesium carbon bricks, and integrate an infrared thermometer to compensate for temperature drift error in real time.

[0084] The specific implementation of a polarization filtering system includes the following steps:

[0085] Step D1: The laser emitting end integrates a wavelength laser and a wire grid polarizer, with the polarization direction at 0°;

[0086] Step D2: Set up an analyzer at the receiving end with a polarization direction of 62°, which is orthogonal to the principal crystal orientation of the graphite layer on the surface of the magnesia-carbon brick.

[0087] Step D3: The polarizer operates at a temperature range of 150℃ and has a transmittance of ≥90%.

[0088] Step A3: Preprocess the point cloud data using GPU-accelerated parallel computing. The preprocessing includes outlier removal, Gaussian filtering for noise reduction, multi-view point cloud registration and stitching, and construction of a complete 3D model.

[0089] Point cloud registration and stitching are specifically as follows:

[0090] A fast point feature histogram descriptor with dimension 33 was extracted from the point cloud. A random sampling algorithm was used for initial matching, with ≥1000 iterations. The data was then fine-tuned by nonlinear optimization of the transformation matrix, and the final registration error was ≤0.02mm.

[0091] Step A4: Perform defect detection.

[0092] Surface defect identification: Input point cloud data into a pre-trained U-Net deep learning model to segment cracks, pores and spalling areas;

[0093] Transfer learning optimizations for the U-Net deep learning model include:

[0094] The base model is initialized on ImageNet pre-trained weights;

[0095] Synthetic data augmentation was applied to the real point cloud: Gaussian noise σ = 0.05 mm, random affine transformation rotation ±5°, and translation ±2 mm were applied.

[0096] The training samples include a 600℃ high-temperature point cloud dataset, labeled with temperature tags by an infrared thermometer;

[0097] Dimensional accuracy calculation: By comparing the point cloud with the CAD model through an iterative nearest point algorithm, the length, width, height, flatness, and verticality deviations are dynamically calculated;

[0098] By combining a support vector machine classifier to classify defect types, the sorting mechanism is triggered in real time when the size deviation is greater than ±0.05mm or the defect area exceeds the threshold.

[0099] The threshold is specifically a dynamic matrix threshold, which is set differently based on the service location and defect type of the magnesia-carbon brick. Its setting type is as follows:

[0100] Crack-type defects: maximum allowable length ≤ 3mm, and the total area of ​​branches ≤ 0.5% of the brick surface area;

[0101] Porosity or spalling defects: Individual defect area ≤ 4mm 2 The total defect area per unit area is ≤1.5%;

[0102] For brickwork working surfaces: the total defect area ratio is ≤0.8%, and for non-working surfaces it is ≤2.0%.

[0103] The dynamic matrix threshold is calculated as follows:

[0104]

[0105] Wherein, the T 温度 The A value is the real-time temperature value from the infrared thermometer. 砖 The actual surface area of ​​the current test surface of the magnesia-carbon brick, the L 服役 The position coefficient of the brick in the ladle.

[0106] Through dynamic adaptive threshold technology, the system intelligently adjusts the judgment criteria based on the real-time temperature of the magnesia-carbon brick, its service location, and the type of defect. In high-temperature environments, it effectively suppresses false positives caused by thermal expansion, significantly reducing the false detection rate and improving the ability to detect surface cracks.

[0107] By integrating polarization extinction technology with a multi-source interference compensation mechanism, the problem of data acquisition distortion on highly reflective surfaces is solved, ensuring point cloud integrity, actively suppressing interference factors such as production line vibration and dust, and guaranteeing detection stability.

[0108] Threshold setting principle;

[0109]

[0110] During operation, the operator places the magnesia-carbon bricks to be inspected on the testing platform. The system automatically starts a six-axis robotic arm to preheat the equipment and calibrate the infrared thermometer at zero point. Simultaneously, it loads the CAD model and dynamic threshold parameters of the batch of bricks. Based on the CAD model, the robotic arm generates a 0.1mm precision three-dimensional voxel mesh and plans seven collision-free scanning paths using a fast-expanding random tree algorithm. The laser sensor moves at a speed of 0.1m / s, ensuring that the scanning axis maintains an angle of 87°±2° with the brick surface normal to avoid data distortion.

[0111] A line laser scanner scans the brick surface from multiple angles, while a polarization filtering system is activated: a 0° polarizer integrated into the laser emitter suppresses specular reflection, a 58° analyzer at the receiver filters stray graphite light, and an infrared thermometer monitors the surface temperature in real time and compensates for thermal expansion coordinate drift, generating 3D point cloud data with a resolution ≤0.1mm. After scanning, the system preprocesses the point cloud using GPU parallel computing: outliers are first removed by three times the standard deviation, then noise is reduced using a 5×5 Gaussian filter kernel, followed by the extraction of a 33-dimensional fast point feature histogram descriptor. Initial registration of the multi-view point cloud is performed through 1200 iterations of a random sampling consensus algorithm, and finally, nonlinear optimization compresses the registration error to within 0.018mm, outputting a complete 3D model.

[0112] In the defect detection stage, the preprocessed point cloud is converted into a 2D depth map and input into a pre-trained U-Net model (based on ImageNet weight initialization and using synthetic data augmentation: adding σ=0.05mm Gaussian noise and ±5° rotation). The model segments defect areas such as cracks and pores. Simultaneously, the nearest-point algorithm is used to compare the point cloud with the CAD model, dynamically calculating dimensional deviations. The system calls the dynamic matrix thresholding module, combining infrared thermometry values, the actual surface area of ​​the brick, and location coefficients to calculate the current allowable defect limits: crack length threshold ≤ 3.0mm, total pore area threshold ≤ 12mm. 2 The flatness threshold is ≤0.05mm / m. The support vector machine classifier determines defects based on the measured values, triggers the pneumatic pusher to sort the bricks to the product area, and generates a JSON format report containing temperature, dimensional deviation, defect distribution, and judgment results.

[0113] Experimental verification data;

[0114] Test conditions Crack detection threshold Pore ​​area threshold False positive rate Normal temperature (25℃) 0.5% surface area 1.5% local area 0.3％ High temperature (600℃) 0.7% surface area 2.1% local area 0.9％ Traditional fixed threshold in the industry 1.0% surface area 3.0% local area 4.2％

[0115] Conclusion: The dynamic threshold scheme reduces the false judgment rate by 78% compared with the fixed threshold under high temperature conditions.

[0116] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0117] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An automatic detection method for magnesia-carbon bricks based on a laser sensor, characterized in that: Includes the following steps: Step A1: Scan the surface of the magnesia-carbon brick from multiple angles using a line laser scanner to generate high-precision three-dimensional point cloud data; Step A2: Use a polarization filter system to suppress the interference of graphite reflection on the surface of magnesium carbon bricks, and integrate an infrared thermometer to compensate for temperature drift error in real time. Step A3: Preprocess the point cloud data using GPU-accelerated parallel computing. The preprocessing includes outlier removal, Gaussian filtering for noise reduction, multi-view point cloud registration and stitching, and construction of a complete 3D model. Step A4: Perform defect detection. Surface defect identification: Input point cloud data into a pre-trained U-Net deep learning model to segment cracks, pores and spalling areas; Dimensional accuracy calculation: By comparing the point cloud with the CAD model through an iterative nearest point algorithm, the length, width, height, flatness, and verticality deviations are dynamically calculated; By combining a support vector machine classifier to classify defect types, the sorting mechanism is triggered in real time when the size deviation is greater than ±0.05mm or the defect area exceeds the threshold.

2. The automatic detection method for magnesia-carbon bricks based on a laser sensor according to claim 1, characterized in that: The line laser scanner in step A1 is completed by a six-axis robotic arm equipped with a laser sensor. The motion path planning of the six-axis robotic arm specifically includes the following steps: Step S1: Generate a three-dimensional voxel mesh based on the CAD model of magnesia-carbon bricks; Step S2: Calculate the collision-free scan trajectory using the fast extended random tree algorithm; Step S3: The scanning axis of the laser sensor always maintains an angle of 85-90 degrees with the normal vector of the magnesium carbon brick surface.

3. The automatic detection method for magnesia-carbon bricks based on a laser sensor according to claim 1, characterized in that: The implementation of the polarization filtering system in step A2 specifically includes the following steps: Step D1: The laser emitting end integrates a wavelength laser and a wire grid polarizer, with a polarization direction of 0°; Step D2: Set up an analyzer at the receiving end with a polarization direction of 60±2°, and its polarization direction is orthogonal to the principal crystal orientation of the graphite layer on the surface of the magnesia-carbon brick. Step D3: The polarizer operates at a temperature range of -20℃ to 150℃, with a transmittance of ≥90%.

4. The automatic detection method for magnesia-carbon bricks based on a laser sensor according to claim 1, characterized in that: The point cloud registration and stitching in step A3 are specifically as follows: A fast point feature histogram descriptor with dimension 33 is extracted from the point cloud. A random sampling algorithm is used for initial matching, with ≥1000 iterations. The data is then fine-tuned by nonlinear optimization of the transformation matrix.

5. The automatic detection method for magnesia-carbon bricks based on a laser sensor according to claim 1, characterized in that: The transfer learning optimization of the U-Net deep learning model in step A4 includes: The base model is initialized on ImageNet pre-trained weights; Using synthetic data augmentation: applying Gaussian noise and random affine transformation rotation to the real point cloud; The training samples contain a high-temperature point cloud dataset ranging from 200℃ to 600℃, with temperature labels annotated by an infrared thermometer.

6. The automatic detection method for magnesia-carbon bricks based on a laser sensor according to claim 5, characterized in that: In step A4, the threshold is specifically a dynamic matrix threshold, which is set differently based on the service location and defect type of the magnesia-carbon brick. The specific setting type is as follows: Crack-type defects: maximum allowable length ≤ 3mm, and the total area of ​​branches ≤ 0.5% of the brick surface area; Porosity or spalling defects: Individual defect area ≤ 4mm 2 The total defect area per unit area is ≤1.5%; For brickwork working surfaces: the total defect area ratio is ≤0.8%, and for non-working surfaces it is ≤2.0%.

7. The automatic detection method for magnesia-carbon bricks based on a laser sensor according to claim 6, characterized in that: The dynamic matrix threshold is calculated as follows: Wherein, the T 温度 The A value is the real-time temperature value from the infrared thermometer. 砖 The actual surface area of ​​the current test surface of the magnesia-carbon brick, the L 服役 The position coefficient of the brick in the ladle.

Images