Intelligent detection method for refractory brick production line

Abstract

The invention relates to the technical field of refractory materials, in particular to an intelligent detection method for a refractory brick production line, which comprises the following steps of: detecting a refractory brick rough blank after a firing process to obtain brick body information and an environment definition index; calculating a feature defect index representing the severity of the defect, a feature pose index representing the risk of the defect position and a feature confidence coefficient representing the reliability of the data, comparing the feature confidence coefficient with a preset threshold value, judging the reliability of the data, if the data is reliable, determining a cooling operation execution strategy, and if the data is not reliable, executing the cooling operation strategy. Determining a final cooling parameter according to a comparison result of the feature defect index and the feature pose index; and if the data is not reliable, a detection optimization strategy is determined to be executed, and the detection system carries out self-optimization and then carries out detection so as to obtain reliable brick body information. According to the method, the quality consistency, the yield and the intelligent level of a production line of refractory brick production are ensured by executing pre-evaluation of data reliability and precise regulation and control of cooling parameters.

Description

Technical Field

[0001] This invention relates to the field of refractory materials technology, and in particular to an intelligent detection method for refractory brick production lines. Background Technology

[0002] With the development of intelligent manufacturing technology, the introduction of automated inspection systems into production lines to replace traditional manual visual inspection has become an industry trend. Existing technical solutions typically focus on using machine vision, laser scanning, and other methods to identify defects and measure dimensions of fired refractory brick blanks. For example, some solutions use image processing algorithms to identify surface cracks and perform simple classification and alarms based on characteristics such as crack length and width; other solutions use 3D scanning to obtain point cloud data of the blanks and compare it with standard CAD models to detect out-of-tolerance dimensions.

[0003] Chinese Patent Publication No. CN115954135A discloses an intelligent control method and system for a refractory brick production line. The relevant technical solution aims to achieve monitoring and control of the production process by dividing the production process, optimizing the process and the detection process, and constructing negative feedback and positive feedback paths.

[0004] However, the above solutions still have the following problems: Existing technical solutions usually assume that the testing equipment operates in a constant ideal environment and directly accept the test results, which can easily lead to misjudgment, missed detection, or inaccurate dimensional measurement of defects. At the same time, the existing methods for defect assessment are mostly limited to the geometric characteristics of the brick itself, and fail to assess the risk of defects in the context of the overall structure and stress of the brick body, and do not form a closed-loop linkage with downstream process detection and control systems such as cooling.

[0005] Therefore, there is an urgent need for an intelligent detection method that can intelligently assess the reliability of the detection data itself and transform multi-dimensional defect information into process adjustment instructions, so as to improve the quality control level and automation of refractory brick production lines. Summary of the Invention

[0006] Therefore, the present invention provides an intelligent detection method for refractory brick production lines to overcome the problems in the prior art where the reliability of detection data cannot be guaranteed due to environmental factors, and where defect assessment does not consider the risk factors of defect location.

[0007] To achieve the above objectives, the present invention provides an intelligent detection method for a refractory brick production line, comprising:

[0008] The rough blanks of refractory bricks are tested to obtain information about the brick body and clarity indicators;

[0009] Based on the brick information, the feature defect index and feature pose index are determined, wherein the brick information includes surface defect information and pose information;

[0010] The feature confidence level is determined based on the sharpness index, wherein the sharpness index includes a first sharpness index and a second sharpness index.

[0011] Based on the comparison result between the stated feature confidence level and the preset feature confidence level, the reliability of the brick information is determined, and a corresponding control strategy is generated based on the determination result.

[0012] If the brick information is determined to be reliable, a cooling operation strategy is determined to be executed. The execution process includes determining whether to adjust the cooling parameters based on the comparison result between the feature defect index and the corresponding preset feature index, and adjusting the cooling parameters based on the comparison result between the feature pose index and the corresponding preset feature index when adjusting the cooling parameters. The cooling parameters include cooling time and cooling rate, and the preset feature index includes a preset defect index and a preset pose index.

[0013] If the brick information is determined to be unreliable, a detection optimization strategy is executed to obtain an optimized feature defect index and an optimized feature pose index. Based on the comparison result between the optimized feature defect index and the preset feature index, it is re-determined whether to adjust the cooling parameters. When adjusting the cooling parameters, the cooling parameters are adjusted based on the comparison result between the optimized feature pose index and the preset feature index.

[0014] Furthermore, the process of determining whether the brick information is reliable based on the comparison result between the feature confidence level and the preset feature confidence level includes:

[0015] If the confidence level of the feature is greater than the preset confidence level of the feature, the brick information is determined to be reliable, and the cooling operation strategy is executed.

[0016] If the feature confidence level is less than or equal to the preset feature confidence level, the brick information is determined to be unreliable, and the detection optimization strategy is executed.

[0017] Furthermore, the process of executing the cooling operation strategy includes:

[0018] If the characteristic defect index is less than or equal to the preset defect index, the initial cooling parameter is determined as the final cooling parameter;

[0019] If the feature defect index is greater than the preset defect index, determine whether to adjust the initial cooling parameters based on the comparison result between the feature pose index and the preset pose index.

[0020] Furthermore, if the feature pose index is less than or equal to the preset pose index, the final cooling parameter is determined to be the initial cooling parameter.

[0021] If the feature pose index is greater than the preset pose index, the initial cooling parameters are adjusted to obtain the final determined cooling parameters.

[0022] Furthermore, the process of adjusting the initial cooling parameters to obtain the final determined cooling parameters includes:

[0023] Calculate the difference between the feature pose index and the preset pose index and record it as the first pose difference value;

[0024] The adjustment range of the initial cooling parameters is determined based on the first position difference value, wherein the extension of the cooling time is positively correlated with the first position difference value, and the reduction of the cooling rate is positively correlated with the first position difference value.

[0025] Furthermore, the process of determining to execute the detection optimization strategy to obtain the optimized feature defect index and the optimized feature pose index includes:

[0026] Based on the determination that the brick information is unreliable, corresponding detection optimization operations are performed. These operations include increasing the number of sensors, switching the feature recognition algorithm, adjusting sensor parameters, or enabling auxiliary lighting.

[0027] The optimized feature defect index and the optimized feature pose index are determined based on the brick information reacquired after the detection optimization operation.

[0028] The cooling operation strategy is executed based on the optimized feature defect index and the optimized feature pose index.

[0029] Furthermore, the process of executing the cooling operation strategy based on the optimized feature defect index and the optimized feature pose index includes:

[0030] If the optimized feature defect index is less than or equal to the preset defect index, the initial cooling parameter is determined as the final cooling parameter;

[0031] If the optimized feature defect index is greater than the preset defect index, determine whether to adjust the initial cooling parameters based on the comparison result between the optimized feature pose index and the preset pose index.

[0032] Furthermore, if the optimized feature pose index is less than or equal to the preset pose index, the final cooling parameter is determined to be the initial cooling parameter.

[0033] If the optimized feature pose index is greater than the preset pose index, the difference between the optimized feature pose index and the preset pose index is calculated and recorded as the second pose difference.

[0034] The adjustment range of the initial cooling parameters is determined based on the second position difference value, wherein the extension of the cooling time is positively correlated with the second position difference value, and the reduction of the cooling rate is positively correlated with the second position difference value.

[0035] Furthermore, the process of determining the feature defect index and feature pose index based on the brick information includes:

[0036] The inspection system includes a first vision inspection mechanism and a second vision inspection mechanism;

[0037] The surface defect information is obtained through the first visual inspection mechanism, wherein the surface defect information includes at least one of crack defect data or hole defect data;

[0038] The pose information is obtained through the second visual detection mechanism, wherein the pose information includes at least the position information of the crack defect data or the hole defect data on the refractory brick blank.

[0039] Furthermore, the process of determining feature confidence based on the sharpness index includes:

[0040] The feature confidence level is determined by a weighted calculation based on the first sharpness index and the second sharpness index;

[0041] The first sharpness index is obtained based on the image quality assessment of the brick information, wherein the image quality assessment includes calculating image contrast and image sharpness;

[0042] The second clarity index is obtained by evaluating the detection environment in which the brick information is collected, wherein the evaluation of the detection environment includes calculating the dust concentration index and the ambient light interference index.

[0043] Compared with existing technologies, the intelligent detection method for refractory brick production lines of the present invention has the following advantages: It acquires brick information through a detection system and calculates a sharpness index to evaluate the quality of the brick information data. Based on the sharpness index, it determines the feature confidence level. By setting a preset feature confidence level, it evaluates the generation environment of the detection data and the image quality of the detection data. Based on the comparison results, it determines the corresponding control strategy. When the data is reliable, it determines to execute a cooling operation strategy. It calculates a high-confidence feature defect index and feature pose index using high-confidence detection data. Then, based on the comparison results of the feature defect index and feature pose index with corresponding thresholds, it determines whether to adjust the cooling parameters. When the data is unreliable, it executes a detection optimization strategy to perform a self-optimization strategy for the detection system, obtaining an optimized feature defect index and optimized feature pose index. This ensures that adjustment commands are generated based on reliable detection data, thereby improving product consistency and stability in overall production.

[0044] Furthermore, this invention uses a feature confidence level greater than a first preset threshold as the execution standard for determining data reliability. If the data is reliable, it is initially screened using a feature defect index. When the defect exceeds the threshold, the location risk of the defect is further determined based on the feature pose index. This setting achieves maximum energy and time savings while suppressing defect propagation.

[0045] Furthermore, the present invention also improves the data quality of the detection data in a targeted manner based on a variety of optimized detection methods. At the same time, by using the feature confidence calculated by weighted fusion of image quality and environmental quality, the reliability of the detection data is judged more accurately.

[0046] Furthermore, the present invention also calculates the characteristic defect index by quantifying defect size and quantity, and calculates the characteristic pose index by the correspondence between defect location and the stress area of ​​the billet, ensuring the consistency and comparability of the evaluation results and making the detection data more accurate. Attached Figure Description

[0047] Figure 1 This is a flowchart illustrating the intelligent detection method for a refractory brick production line in an embodiment of the present invention.

[0048] Figure 2 This is a flowchart illustrating the logic of determining the reliability of brick information based on feature confidence in an embodiment of the present invention.

[0049] Figure 3 This is a flowchart illustrating the logic of executing a cooling operation strategy based on the feature defect index and the feature pose index in an embodiment of the present invention.

[0050] Figure 4 This is a flowchart illustrating the logic of the cooling operation strategy based on the optimized feature defect index and the optimized feature pose index in an embodiment of the present invention. Detailed Implementation

[0051] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0052] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0053] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0054] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0055] Please see Figure 1 The diagram shown is a flowchart of an intelligent detection method for a refractory brick production line according to an embodiment of the present invention. The detection method includes at least the following steps:

[0056] S1: The refractory brick blanks are inspected through the detection system to obtain brick information and clarity indicators;

[0057] S2: Determine the feature defect index and feature pose index based on brick information, where brick information includes surface defect information and pose information;

[0058] S3: Determine feature confidence based on sharpness metrics, where sharpness metrics include a first sharpness metric and a second sharpness metric;

[0059] S4: Determine whether the brick information is reliable based on the comparison result between the feature confidence level and the preset feature confidence level, and generate the corresponding control strategy based on the determination result;

[0060] S51: If the brick information is determined to be reliable, determine to execute the cooling operation strategy;

[0061] S61: Determine whether to adjust the cooling parameters based on the comparison result between the feature defect index and the corresponding preset feature index, and adjust the cooling parameters based on the comparison result between the feature pose index and the corresponding preset feature index when adjusting the cooling parameters. The cooling parameters include cooling time and cooling rate, and the preset feature index includes preset defect index and preset pose index.

[0062] S52: If the brick information is determined to be unreliable, determine to execute the detection optimization strategy to obtain the optimized feature defect index and the optimized feature pose index;

[0063] S62: Based on the comparison results between the optimized feature defect index and the preset feature index, re-determine whether to adjust the cooling parameters, and when determining to adjust the cooling parameters, adjust the cooling parameters based on the comparison results between the optimized feature pose index and the preset feature index.

[0064] In this embodiment, a cooling operation is also performed based on the adjusted and finally determined cooling parameters to obtain the finished refractory bricks.

[0065] Specifically, an industrial camera is used as the first visual inspection mechanism in the detection system of this invention to acquire two-dimensional images of crack or hole defect data on the surface of the refractory brick blank, thereby obtaining surface image information; a 3D vision camera is used as the second visual inspection mechanism in the detection system of this invention to acquire three-dimensional point cloud data of the refractory brick blank, thereby obtaining the positional information of the crack or hole defect data on the refractory brick blank; the detection system uses environmental sensors to detect the environmental state of the refractory brick blank in real time to obtain dust concentration data and ambient light intensity data.

[0066] In a specific implementation, the first sharpness index S1 is calculated based on the image contrast and sharpness of the frame image of the brick information. The higher the value, the better the quality of the acquired brick image. The image contrast index Cs is calculated using the formula: Cs = (Ca - Cmin) / (Cmax - Cmin), which characterizes the richness of the image's grayscale levels. A larger value indicates a more ideal image contrast and easier feature recognition. Ca represents the image contrast, while Cmax and Cmin are determined based on statistical analysis of a large number of high-quality brick images acquired under ideal lighting conditions. Specifically, in a laboratory environment, lighting and camera parameters are systematically adjusted, images with different contrasts are acquired, and defects are manually labeled. A curve showing the relationship between contrast and recognition accuracy is established. The inflection point where accuracy drops sharply is selected as Cmin, and the peak contrast of mainstream high-quality images is used as Cmax. The contrast distribution is statistically analyzed, and the 95th percentile is taken as Cmax, and the 5th percentile as Cmin. For example, Cmax = 0.8 and Cmin = 0.2 can be set.

[0067] The image sharpness index Cr is calculated as follows: Cr = (Sr - Smin) / (Smax - Smin), which represents the clarity of the image edges. The larger the value, the clearer the image, the better the details are preserved, and the more distinct the defect boundaries are. Sr is the image sharpness. Smax and Smin are determined based on the texture shooting conditions of the camera used in this embodiment under optimal focusing and vibration-free conditions. The specific method is as follows: using a resolution test card and the brick blank to be tested, at a fixed distance, the camera focal length is finely adjusted by a motor to collect a series of images from clear to blurry, calculate its Sr and correspond it with the result of manual evaluation to determine the threshold. For example, Smax = 800 and Smin = 50 can be set.

[0068] The first sharpness index S1 is calculated as follows: S1 = c1·Cs + c2·Cr, where c1 and c2 are preset weight coefficients, and c1 + c2 = 1, which is used to balance the contribution of contrast and sharpness to image quality assessment. Based on the feature extraction algorithm's greater sensitivity to edge information, c1 = 0.4 and c2 = 0.6 can be set.

[0069] The second clarity index, S2, is calculated based on dust concentration data and ambient light intensity data acquired by environmental sensors. A higher S2 value indicates a harsher detection environment and a greater negative impact on the reliability of data acquisition by the detection system. The dust concentration index, Dsub, is determined using the formula: Dsub=(Draw-Da) / (Db-Da), and characterizes the severity of dust interference in the detection environment. A higher Dsub value indicates a higher dust concentration, stronger obstruction and scattering effects on optical detection, and lower data reliability. Draw is the real-time airborne particulate matter mass concentration Da measured by the dust sensor, and Db is based on... The data is determined based on the manual's recommendations and long-term on-site observation data. Specifically, Da is set according to the upper limit of the clean working environment for optical equipment, and Db is determined based on the dust concentration at which the measured image quality begins to significantly deteriorate. For example, Da = 0.1 mg / m³ and Db = 1.0 mg / m³. The ambient light interference index Lsub is determined based on the formula: Lsub = |Lraw - Lopt| / Lopt. It characterizes the degree to which the ambient light deviates from ideal conditions. The larger the value, the greater the interference caused by excessively dark or bright ambient light, which may cause overexposure, underexposure, or shadows in the image, affecting feature extraction. Lraw is the illuminance value measured in real time by the ambient light sensor, and Lopt is determined based on the preset optimal detection illuminance of the system. Specifically, it is determined by testing the image quality under different illuminance levels. For example, Lopt = 500 lux.

[0070] The formula for calculating the second clarity index S2 is: S2 = wd·Dsub + wl·Lsub. Here, wd and wl are preset weighting coefficients, and wd + wl = 1, used to balance the impact of dust and light interference on the overall environmental quality assessment. Based on the fact that dust is the main source of interference in the on-site environment of this embodiment, the specific method is as follows: during the production line debugging phase or under typical operating conditions, while maintaining ideal light intensity Lsub≈0, the dust concentration is gradually increased, and the accuracy decrease curve of the feature extraction algorithm is recorded as Dsub increases from 0 to 1; while maintaining low dust Dsub≈0, the ambient light intensity is gradually changed, and the accuracy decrease curve of the same feature extraction algorithm is recorded as Lsub increases from 0 to 1. By analyzing the slope of the two decreasing curves or the ratio of Dsub to Lsub when their impact on accuracy is equal, the relative weight of their influence can be determined. For example, wd = 0.6 and wl = 0.4 can be set.

[0071] Please see Figure 2 The diagram shows the logical flowchart for determining the reliability of brick information based on feature confidence in an embodiment of the present invention. In this embodiment, feature confidence should be positively correlated with the image quality of the brick information itself and negatively correlated with the degree of environmental interference. Therefore, by comprehensively calculating S1 and S2, the image quality of the brick information and the detection environment are quantified into the interval [0,1]. The formula for calculating feature confidence C is: C=k1·S1+k2·(1-S2), where k1 and k2 are preset weight coefficients set based on experimental calibration or experience. Specifically, training is performed using historical production data or specially collected sample datasets. These sample datasets contain a large number of samples, each with its corresponding S1 and S2 values, to indicate whether the process decision made based on the detection data is verified as "correct" by subsequent product quality. The dataset is fitted using machine learning methods such as logistic regression. The coefficients of S1 and S2 in the best-fit model are normalized and can be used as the values ​​of k1 and k2. For example, k1=0.7 and k2=0.3. C represents the reliability of brick information feature extraction. The closer its value is to 1, the higher the image quality and the less environmental interference.

[0072] Furthermore, the detection and control system compares the calculated feature confidence level C with the preset feature confidence level Cth. The preset feature confidence level Cth is set based on experimental calibration or experience. Specifically, it is trained using historical production data or a specially collected sample dataset containing a large number of samples, each with its corresponding S1 and S2 values ​​to indicate whether the process decision made based on the detection data is verified as "correct" by subsequent product quality. The dataset is then fitted using machine learning methods such as logistic regression. The coefficients of S1 and S2 in the best-fit model, after normalization, can be used as the values ​​of k1 and k2. For example, they can be set to 0.75.

[0073] If C > Cth, it indicates that the image quality obtained by the current detection is good and the environmental interference is small. Therefore, the brick information is determined to be reliable, and the detection control system determines the control strategy to be the cooling operation strategy.

[0074] If C≤Cth, the brick information is deemed unreliable, indicating poor image quality or severe interference from the detection environment. The detection control system then determines the control strategy as the detection optimization strategy.

[0075] In this embodiment, when the detection and control system determines that the brick information is reliable based on the feature confidence level C, it determines to execute a cooling operation strategy, and determines the final cooling parameters based on the comparison results of the feature defect index F and the feature pose index P with the preset threshold.

[0076] The characteristic defect index F is used to characterize the severity of surface defects in refractory brick blanks. Preferably, the characteristic defect index F is calculated using brick crack data in the surface image information as surface defects. Based on the surface image information collected by the first vision inspection agency, the calculation formula is: F=Lmax / Lth, where Lmax is the longest crack length among all cracks identified from the surface image, and Lth is a preset crack length threshold. Lth is based on fracture mechanics analysis and historical production data statistics. Specifically, it provides data on blanks scrapped due to cracks in past production cycles, statistically analyzes the distribution of their maximum crack lengths, and takes the lower limit as an empirical threshold. F is a dimensionless value; the larger the value, the more significant the surface defects of the blank.

[0077] The feature pose index P is used to characterize the degree of clustering of defects at critical risk locations in the brick. By combining the surface image information identified by the first vision inspection mechanism with camera calibration parameters and the pose information obtained by the second vision inspection mechanism, an XYZ coordinate system is established and precisely mapped onto the three-dimensional blank model to determine the specific location of each defect in three-dimensional space. The critical risk locations in the brick are pre-determined based on the geometric structure and stress analysis of the refractory brick. The brick's edges and corners are divided into high-stress risk zones, and the center of the plane is divided into low-stress risk zones, based on experimental calibration or empirical settings. The specific formula is: P = ( The formula is Wh*Nh+Wm*Nm+Wl*Nl) / (Wh*Nt), where Nh, Nm, and Nl represent the number of cracks located in the high, medium, and low stress risk zones, respectively; Nt is the total number of detected cracks, Nt=Nh+Nm+Nl; Wh, Wm, and Wl are the preset risk weights for the corresponding high, medium, and low risk zones, determined based on historical production data. Specifically, within a complete production cycle, the frequency of cracks occurring in the high, medium, and low risk zones in failure cases is collected, and the frequency is approximately 70% in the high-risk zone, approximately 25% in the medium-risk zone, and approximately 5% in the low-risk zone. This frequency distribution reflects the relative "contribution" or "risk" of different zones to initiating failure. Accordingly, the weight ratio is set to be positively correlated with this contribution and normalized so that its maximum value Wh is 1.0. For example, Wh=1.0, Wm=0.6, and Wl=0.2 can be set. The higher the P-value, the more concentrated the defect data is in the high-stress risk area, and the greater the possibility of defect expansion due to thermal stress concentration during subsequent cooling.

[0078] Please see Figure 3As shown, it is a logical flowchart of the cooling operation strategy based on the feature defect index and feature pose index in an embodiment of the present invention. Specifically, if C > Cth, when the detection and control system determines to execute the cooling operation strategy, it determines whether to adjust the cooling parameters based on the comparison results of the characteristic defect index F and the characteristic pose index P with the corresponding preset characteristic indices. The preset characteristic indices include the preset defect index Fth and the preset pose index Pth. The preset defect index Fth is based on historical production data statistics and process experiment calibration. Specifically, it collects the F value data of all billets in the past continuous production cycles and performs correlation analysis with the final product quality. Based on the fact that about 95% of the billets in the final qualified products have an F value lower than this threshold, while about 80% of the billets scrapped due to cracks have an F value higher than this threshold, this threshold is determined as the statistical reference benchmark for Fth. For example, it can be set to 0.15. The preset pose index Pth is based on experimental calibration or historical data setting. Specifically, it extracts historical data of billets scrapped due to cooling cracks from the production database and statistically analyzes the distribution of P values ​​of failure cases. 90% of the cases have a P value greater than 0.6. For example, it can be set to 0.6.

[0079] The comparison process based on the feature defect index F and the preset defect index Fth, as well as the feature pose index P and the preset pose index Pth, is as follows:

[0080] If F ≤ Fth, the current crack defect is considered to be relatively minor, indicating that the defect itself is small in size and its stress concentration effect is weak. Therefore, it is insufficient to drive crack propagation in the thermal stress field generated during subsequent cooling. Thus, the initial cooling parameters are determined as the final cooling parameters.

[0081] If F > Fth, then the current crack defect is considered to be severe, indicating that the defect has the potential to expand. Therefore, it is necessary to determine the local stress state at the location of the defect.

[0082] Furthermore, when the detection and control system determines that the crack defect is severe based on the current judgment result, it determines whether to adjust the initial cooling parameters by comparing the characteristic pose index P with the preset pose index Pth.

[0083] If P≤Pth, it is determined that the current crack defect is severe but the location risk is low. Although the defect size is large, since most of the crack defects are located in the low stress risk area of ​​the brick, they are insufficient to drive the crack to propagate. Therefore, the initial cooling parameters are determined as the final cooling parameters.

[0084] If P > Pth, it is determined that the current crack defect is severe and the location risk is high, indicating that the refractory brick blank has a large defect size and many high stress risk areas, which makes the crack easy to propagate. Therefore, the initial cooling parameters should be adjusted.

[0085] In one specific embodiment, the initial cooling parameters include initial cooling time T0 and initial cooling rate V0. These initial cooling parameters are determined based on the material properties of the refractory bricks and historical production data. Specifically, from the historical database of the production management system, the cooling process parameters corresponding to blanks determined to be defect-free or low-risk from the production records of all refractory brick products that ultimately meet quality standards are selected. The dataset of cooling time T and cooling rate V is statistically analyzed, and their mode or median is calculated. The obtained values ​​are set as the baseline initial cooling parameters T0 and V0. For example, T0 can be set to 300 minutes and V0 to 50℃ / h. The detection and control system calculates the difference between the characteristic pose index P and the preset pose index Pth to obtain the first pose difference value ΔP, and determines the adjustment range of the initial cooling parameters based on the first pose difference value ΔP. Here, ΔP characterizes the degree to which the defects of the current refractory brick blank exceed the stress risk boundary.

[0086] Specifically, when it is determined that the initial cooling parameters need to be adjusted, the detection and control system adjusts the initial cooling parameters by calculating the first attitude difference value ΔP, thereby slowing down the cooling and reducing the thermal stress level. Therefore, the adjusted cooling time T and cooling rate V are respectively: T=T0·(1+kt·ΔP), V=V0 / (1+kv·ΔP).

[0087] Wherein, kt and kv are adjustment coefficients determined based on experiments and simulations. For example, kt=0.8 and kv=1.0 can be set, indicating that the higher the location risk, i.e. the larger ΔP, the slower the cooling, i.e. the more T increases and V decreases, thus selectively suppressing thermal stress within a range that prevents crack propagation.

[0088] In this embodiment, if the calculated value of C=0.82>Cth=0.75 indicates that the brick information is reliable, then the calculated value of F=0.12<Fth=0.15 indicates that the current crack defect is relatively minor, and the initial cooling parameters are determined to be the final cooling parameters, i.e., T=300 minutes, V=50℃ / h;

[0089] If the test results show that C=0.88 > Cth=0.75, the brick information is considered reliable. If the calculated F=0.22 > Fth=0.15, the current crack defect is considered severe. If the calculated P=0.55 ≤ Pth=0.6, the current crack defect location is considered to have a low risk. Therefore, the initial cooling parameters are determined to be the final cooling parameters, i.e., T=300 minutes, V=50℃ / h.

[0090] If the detection calculation shows C=0.90>Cth=0.75, indicating the brick information is reliable, then the calculated F=0.22>Fth=0.15 indicates the current crack defect is severe, and the calculated P=0.75>Pth=0.6 indicates the current crack defect is severe and the location risk is high. Therefore, the initial cooling parameters are adjusted, and the adjusted cooling parameters are used as the final cooling parameters. The calculated ΔP=0.15 is substituted into the adjustment formula to determine the adjusted cooling parameters as: T=336 minutes, V≈43.5℃ / h.

[0091] The detection and control system controls the cooling equipment to execute the corresponding cooling process based on the final determined cooling parameters, and finally obtains refractory brick products that meet the quality requirements, effectively avoiding the increase in scrap rate caused by crack propagation.

[0092] In one specific embodiment, if C≤Cth, when the detection control system determines to execute the detection optimization strategy, it performs an optimization operation to actively intervene and improve the detection quality of the detection data acquisition stage, so as to obtain reliable brick information, thereby providing accurate data for subsequent cooling parameter adjustment, including,

[0093] The system automatically adjusts the exposure time, gain, or aperture of industrial cameras to adapt to current lighting conditions, improving image brightness and contrast. If the system determines that the overall image is too dark, it increases the exposure time. It also controls or enhances the brightness of dedicated LED light sources at the inspection station to provide stable and uniform illumination, eliminating shadows or overexposure. While enhancing illumination and adjusting sensor parameters, the decision-making system can increase the number of sensors to improve data acquisition density and coverage. Furthermore, if conventional optimization operations fail, the vision inspection agency can temporarily switch to a backup feature extraction algorithm that is not sensitive to noise, or perform noise reduction and fusion processing on several consecutively acquired frames to improve the signal-to-noise ratio.

[0094] Please see Figure 4 As shown, this is a flowchart illustrating the logic of the cooling operation strategy based on the optimized feature defect index and optimized feature pose index in an embodiment of the present invention. In one specific embodiment, after the detection system completes the predetermined optimization operation, the detection control system will re-execute the detection and information acquisition steps to acquire new brick information and sharpness indicators, and recalculate the new feature confidence level Cnew.

[0095] If Cnew > Cth, the optimization is deemed effective, the brick information is reliable, indicating that the currently detected image quality is good and environmental interference is minimal. The detection control system then determines the control strategy as a cooling operation strategy. Based on the current reliable brick information, the detection control system calculates the optimized feature defect index Fnew and the optimized feature pose index Pnew, and continues to compare Fnew with the preset defect index Fth.

[0096] If Fnew≤Fth, then the final cooling parameters are determined to be the initial cooling parameters (T0,V0).

[0097] If Fnew > Fth, determine whether to adjust the initial cooling parameters by comparing the optimized feature pose index Pnew with the preset pose index Pth.

[0098] If Pnew≤Pth, the final cooling parameters are determined to be the initial cooling parameters (T0,V0).

[0099] If Pnew>Pth, then determine the adjustment of the initial cooling parameters, calculate the new second pose difference value ΔPnew=Pnew-Pth, and calculate the new cooling time and new cooling rate after the final optimization detection adjustment according to the adjustment formula described above: Tnew=T0·(1+kt·ΔPnew), Vnew=V0 / (1+kv·ΔPnew).

[0100] The detection and control system adjusts the initial cooling parameters to determine the final cooling parameters to be executed.

[0101] If Cnew≤Cth, the optimization is deemed invalid and the brick information is unreliable. This indicates that the image quality of the brick information is poor or there is serious interference in the detection environment. The detection control system marks the brick as an abnormality and diverts or issues a manual maintenance alarm.

[0102] In one specific embodiment, if the calculated value of C = 0.70 ≤ Cth = 0.75, the brick information is determined to be unreliable. The control strategy is then set to an optimized detection strategy. After re-detecting the optimized brick information, Cnew = 0.88 > Cth = 0.75, indicating that the brick information is reliable. Based on the latest brick information, Fnew = 0.25 > Fth = 0.15, indicating that the current crack defect is severe. Pnew = 0.65 > Pth = 0.6, indicating that the current crack defect is severe and the location risk is high. The initial cooling parameters are then adjusted. ΔPnew = 0.05 is calculated and substituted into the adjustment formula to determine the optimized detection cooling parameters as follows: Tnew = 312 minutes, Vnew ≈ 47.6℃ / h.

[0103] If the calculated value of C=0.68≤Cth=0.75 indicates that the brick information is unreliable, the control strategy is determined to be the detection optimization strategy. The confidence level of the optimized brick information is then re-detected. If the value of Cnew=0.72≤Cth=0.75, the detection optimization operation is deemed invalid. The detection control system then determines that the brick is an abnormality and diverts it to the manual inspection area, or issues an abnormality alarm to prompt technicians to conduct an inspection.

[0104] All technologies not mentioned in the above embodiments are applicable to existing technologies. It is understood that no specific limitation is made to any preset parameter or critical parameter in the embodiments of the present invention, and the above values ​​are not limited thereto. Those skilled in the art can adjust the preset parameters or critical parameters accordingly based on actual needs, analysis of historical data, or equipment usage.

[0105] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. An intelligent detection method for a refractory brick production line, characterized in that, include: The rough blanks of refractory bricks are tested to obtain information about the brick body and clarity indicators; Based on the brick information, the feature defect index and feature pose index are determined, wherein the brick information includes surface defect information and pose information; The feature confidence level is determined based on the sharpness index, wherein the sharpness index includes a first sharpness index and a second sharpness index. Based on the comparison result between the stated feature confidence level and the preset feature confidence level, the reliability of the brick information is determined, and a corresponding control strategy is generated based on the determination result. If the brick information is determined to be reliable, a cooling operation strategy is determined to be executed. The execution process includes determining whether to adjust the cooling parameters based on the comparison result between the feature defect index and the corresponding preset feature index, and adjusting the cooling parameters based on the comparison result between the feature pose index and the corresponding preset feature index when adjusting the cooling parameters. The cooling parameters include cooling time and cooling rate, and the preset feature index includes a preset defect index and a preset pose index. If the brick information is determined to be unreliable, a detection optimization strategy is executed to obtain an optimized feature defect index and an optimized feature pose index. Based on the comparison result between the optimized feature defect index and the preset feature index, it is re-determined whether to adjust the cooling parameters. When adjusting the cooling parameters, the cooling parameters are adjusted based on the comparison result between the optimized feature pose index and the preset feature index.

2. The intelligent detection method for a refractory brick production line according to claim 1, characterized in that, The process of determining whether the brick information is reliable based on the comparison result between the feature confidence level and the preset feature confidence level includes: If the confidence level of the feature is greater than the preset confidence level of the feature, the brick information is determined to be reliable, and the cooling operation strategy is executed. If the feature confidence level is less than or equal to the preset feature confidence level, the brick information is determined to be unreliable, and the detection optimization strategy is executed.

3. The intelligent detection method for a refractory brick production line according to claim 2, characterized in that, The process of executing the cooling operation strategy includes: If the characteristic defect index is less than or equal to the preset defect index, the initial cooling parameter is determined as the final cooling parameter; If the feature defect index is greater than the preset defect index, determine whether to adjust the initial cooling parameters based on the comparison result between the feature pose index and the preset pose index.

4. The intelligent detection method for a refractory brick production line according to claim 3, characterized in that, If the feature pose index is less than or equal to the preset pose index, the final cooling parameter is determined to be the initial cooling parameter. If the feature pose index is greater than the preset pose index, the initial cooling parameters are adjusted to obtain the final determined cooling parameters.

5. The intelligent detection method for a refractory brick production line according to claim 4, characterized in that, The process of adjusting the initial cooling parameters to obtain the final determined cooling parameters includes: Calculate the difference between the feature pose index and the preset pose index and record it as the first pose difference value; The adjustment range of the initial cooling parameters is determined based on the first position difference value, wherein the extension of the cooling time is positively correlated with the first position difference value, and the reduction of the cooling rate is positively correlated with the first position difference value.

6. The intelligent detection method for a refractory brick production line according to claim 2, characterized in that, The process of determining to execute the detection optimization strategy to obtain the optimized feature defect index and the optimized feature pose index includes: Based on the determination that the brick information is unreliable, corresponding detection optimization operations are performed. These operations include increasing the number of sensors, switching the feature recognition algorithm, adjusting sensor parameters, or enabling auxiliary lighting. The optimized feature defect index and the optimized feature pose index are determined based on the brick information reacquired after the detection optimization operation. The cooling operation strategy is executed based on the optimized feature defect index and the optimized feature pose index.

7. The intelligent detection method for a refractory brick production line according to claim 6, characterized in that, The process of executing the cooling operation strategy based on the optimized feature defect index and the optimized feature pose index includes: If the optimized feature defect index is less than or equal to the preset defect index, the initial cooling parameter is determined as the final cooling parameter; If the optimized feature defect index is greater than the preset defect index, determine whether to adjust the initial cooling parameters based on the comparison result between the optimized feature pose index and the preset pose index.

8. The intelligent detection method for a refractory brick production line according to claim 7, characterized in that, If the optimized feature pose index is less than or equal to the preset pose index, the final cooling parameter is determined to be the initial cooling parameter. If the optimized feature pose index is greater than the preset pose index, the difference between the optimized feature pose index and the preset pose index is calculated and recorded as the second pose difference. The adjustment range of the initial cooling parameters is determined based on the second position difference value, wherein the extension of the cooling time is positively correlated with the second position difference value, and the reduction of the cooling rate is positively correlated with the second position difference value.

9. The intelligent detection method for a refractory brick production line according to claim 1, characterized in that, The process of determining the feature defect index and feature pose index based on the brick information includes: The inspection system includes a first vision inspection mechanism and a second vision inspection mechanism; The surface defect information is obtained through the first visual inspection mechanism, wherein the surface defect information includes at least one of crack defect data or hole defect data; The pose information is obtained through the second visual detection mechanism, wherein the pose information includes at least the position information of the crack defect data or the hole defect data on the refractory brick blank.

10. The intelligent detection method for a refractory brick production line according to claim 1, characterized in that, The process of determining feature confidence based on the sharpness index includes: The feature confidence level is determined by a weighted calculation based on the first sharpness index and the second sharpness index; The first sharpness index is obtained based on the image quality assessment of the brick information, wherein the image quality assessment includes calculating image contrast and image sharpness; The second clarity index is obtained by evaluating the detection environment in which the brick information is collected, wherein the evaluation of the detection environment includes calculating the dust concentration index and the ambient light interference index.

Images