Blast furnace displacement monitoring and repairing method, system and device based on computer vision

Abstract

The invention provides a blast furnace displacement monitoring and repairing method, system and device based on computer vision, and relates to the technical field of blast furnace displacement monitoring. The method mainly comprises the steps that a first camera is arranged near a blast furnace and used for shooting blast furnace images from the side view angle; based on the blast furnace image # imgabs 1 # of the # imgabs 0 # frame, identifying a rough position # imgabs 2 # of a preset target point of the blast furnace through a filter; through a feature point tracking method, the repair displacement # imgabs3 # of the target point is calculated; and when the repair displacement # imgabs4 # exceeds a preset threshold value, the blast furnace is reversely corrected through an adjusting device arranged at the blast furnace base. According to the scheme, the displacement of the blast furnace can be monitored in real time, so that the investment is low, and the problems of low efficiency, influence on productivity, unstable detection result and the like caused by manual detection and repair are avoided; error compensation can be carried out by arranging a second camera, so that the error of the first camera caused by the change of the external environment is eliminated; according to the scheme, the displacement of the blast furnace can be reliably monitored and repaired for a long time.

Description

Technical Field

[0001] The present invention relates to the technical field of blast furnace displacement monitoring, and in particular to a blast furnace displacement monitoring and repair method, system and device based on computer vision. Background Art

[0002] The blast furnace is the core equipment in the steel production process. Its main function is to reduce raw materials such as iron ore, coke and limestone at high temperature to produce liquid iron and slag.

[0003] During the service life of a blast furnace, its interior needs to withstand the coupled movement of solid-liquid-gas three-phase materials and high temperature and high pressure. Once the bottom of the blast furnace has problems such as uneven temperature distribution, untimely slag cleaning, support subsidence, construction defects, etc., it is very easy for the overall tilting and deformation to occur, resulting in deformation of the furnace bottom, local gas leakage, poor cooling effect of the furnace bottom, etc., which seriously affects the utilization rate and service life of the blast furnace, and may even cause an overall collapse accident. For this reason, large equipment such as blast furnaces need to be manually inspected and repaired regularly, that is, the inspectors make displacement judgments through visual inspection and professional equipment, but this not only affects production, but also has a huge workload and unstable test results.

[0004] At present, there have been preliminary studies and applications of displacement monitoring methods for large public buildings, large bridges and other infrastructure based on computer vision technology. However, in harsh outdoor environments, computer vision monitoring results often cannot meet engineering needs. External factors such as lighting changes caused by sunlight angles, image distortion caused by uneven air heating, ground vibrations or strong winds may affect the accuracy of computer vision monitoring results. Therefore, how to conduct long-term and reliable displacement monitoring of large equipment such as blast furnaces has become a major issue that needs to be solved in the industry. Summary of the invention

[0005] The object of the present invention is to provide a blast furnace displacement monitoring and repair method, system and device based on computer vision to solve at least one of the above-mentioned technical problems existing in the prior art.

[0006] In the first aspect, in order to solve the above technical problems, the present invention provides a blast furnace displacement monitoring and repair method based on computer vision, comprising the following steps: Step 1: A first camera is arranged near the blast furnace to capture an image of the blast furnace from a side angle.

[0007] Step 2: Based on Frame of blast furnace image , through the filter, identify the rough position of the preset target point of the blast furnace ; Calculate the restoration displacement of the target point through the feature point tracking method . In a feasible implementation manner, the filter is an SCF filter (SVM based correlation filters, correlation filter based on support vector machine); the feature point tracking method is a KLT (Kanade-Lucas-Tomasi) algorithm.

[0008] In a feasible implementation manner, the step 2 specifically includes: Step 21, from In the frame blast furnace image, several regions with objects of interest (explicit features) are selected as object regions; Step 22: From the object area, select a square patch containing the target point to obtain the grayscale matrix ; The size of the square patch is Pixel; the specific format of the grayscale matrix is: ; in, Represents the gray value of the pixel in the nth row and nth column in the square patch; Through the Hann function, the weight of the center of the square patch is enhanced to obtain the basic sample. The specific formula is: ; in, Represents the input value of the SCF filter; Step 23: Input a value for one of the square patches , through the circulation operator matrix, converted to dimensional input vector , the specific formula is: ; For the remaining input values ​​in the square patch, the operator matrix , build Generate samples and get the circulant matrix of the input vector , the specific formula is: ; The circulant operator matrix The size is , the specific format is: ; Based on the samples, reshape the image and train the SVM classifier; In this way, the good characteristics of the circulant matrix can be used to generate more samples containing the target point, which is conducive to training an SVM classifier that can converge quickly; Step 24, convolution operation is performed on the trained SVM classifier and the corresponding square patch in the next frame of blast furnace image, and the coordinates at the highest point of the convolution value are used as the position of the target point in the frame of blast furnace image; iteratively execute step 21, and the rough position self-consistent field of the target point of the blast furnace can be calculated in real time until the iteration termination condition is reached, and step 25 is executed; Step 25: Calculate the Rough position of the preset target point in the blast furnace image , through the correspondence between frames and time, Frame conversion to After the moment, the specific calculation formula is: ; in, Indicates the change in the rough position of each frame of blast furnace image; From the beginning to the The number of frames at the moment; Step 26: Use the improved KLT algorithm to identify the center point coordinate position of the square patch. The specific formula includes: Define the target point In the The intensity at the frame is ; represents the horizontal coordinate; Indicates the vertical coordinate; The residual difference in intensity between the current frame and the previous frame is defined as ; Define Repair Displacement The calculation formula is: ; in, Indicates the change in the horizontal coordinate; Indicates the change in the longitudinal coordinate; The residual is then approximately represented by the Taylor series truncated to the linear term. The specific formula is: ; in, Represents the partial derivative vector, and the specific expression is: ; in, express for The partial derivative of express for The partial derivative of definition The specific calculation formula is: ; in, represents the first parameter of the KLT algorithm, Represents the second parameter of the KLT algorithm; The specific calculation formula is: ; in, Represents the third parameter of the KLT algorithm, Represents the fourth parameter of the KLT algorithm; The specific calculation formula is: .

[0009] In a feasible implementation manner, the specific method for training the SVM classifier in step 23 is: Step 231: Based on the similarity between each generated sample and the basic sample, a classification label is defined and the continuous value of the classification label is calculated. The specific calculation formula is: ; in, , and These are all hyperparameters of the SVM classifier; represents the norm of 2; Indicates (1≤ ≤ ) continuous value of the classification label of the samples; Indicates the coordinate position of the center point of the generated sample The coordinate position of the center point of the basic sample The specific calculation formula for the distance between is: ; in, In the loop matrix The specific operations are: ; It represents the remainder after dividing the first term by the second term; Indicates rounding down; Step 232: Based on the continuous value, a SVM classifier is used to classify the continuous value to obtain a discrete classification label. The specific formula is: ; in, Represents discrete categorical labels; express The complex conjugate function of ; express Discrete Fourier Transform (DFT) of By alternating the optimization method, update and , solve , the specific calculation formula includes: ; ; in, Represents the first optimization parameter, and the specific calculation formula is: ; express The discrete Fourier transform of express The average value of represents the second optimization parameter.

[0010] In a feasible implementation manner, the alternating optimization method in step 232 includes: Step 2321, initialization Frame Blast Image and , set the number of iterations is equal to 1; Step 2322: When the SVM classifier does not converge, update the current number of iterations. The specific formula is: ; ; in, express The vector form of Indicates taking the maximum value; Step 2323: Update the current number of iterations value, Value and The specific formula includes: ; ; ; in, Indicates taking the average; represents the third optimization parameter; Step 2324: Update the current number of iterations The specific formula is: ; in, represents the hyperparameters of the SVM classifier; Let the number of iterations Add 1, execute step 2322, until the Convergence of the SVM classifier on the blast furnace image.

[0011] In a feasible implementation manner, the step 2 further includes step 27: Fix the second camera on the first camera, and fix the target card on the reference object within the field of view of the second camera; calibrate the two cameras; collect the target card image in real time through the second camera, calculate the false displacement deviation and compensate the target point coordinate position of the blast furnace image collected by the first camera (in step 26), and obtain the real target point coordinate; obtain the repaired displacement according to the calculation method in step 26 In this way, the false displacement deviation of the second camera can be brought into the image processing process of the first camera, thereby eliminating the shooting error of the first camera caused by environmental changes (for example, the increase in ambient temperature causes a longitudinal displacement error in the first camera).

[0012] In a feasible implementation manner, the method for calibrating the two cameras in step 27 is: Step 271: Calculate the projective transformation of the first camera image before and after displacement by direct linear transformation. The specific formula is: ; in, Indicates The image coordinate system of the first camera and the second camera in the frame image; Indicated in Middle Frame time target point Direction coordinates; Indicated in Middle Frame time target point Direction coordinates; represents the world coordinate system defined by the initial position of the target photographed by the first camera; Indicated in Middle Frame time target point Direction coordinates; Indicated in Middle Frame time target point Direction coordinates; Indicates from arrive The projective transformation of (assuming only two-dimensional motion in the image plane); Indicates the number of target points (feature points); Step 272: Calibrate the second camera using the Zhang Zhengyou method to determine the internal parameters of the second camera. And the projection function, the specific formula of the projection function is: ; in, The first Projection function of the frame image from the time-varying coordinate system to the image coordinate system; Indicates The image coordinate system of the first camera in the frame image; represents the time-varying coordinate system due to camera offset; Step 273, calibrating unknown parameters of the first camera, including , , , , , , , and the second camera The world coordinate system position of the target point in the first frame image , the specific calibration formula is: ; in, Indicates The target point is Frame Direction coordinates; Indicates The target point is Frame Towards coordinates.

[0013] In a feasible implementation manner, the method for calculating the real target point coordinates in step 27 includes: Step 274: Get the first camera's The coordinates of each target point in the frame image; Step 275: Calculate the second camera's corresponding The coordinates of the target point when the frame image is taken; based on the calibration formula in step 273, the false displacement deviation is calculated, and the specific formula is: ; in, express arrive The rotation matrix of represents the identity matrix; express arrive The translation matrix of Step 276: Calculate the first camera's In the frame image, the actual target point coordinates of each target point are as follows: ; in, express The conjugate transpose of .

[0014] Step 3: When repairing displacement When the preset threshold is exceeded, an alarm message is generated; based on the alarm message, the adjustment device arranged at the blast furnace base is used to repair the displacement according to the above , in the coordinate plane of the blast furnace image, reversely correct the blast furnace until the displacement is repaired Less than or equal to the preset threshold.

[0015] In a feasible embodiment, the adjusting device includes: a plurality of circumferentially distributed upper stiffening plates are arranged on the outer circumferential side wall of the blast furnace body; an upper ring plate is fixed on the upper stiffening plate along the circumferential direction of the outer blast furnace body; a plurality of circumferentially distributed lower stiffening plates are arranged on the outer circumferential side wall of the blast furnace base; a lower ring plate is fixed on the side of the lower stiffening plate away from the blast furnace base; between the upper ring plate and the lower ring plate, a plurality of jacks and bolt assemblies used in pairs are symmetrically arranged along the circumferential direction; the jack is used to increase the single-sided distance between the upper ring plate and the lower ring plate; the bolt assembly is used to reduce the single-sided distance between the upper ring plate and the lower ring plate, and to lock it.

[0016] In a feasible implementation manner, the adjustment method of the adjustment device is: The two jacks in the coordinate plane of the blast furnace image are defined as a left jack and a right jack respectively; the two bolt assemblies in the coordinate plane of the blast furnace image are defined as a left bolt assembly and a right bolt assembly respectively; When repairing displacement When moving to the left, loosen the left bolt assembly, lower the right jack, then raise the output shaft of the left jack, and lift the left end of the upper ring plate until the displacement is repaired. Less than or equal to the preset threshold; then, tighten the left bolt assembly, raise the output shaft of the right jack to the lower surface of the right end of the upper ring plate, and then tighten the right bolt assembly to complete the blast furnace displacement repair; When repairing displacement When moving to the right, loosen the right bolt assembly, lower the left jack, then raise the output shaft of the right jack, and lift the right end of the upper ring plate until the displacement is repaired. Less than or equal to the preset threshold; then, tighten the right bolt assembly, raise the output shaft of the left jack to the lower surface of the left end of the upper ring plate, and then tighten the left bolt assembly to complete the blast furnace displacement repair.

[0017] In the second aspect, based on the same inventive concept, the present application also provides a blast furnace displacement monitoring and repair system based on computer vision, including a data acquisition module, a data processing module and a result generation module; The data acquisition module includes a first acquisition unit for acquiring a blast furnace image from a side view angle through a first camera; The data processing module includes a filter unit, a feature tracking unit and an alarm unit; The filter unit is based on Frame of blast furnace image , through the filter, identify the rough position of the preset target point of the blast furnace ; The feature tracking unit, based on the rough position , calculate the restoration displacement of the target point through the feature point tracking method ; The alarm unit, when the repair displacement When the preset threshold is exceeded, an alarm message is generated; the alarm message includes the repair displacement ; The result generation module is used to send the alarm information externally.

[0018] In a feasible implementation manner, the data acquisition module further includes a second acquisition unit, which is used to acquire the target card image through a second camera; The data processing module also includes a compensation unit, which calculates the false displacement deviation based on the target card image and compensates the target point coordinate position of the blast furnace image to obtain the real target point coordinates and sends them to the filter unit.

[0019] On the third aspect, based on the same inventive concept, the present application also provides a blast furnace displacement monitoring and repair device based on computer vision, including a camera, a processor, a memory and a bus, wherein the memory stores instructions and data that can be read by the processor, and the processor is used to call the instructions and data in the memory to execute the blast furnace displacement monitoring and repair method based on computer vision as described above, and the bus connects the functional components for transmitting information.

[0020] By adopting the above technical solution, the present invention has the following beneficial effects: The present invention provides a computer vision-based blast furnace displacement monitoring and repair method, system and device, which can monitor the blast furnace displacement in real time. Not only does it require less equipment investment, but it also avoids the problems of low efficiency, impact on production capacity, unstable detection results, etc. caused by manual detection and repair. By setting a second camera, error compensation can also be performed to eliminate the error of the first camera caused by changes in the external environment. This solution can monitor and repair the blast furnace displacement for a long time and reliably. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0022] Figure 1 A flow chart of a blast furnace displacement monitoring and repair method based on computer vision provided by an embodiment of the present invention; Figure 2 An example of a blast furnace image provided by an embodiment of the present invention; Figure 3 An example diagram of a square plaque provided in an embodiment of the present invention; Figure 4 A basic sample example diagram provided for an embodiment of the present invention; Figure 5 A diagram illustrating the principle of step 23 provided in an embodiment of the present invention; Figure 6 A diagram illustrating the repair displacement provided by an embodiment of the present invention; Figure 7 A second camera layout diagram provided by an embodiment of the present invention; Figure 8 A diagram illustrating longitudinal displacement errors provided by an embodiment of the present invention; Fig. 9 A structural diagram of an adjusting device provided in an embodiment of the present invention; Fig.10 A top view layout diagram of the lower ring plate provided in an embodiment of the present invention; Fig.11 A diagram of a blast furnace displacement monitoring and repair system based on computer vision provided by an embodiment of the present invention; Reference numerals: 1- blast furnace body; 2- blast furnace base; 11- upper stiffening plate; 12- upper ring plate; 21- lower stiffening plate; 22- lower ring plate; 3- jack; 31- left jack; 32- right jack; 4- bolt assembly; 5- first camera; 6- second camera; 7- target card. DETAILED DESCRIPTION

[0023] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0024] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", and "third" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance.

[0025] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0026] The present invention is further explained below in conjunction with specific implementation modes.

[0027] It should also be noted that the following specific embodiments or specific implementations are a series of optimized settings listed in the present invention to further explain the specific content of the invention, and these settings can be used in combination or in association with each other.

[0028] Embodiment 1: like Figure 1 As shown, this embodiment provides a blast furnace displacement monitoring and repair method based on computer vision, including the following steps: Step 1: A first camera is arranged near the blast furnace to capture an image of the blast furnace from a side angle.

[0029] Step 2: Based on Frame of blast furnace image , through the filter, identify the rough position of the preset target point of the blast furnace ; Calculate the restoration displacement of the target point through the feature point tracking method In this way, the restoration displacement calculation can be directly performed through the blast furnace image, avoiding the need to set up a special image card on the blast furnace.

[0030] Furthermore, the filter is an SCF filter (SVM based correlation filters, correlation filter based on support vector machine); the feature point tracking method is a KLT (Kanade-Lucas-Tomasi) algorithm.

[0031] Furthermore, the step 2 specifically includes: Step 21: Figure 2 As shown, from In the frame blast furnace image, several regions with objects of interest (explicit features) are selected as object regions; Step 22. From the object area, select a square patch containing the target point (e.g., centered at the Harris corner point), such as Figure 3 As shown, the grayscale matrix is ​​obtained ; The square plaque ( Figure 3 The size of the green box in (41*41) pixels; the specific format of the grayscale matrix is: ; in, Represents the gray value of the pixel in the nth row and nth column in the square patch; Through the Hann function, the weight of the center of the square patch is enhanced to obtain the basic sample, such as Figure 4 As shown, the specific formula is: ; in, Represents the input value of the SCF filter; Step 23: Figure 5 As shown, for an input value in the square patch , through the circulation operator matrix, converted to dimensional input vector , the specific formula is: ; For the remaining input values ​​in the square patch, the operator matrix , build Generate samples and get the circulant matrix of the input vector , the specific formula is: ; The circulant operator matrix The size is , the specific format is: ; Based on the samples, reshape the image and train the SVM classifier; In this way, the good characteristics of the circulant matrix can be used to generate more samples containing the target point, which is conducive to training an SVM classifier that can converge quickly; Step 24, convolve the trained SVM classifier with the corresponding square patch in the next frame of blast furnace image, and use the coordinates at the highest point of the convolution value as the position of the target point in the frame of blast furnace image; iterate step 21 to calculate the rough position self-consistent field of the target point of the blast furnace in real time until the iteration termination condition is reached (for example, an instruction to stop blast furnace displacement detection is received), and then execute step 25; Step 25: Calculate the Rough position of the preset target point in the blast furnace image , through the correspondence between frames and time (generally industrial cameras have 100 frames per second), Frame conversion to After the moment, the specific calculation formula is: ; in, Indicates the change in the rough position of each frame of blast furnace image; From the beginning to the The number of frames at the moment; Step 26: Use the improved KLT algorithm to identify the center point coordinate position of the square patch. The specific formula includes: Define the target point In the The intensity at the frame is (It can be the same as the input value of the SCF filter calculated by the Hann function in step 22); represents the horizontal coordinate; Indicates the vertical coordinate; The residual difference in intensity between the current frame and the previous frame is defined as ; like Figure 6 As shown, define the repair displacement The calculation formula is: ; in, Indicates the change in the horizontal coordinate; Indicates the change in the longitudinal coordinate; The residual is then approximately represented by the Taylor series truncated to the linear term. The specific formula is: ; in, Represents the partial derivative vector, and the specific expression is: ; in, express for The partial derivative of express for The partial derivative of definition The specific calculation formula is: ; in, represents the first parameter of the KLT algorithm, Represents the second parameter of the KLT algorithm; The specific calculation formula is: ; in, Represents the third parameter of the KLT algorithm, Represents the fourth parameter of the KLT algorithm; The specific calculation formula is: .

[0032] Furthermore, the specific method of training the SVM classifier in step 23 is: Step 231: Based on the similarity between each generated sample and the basic sample, a classification label is defined and the continuous value of the classification label is calculated. The specific calculation formula is: ; in, , and These are all hyperparameters of the SVM classifier; represents the norm of 2; Indicates (1≤ ≤ ) continuous value of the classification label of the samples; Indicates the coordinate position of the center point of the generated sample The coordinate position of the center point of the basic sample The specific calculation formula for the distance between is: ; in, In the loop matrix The specific operations are: ; It represents the remainder after dividing the first term by the second term; Indicates rounding down; For example, when Take 293, When taking 41, Value is ; Step 232: Based on the continuous value, a SVM classifier is used to classify the continuous value to obtain a discrete classification label. The specific formula is: ; in, Represents discrete classification labels (for example, it can be divided into three categories: displacement safety label, displacement qualified label and displacement exceeding limit label); express The complex conjugate function of ; express Discrete Fourier Transform (DFT) of By alternating the optimization method, update and , solve , the specific calculation formula includes: ; ; in, Represents the first optimization parameter, and the specific calculation formula is: ; express The discrete Fourier transform of express The average value of represents the second optimization parameter.

[0033] Furthermore, the alternating optimization method in step 232 includes: Step 2321, initialization Frame Blast Image and , set the number of iterations is equal to 1; Step 2322: When the SVM classifier does not converge, update the current number of iterations. The specific formula is: ; ; in, express The vector form of Indicates taking the maximum value; Step 2323: Update the current number of iterations value, Value and The specific formula includes: ; ; ; in, Indicates taking the average; represents the third optimization parameter; Step 2324: Update the current number of iterations The specific formula is: ; in, represents the hyperparameters of the SVM classifier; Let the number of iterations Add 1, execute step 2322, until the Convergence of the SVM classifier on the blast furnace image.

[0034] Furthermore, the step 2 further includes step 27: A second camera 6 is fixed on the first camera 5, and a target card 7 (for example, a black and white chessboard) is fixed on a reference object within the field of view of the second camera 6. Figure 7 As shown; calibrate the two cameras; collect the image of the target card 7 in real time through the second camera 6, calculate the false displacement deviation and compensate (in step 26) the target point coordinate position of the blast furnace image collected by the first camera 5 to obtain the real target point coordinate; obtain the repair displacement according to the calculation method in step 26 In this way, the false displacement deviation of the second camera 6 can be brought into the image processing process of the first camera 5, thereby eliminating the shooting error of the first camera 5 caused by environmental changes (for example, the increase in ambient temperature causes the first camera 5 to have a longitudinal displacement error, such as Figure 8 as shown).

[0035] Furthermore, the method for calibrating the two cameras in step 27 is: Step 271: Calculate the projective transformation of the first camera image before and after displacement by direct linear transformation. The specific formula is: ; in, Indicates The image coordinate system of the first camera and the second camera in the frame image; Indicated in Middle Frame time target point Direction coordinates; Indicated in Middle Frame time target point Direction coordinates; represents the world coordinate system defined by the initial position of the target photographed by the first camera; Indicated in Middle Frame time target point Direction coordinates; Indicated in Middle Frame time target point Direction coordinates; Indicates from arrive The projective transformation of (assuming only two-dimensional motion in the image plane); Indicates the number of target points (feature points); Step 272: Calibrate the second camera using the Zhang Zhengyou method to determine the internal parameters of the second camera. And the projection function, the specific formula of the projection function is: ; in, The first Projection function of the frame image from the time-varying coordinate system to the image coordinate system; Indicates The image coordinate system of the first camera in the frame image; represents the time-varying coordinate system due to camera offset; Step 273, calibrating unknown parameters of the first camera, including , , , , , , , and the second camera The world coordinate system position of the target point in the first frame image , the specific calibration formula is: ; in, Indicates The target point is Frame Direction coordinates; Indicates The target point is Frame Towards coordinates.

[0036] Furthermore, the calculation method of the real target point coordinates in step 27 includes: Step 274: Get the first camera's The coordinates of each target point in the frame image; Step 275: Calculate the second camera's corresponding The coordinates of the target point when the frame image is taken; based on the calibration formula in step 273, the false displacement deviation is calculated, and the specific formula is: ; in, express arrive The rotation matrix of represents the identity matrix; express arrive The translation matrix of Step 276: Calculate the first camera's In the frame image, the actual target point coordinates of each target point are as follows: ; in, express The conjugate transpose of .

[0037] Step 3: When repairing displacement When the preset threshold is exceeded, an alarm message is generated; based on the alarm message, the adjustment device arranged at the blast furnace base is used to repair the displacement according to the above , in the coordinate plane of the blast furnace image, reversely correct the blast furnace until the displacement is repaired Less than or equal to the preset threshold.

[0038] Furthermore, if Fig. 9 As shown, the adjusting device includes: a plurality of circumferentially distributed upper stiffening plates 11 are arranged on the outer circumferential side wall of the blast furnace body 1; an upper ring plate 12 is fixed on the upper stiffening plate 11 along the outer circumferential direction of the blast furnace body 1; a plurality of circumferentially distributed lower stiffening plates 21 are arranged on the outer circumferential side wall of the blast furnace base 2; a lower ring plate 22 is fixed on the side of the lower stiffening plate 21 away from the blast furnace base; between the upper ring plate 12 and the lower ring plate 22, a plurality of jacks 3 and bolt assemblies 4 used in pairs are symmetrically arranged along the circumferential direction; the jack 3 is used to increase the single-sided distance between the upper ring plate 12 and the lower ring plate 22; the bolt assembly 4 is used to reduce the single-sided distance between the upper ring plate 12 and the lower ring plate 22, and to lock it.

[0039] Furthermore, if Fig.10As shown, the adjustment method of the adjustment device is: The two jacks in the coordinate plane of the blast furnace image are defined as the left jack 31 and the right jack 32 respectively; the two bolt assemblies ( Fig.10 ), respectively defined as a left bolt assembly and a right bolt assembly; When repairing displacement When moving to the left, loosen the left bolt assembly, lower the output shaft of the right jack 32, then raise the output shaft of the left jack 31, and lift the left end of the upper ring plate 12 until the displacement is repaired. Less than or equal to the preset threshold; then, tighten the left bolt assembly, raise the output shaft of the right jack 32 to the lower surface of the right end of the upper ring plate 12, and then tighten the right bolt assembly to complete the blast furnace displacement repair; When repairing displacement When moving to the right, loosen the right bolt assembly, lower the output shaft of the left jack 31, then raise the output shaft of the right jack 32, and lift the right end of the upper ring plate 12 until the displacement is repaired. Less than or equal to the preset threshold; then, tighten the right bolt assembly, lift the output shaft of the left jack 31 to the lower surface of the left end of the upper ring plate 12, and then tighten the left bolt assembly to complete the blast furnace displacement repair.

[0040] Embodiment 2: like Fig.11 As shown, this embodiment provides a blast furnace displacement monitoring and repair system based on computer vision, including a data acquisition module, a data processing module and a result generation module; The data acquisition module includes a first acquisition unit for acquiring a blast furnace image from a side view angle through a first camera; The data processing module includes a filter unit, a feature tracking unit and an alarm unit; The filter unit is based on Frame of blast furnace image , through the filter, identify the rough position of the preset target point of the blast furnace ; The feature tracking unit, based on the rough position , calculate the restoration displacement of the target point through the feature point tracking method ; The alarm unit, when the repair displacement When the preset threshold is exceeded, an alarm message is generated; the alarm message includes the repair displacement ; The result generation module is used to send the alarm information externally.

[0041] Furthermore, the data acquisition module further includes a second acquisition unit, which is used to acquire the target card image through a second camera; The data processing module also includes a compensation unit, which calculates the false displacement deviation based on the target card image and compensates the target point coordinate position of the blast furnace image to obtain the real target point coordinates and sends them to the filter unit.

[0042] Embodiment three: The present embodiment provides a blast furnace displacement monitoring and repair device based on computer vision, including a camera, a processor, a memory and a bus. The memory stores instructions and data that can be read by the processor. The processor is used to call the instructions and data in the memory to execute the blast furnace displacement monitoring and repair method based on computer vision as described above. The bus connects the functional components for transmitting information.

[0043] Furthermore, it also includes the adjustment device mentioned in the first embodiment.

[0044] In another implementation, the present solution can also be implemented by an integrated device, which may include corresponding modules for executing each or several steps in the above-mentioned various implementations. The module may be one or more hardware modules specially configured to execute the corresponding steps, or implemented by a processor configured to execute the corresponding steps, or stored in a computer-readable medium for implementation by a processor, or implemented by some combination.

[0045] The processor performs the various methods and processes described above. For example, the method implementation in the present solution can be implemented as a software program, which is tangibly contained in a machine-readable medium, such as a memory. In some embodiments, part or all of the software program can be loaded and / or installed via a memory and / or a communication interface. When the software program is loaded into the memory and executed by the processor, one or more steps in the method described above can be executed. Alternatively, in other embodiments, the processor can be configured to perform one of the above methods by any other appropriate means (for example, by means of firmware).

[0046] The device can be implemented using a bus architecture. The bus architecture can include any number of interconnecting buses and bridges, depending on the specific application and overall design constraints of the hardware. The bus connects various circuits including one or more processors, memories, and / or hardware modules together. The bus can also connect various other circuits such as peripherals, voltage regulators, power management circuits, external antennas, etc.

[0047] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Component (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc.

[0048] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A blast furnace displacement monitoring and repair method based on computer vision, characterized in that: include: Step 1: a first camera is arranged near the blast furnace to capture an image of the blast furnace from a side view angle; Step 2: Based on Frame of blast furnace image , through the filter, identify the rough position of the preset target point of the blast furnace ; By using the feature point tracking method, the restoration displacement of the target point is calculated. ; Step 3: When repairing displacement When the preset threshold is exceeded, an alarm message is generated; Based on the alarm information, the adjustment device set at the blast furnace base is used to repair the displacement according to the , in the coordinate plane of the blast furnace image, reversely correct the blast furnace until the displacement is repaired Less than or equal to the preset threshold.

2. The method according to claim 1, characterized in that The filter is an SCF filter; the feature point tracking method is a KLT algorithm.

3. The method according to claim 2, characterized in that The step 2 specifically includes: Step 21, from In the frame blast furnace image, several regions with objects of interest are selected as object regions; Step 22: From the object area, select a square patch containing the target point to obtain the grayscale matrix ; The size of the square patch is pixels; the specific format of the grayscale matrix is: ； in, Represents the gray value of the pixel in the nth row and nth column in the square patch; Through the Hann function, the weight of the center of the square patch is enhanced to obtain the basic sample. The specific formula is: ； in, Represents the input value of the SCF filter; Step 23: Input a value for one of the square patches , through the circulation operator matrix, converted to dimensional input vector , the specific formula is: ； For the remaining input values ​​in the square patch, the operator matrix , build Generate samples and get the circulant matrix of the input vector , the specific formula is: ； The circulant operator matrix The size is pixels, the specific format is: ； Based on the samples, reshape the image and train the SVM classifier; Step 24, convolution operation is performed on the trained SVM classifier and the corresponding square patch in the next frame of blast furnace image, and the coordinates at the highest point of the convolution value are used as the position of the target point in the frame of blast furnace image; iteratively execute step 21, and calculate the rough position self-consistent field of the target point of the blast furnace in real time until the iteration termination condition is reached, and execute step 25; Step 25: Calculate the Rough position of the preset target point in the blast furnace image , through the correspondence between frames and time, Frame conversion to After the moment, the specific calculation formula is: ； in, Indicates the change in the rough position of each frame of blast furnace image; From the beginning to the The number of frames at the moment; Step 26: Using the improved KLT algorithm, identify the center point coordinate position of the square patch, specifically including: Define the target point In the The intensity at the frame is ; represents the horizontal coordinate; Indicates the vertical coordinate; The residual difference in intensity between the current frame and the previous frame is defined as ; Define Repair Displacement The calculation formula is: ； in, Indicates the change in the horizontal coordinate; Indicates the change in the longitudinal coordinate; The residual is then expressed by the Taylor series truncated to the linear term. The specific formula is: ； in, Represents the partial derivative vector, and the specific expression is: ； in, express for The partial derivative of express for The partial derivative of definition The specific calculation formula is: ； in, represents the first parameter of the KLT algorithm, Represents the second parameter of the KLT algorithm; The specific calculation formula is: ； in, Represents the third parameter of the KLT algorithm, Represents the fourth parameter of the KLT algorithm; The specific calculation formula is: 。 4. The method according to claim 3, characterized in that The specific method for training the SVM classifier in step 23 is: Step 231: Based on the similarity between each generated sample and the basic sample, a classification label is defined and the continuous value of the classification label is calculated. The specific calculation formula is: ； in, , and These are all hyperparameters of the SVM classifier; represents the norm of 2; Indicates The continuous value of the classification label of the samples; Indicates the coordinate position of the center point of the generated sample The coordinate position of the center point of the basic sample The specific calculation formula for the distance between is: ； in, In the loop matrix The specific operations are: ; It represents the remainder after dividing the first term by the second term; Indicates rounding down; Step 232: Based on the continuous value, a SVM classifier is used to classify the continuous value to obtain a discrete classification label. The specific formula is: ； in, Represents discrete categorical labels; express The complex conjugate function of ; express The discrete Fourier transform of By alternating the optimization method, update and , solve , the specific calculation formula includes: ； ； in, Represents the first optimization parameter, and the specific calculation formula is: ； express The discrete Fourier transform of express The average value of represents the second optimization parameter.

5. The method according to claim 4, characterized in that The alternating optimization method in step 232 includes: Step 2321, initialization Frame Blast Image and , set the number of iterations is equal to 1; Step 2322: When the SVM classifier does not converge, update the current number of iterations. The specific formula is: ； ； in, express The vector form of Indicates taking the maximum value; Step 2323: Update the current number of iterations value, Value and The specific formula includes: ； ； ； in, Indicates taking the average; represents the third optimization parameter; Step 2324: Update the current number of iterations The specific formula is: ； in, represents the hyperparameters of the SVM classifier; Let the number of iterations Add 1, execute step 2322, until the Convergence of the SVM classifier on the blast furnace image.

6. The method according to claim 3, characterized in that The step 2 further comprises a step 27: Fix the second camera on the first camera, and fix the target card on the reference object within the field of view of the second camera; calibrate the two cameras; collect the target card image in real time through the second camera, calculate the false displacement deviation and compensate the target point coordinate position of the blast furnace image collected by the first camera, and obtain the real target point coordinate; obtain the repaired displacement according to the calculation method in step 26 .

7. The method according to claim 6, characterized in that The method for calibrating the two cameras in step 27 is: Step 271: Calculate the projective transformation of the first camera image before and after displacement by direct linear transformation. The specific formula is: ； in, Indicates The image coordinate system of the first camera and the second camera in the frame image; Indicated in Middle Frame time target point Direction coordinates; Indicated in Middle Frame time target point Direction coordinates; represents the world coordinate system defined by the initial position of the target photographed by the first camera; Indicated in Middle Frame time target point Direction coordinates; Indicated in Middle Frame time target point Direction coordinates; Indicates from arrive The projective transformation of Indicates the number of target points; Step 272: Calibrate the second camera using the Zhang Zhengyou method to determine the internal parameters of the second camera. And the projection function, the specific formula of the projection function is: ； in, The first Projection function of the frame image from the time-varying coordinate system to the image coordinate system; Indicates The image coordinate system of the first camera in the frame image; represents the time-varying coordinate system due to camera offset; Step 273, calibrating unknown parameters of the first camera, including , , , , , , , and the second camera The world coordinate system position of the target point in the first frame image , the specific calibration formula is: ； in, Indicates The target point is Frame Direction coordinates; Indicates The target point is Frame Towards coordinates.

8. The method according to claim 7, characterized in that The calculation method of the actual target point coordinates in step 27 includes: Step 274: Get the first camera's The coordinates of each target point in the frame image; Step 275: Calculate the second camera's corresponding The coordinates of the target point when the frame image is taken; based on the calibration formula in step 273, the false displacement deviation is calculated, and the specific formula is: ； in, express arrive The rotation matrix of represents the identity matrix; express arrive The translation matrix of Step 276: Calculate the first camera's In the frame image, the actual target point coordinates of each target point are as follows: ； in, express The conjugate transpose of .

9. A blast furnace displacement monitoring and repair system based on computer vision, characterized in that: It includes a data acquisition module, a data processing module and a result generation module; The data acquisition module includes a first acquisition unit for acquiring a blast furnace image from a side view angle through a first camera; The data processing module includes a filter unit, a feature tracking unit and an alarm unit; The filter unit is based on Frame of blast furnace image , through the filter, identify the rough position of the preset target point of the blast furnace ; The feature tracking unit, based on the rough position , calculate the restoration displacement of the target point through the feature point tracking method ; The alarm unit, when the repair displacement When the preset threshold is exceeded, an alarm message is generated; the alarm message includes the repair displacement ; The result generation module is used to send the alarm information externally.

10. A blast furnace displacement monitoring and repair device based on computer vision, characterized in that: It includes a camera, a processor, a memory and a bus, wherein the memory stores instructions and data read by the processor, the processor is used to call the instructions and data in the memory to execute any method as described in claims 1-8, and the bus connects the functional components for transmitting information.

Images