Intelligent coating method and system for aluminum coil composite coating based on computer vision

Abstract

The present application relates to the technical field of image semantic recognition and image semantic analysis, and specifically to a computer vision-based intelligent coating method and system for aluminum coil composite coating, the method comprising: collecting the coating image of the aluminum coil to obtain the current aluminum coil defect binary image; calculating the stripe defect correction factor to determine each stripe defect area of ​​the current aluminum coil defect binary image; calculating the stripe defect rotation coefficient and stripe defect distribution index of the current aluminum coil defect binary image; calculating the coating pressure adjustment coefficient of the current aluminum coil defect binary image; determining the integral gain parameter in the PID algorithm; adjusting the roller speed ratio of the roller coater during the next aluminum coil coating process to complete the intelligent roller coating of the aluminum coil. The stripe defect characteristics are analyzed by image recognition technology, the integral gain parameter of the PID algorithm is determined, and the roller speed ratio is adjusted according to the output result of the control system to reduce the stripe defects and improve the coating quality of the aluminum coil.

Description

Technical Field

[0001] The present application relates to the technical field of image semantic recognition and image semantic analysis, and in particular to a method and system for intelligent coating of composite coatings on aluminum coils based on computer vision. Background Art

[0002] Aluminum coil coating has always been an important part of the industrial manufacturing field. Aluminum coil is widely used in many industries such as building decoration, transportation, and electronic appliances due to its light weight, corrosion resistance, and good processing performance. In these applications, the coating not only plays a role in aesthetics, but also has the functions of protecting aluminum and extending service life. Therefore, the uniformity, adhesion, wear resistance and other performance indicators of the coating have a decisive influence on the quality of the final product.

[0003] During the coating process, the roller speed ratio determines the amount of paint transferred between the rollers, which can effectively control the wet film thickness. The coating roller and the hanging roller run faster relative to the aluminum coil, and the coating amount per unit time increases, the paint film thickens, and the roller speed is reduced, and the paint film becomes thinner. If the roller speed ratio does not match, it will cause streak defects on the paint film. How to accurately adjust the roller speed ratio is a technical problem of this application. Closed-loop control can be achieved through the PID algorithm, and the characteristics of the streak defects are analyzed through image recognition technology to determine the parameters of the PID algorithm. Secondly, the roller speed ratio is adjusted through the output of the control system to reduce the streak defects and improve the coating quality of the aluminum coil. Summary of the invention

[0004] In order to solve the problem that it is difficult to accurately adjust the roller speed ratio in the prior art and reduce the stripe defects occurring during the coating process of aluminum coils, a computer vision-based intelligent coating method and system for aluminum coil composite coatings are provided to solve the existing problems.

[0005] The solution to the technical problem of this application is to provide a computer vision-based intelligent coating method and system for aluminum coil composite coating, including the following steps:

[0006] In a first aspect, an embodiment of the present application provides a method for intelligent coating of composite coating on aluminum coil based on computer vision, the method comprising the following steps:

[0007] Collect the coating image of the aluminum coil and obtain the binary image of the aluminum coil defect at the current moment;

[0008] Calculate the stripe defect correction factor according to the aspect ratio of each defect area in the aluminum coil defect binary image at the current moment, and determine the stripe defect area of ​​the aluminum coil defect binary image at the current moment;

[0009] According to the angle between each stripe defect and the horizontal axis in the aluminum coil defect binary image at the current moment, a stripe defect rotation coefficient of the aluminum coil defect binary image at the current moment is calculated;

[0010] Calculate the stripe defect distribution index of the aluminum coil defect binary image at the current moment according to the distance between each stripe defect area and the distribution of each stripe defect area in the aluminum coil defect binary image at the current moment;

[0011] Based on the stripe defect rotation coefficient, the stripe defect distribution index and the angle between each stripe defect area and the horizontal axis of the aluminum coil defect binary image at the current moment, the coating pressure adjustment coefficient of the aluminum coil defect binary image at the current moment is obtained;

[0012] The integral gain parameter in the PID algorithm is determined by the coating pressure adjustment coefficient of the aluminum coil defect binary map at the current moment. According to the output of the control system, the roller speed ratio of the roller coater in the aluminum coil coating process at the next moment is adjusted to complete the intelligent roller coating of the aluminum coil.

[0013] Preferably, the stripe defect correction factor is the average of the aspect ratios of all defective regions in the aluminum coil defect binary image at the current moment.

[0014] Preferably, the method for determining each stripe defect area of ​​the aluminum coil defect binary image at the current moment is:

[0015] If the aspect ratio of the defect region is greater than the stripe defect correction factor, the defect region is a stripe defect region.

[0016] Preferably, the calculation method of the stripe defect rotation coefficient of the aluminum coil defect binary image at the current moment is:

[0017] Calculate the angle between one edge of the minimum circumscribed rectangle of each stripe defect area in the aluminum coil defect binary image at the current moment and the horizontal axis as the rotation angle;

[0018] The deviation value of the preset rotation angle is recorded as the deviation angle;

[0019] Calculate the average value of the rotation angles of all stripe defect areas in the aluminum coil defect binary image at the current moment and record it as the average rotation angle;

[0020] The sum of the average rotation angle and the deviation angle is recorded as the rotation deviation angle;

[0021] Perform sine transform and cosine transform on the rotation deviation angle in the binary image of aluminum coil defects at the current moment respectively. If the sine transform result is greater than 0, the sine value is assigned to 1; if the sine transform result is less than 0, the sine value is assigned to 0; if the cosine transform result is greater than 0, the cosine value is assigned to 1; if the cosine transform result is less than 0, the cosine value is assigned to 0; if the sine transform result is equal to 0, both the sine value and the cosine value are assigned to 0; if the cosine transform result is equal to 0, the sine value is assigned to 1 and the cosine value is assigned to 0;

[0022] If the sine value and the cosine value have the same value, the fringe defect rotation coefficient is 1; if the sine value and the cosine value have different value, the fringe defect rotation coefficient is 0.

[0023] Preferably, the calculation of the stripe defect distribution index of the aluminum coil defect binary image at the current moment includes:

[0024] According to the distance between each stripe defect area in the aluminum coil defect binary image at the current moment, the defect area distribution distance of the aluminum coil defect binary image at the current moment is calculated;

[0025] According to the distribution of each stripe defect area in the aluminum coil defect binary image at the current moment, the defect area distribution density of the aluminum coil defect binary image at the current moment is calculated;

[0026] The stripe defect distribution index of the aluminum coil defect binary image at the current moment is negatively correlated with the distribution distance of the defect area, and is positively correlated with the distribution density of the defect area.

[0027] Preferably, the calculation method of the defect area distribution distance of the aluminum coil defect binary image at the current moment is:

[0028] The average value of the horizontal coordinate and the average value of the vertical coordinate of the pixel points with a pixel value of 1 in each stripe defect area in the aluminum coil defect binary image at the current moment are used as the centroid coordinates of each stripe defect area;

[0029] Calculate the distance between the centroids of any two stripe defect areas in the aluminum coil defect binary image at the current moment and record it as the centroid distance;

[0030] The average of all the centroid distances in the aluminum coil defect binary image at the current moment is used as the defect area distribution distance of the aluminum coil defect binary image at the current moment.

[0031] Preferably, the calculation method of the defect area distribution density of the aluminum coil defect binary image at the current moment is:

[0032] The sum of the number of pixels of all stripe defect areas on the aluminum coil defect binary image at the current moment is recorded as the defect area;

[0033] The number of all pixel points of the aluminum coil defect binary image at the current moment is recorded as the binary image area;

[0034] The ratio of the defect region area to the binary image area is calculated as the defect region distribution density of the aluminum coil defect binary image at the current moment.

[0035] Preferably, the calculation of the coating pressure adjustment coefficient of the aluminum coil defect binary image at the current moment includes:

[0036] Preset an adjustment coefficient, where the value range of the adjustment coefficient is (0,1);

[0037] The difference between the adjustment coefficient and the stripe defect rotation coefficient of the aluminum coil defect binary image at the current moment is used as a factor of the coating pressure adjustment coefficient;

[0038] If the stripe defect rotation coefficient is 1, the stripe defect influence factor is the average rotation angle of the aluminum coil defect binary image at the current moment; if the stripe defect rotation coefficient is 0, the stripe defect influence factor is the stripe defect distribution index of the aluminum coil defect binary image at the current moment;

[0039] The coating pressure adjustment coefficient of the aluminum coil defect binary image at the current moment is positively correlated with the factor of the coating pressure adjustment coefficient and the stripe defect influencing factor.

[0040] Preferably, the step of determining the adjusted integral gain parameter in the PID algorithm, adjusting the roller speed ratio of the roller coater during the coating process of the aluminum coil at the next moment, and completing the intelligent roller coating of the aluminum coil comprises:

[0041] Preset the initial value of the proportional gain parameter of the PID algorithm and the initial value of the differential gain parameter of the PID algorithm;

[0042] The sum of the coating pressure adjustment coefficient and the preset initial integral gain coefficient is used as the integral gain parameter of the PID algorithm;

[0043] The PID control algorithm is deployed in the control system of the roller coater to adjust the roller speed ratio of the roller coater to complete the intelligent roller coating of the aluminum coil.

[0044] In the second aspect, an embodiment of the present application also provides an intelligent coating system for composite coatings on aluminum coils based on computer vision, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein when the processor executes the computer program, the steps of any one of the above-mentioned intelligent coating methods for composite coatings on aluminum coils based on computer vision are implemented.

[0045] This application has at least the following beneficial effects:

[0046] The present application calculates the stripe defect correction factor by the aspect ratio in each defect area, and its beneficial effect is to exclude the noise that cannot be removed by Gaussian filtering, distinguish the stripe defect area in the aluminum coil defect binary image, and increase the accuracy of stripe defect calculation; the stripe defect rotation coefficient is calculated by the angle between each stripe defect area and the horizontal axis, and the horizontal defect and the vertical defect can be distinguished. Its beneficial effect is that different parameters can be adjusted for different types of stripe defects; the stripe defect distribution index is calculated by the centroid distance and distribution density of the vertical defect, and its beneficial effect is to exclude the interference of vertical line defects, thereby adjusting the roller speed ratio. Finally, the integral error of the PID algorithm is determined by the coating pressure adjustment coefficient, and the roller speed ratio is adjusted according to the output result of the control algorithm. Its beneficial effect is to analyze the stripe defect characteristics through image recognition technology to adjust the roller speed ratio, reduce the generation of stripe defects, and improve the coating quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The computer vision-based intelligent coating method for aluminum coil composite coating of the present application is further described in detail below in conjunction with the accompanying drawings.

[0048] Figure 1 A flowchart of the steps of the intelligent coating method of aluminum coil composite coating based on computer vision provided in this application;

[0049] Figure 2 An introduction to the types of streak defects that occur during the coating process of aluminum coils provided for this application;

[0050] Figure 3 The process introduction of two-roll coating process and reverse coating method provided for this application;

[0051] Figure 4 A flowchart of the steps of a method for obtaining the stripe defect distribution index of the aluminum coil defect binary image at the current moment provided in this application. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solution and advantages of this application more clear, the following is a further detailed description of the intelligent coating method and system for aluminum coil composite coating based on computer vision proposed in this application in combination with the accompanying drawings and implementation examples. It should be understood that the specific embodiments described herein are only used to explain this application and are not used to limit this application.

[0053] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0054] See also Figure 1 , which shows a flowchart of a method for intelligent coating of aluminum coil composite coating based on computer vision provided by an embodiment of the present application, the method comprising the following steps:

[0055] Step 1: collect the coating image of the aluminum coil and obtain the binary image of the aluminum coil defect at the current moment.

[0056] Roller coating is the core part of aluminum coil coating production, and its equipment includes roller coater, baking oven and cooling system. Roller coater uses coating roller to apply paint on aluminum coil, and the commonly used coating processes are two-roller coating process and three-roller coating process. There are two coating methods: forward coating and reverse coating, and the main difference lies in the different transfer ratio of the coating.

[0057] During the coating process of aluminum coils, stripe defects may occur, including transverse defects and vertical defects. Transverse defects are horizontal stripe defects, and vertical defects are vertical stripe defects and vertical line defects. Figure 2 As shown. Horizontal streak defects usually appear in the entire coating width, generally at regular intervals; the main manifestation is that the thickness of a horizontal line is different from the thickness of the coating before and after, and the defect on a horizontal line is more prominent than the coating before and after, which is a defect with uniform or regular intervals; the reason for the occurrence is generally that when two rollers are reversely coated, the speed of the coating roller is higher than the running speed of the aluminum coil, and the speed of the feeding roller is lower than the running speed of the aluminum coil, resulting in a mismatch between the speeds of the coating roller and the feeding roller. When the speed ratio is too high, horizontal streaks will be generated on the coating film; the solution is to reduce the roller speed ratio. Among them, the roller speed ratio refers to the speed ratio of the coating roller to the feeding roller.

[0058] Secondly, vertical streak defects are similar to horizontal streak defects and appear in the length direction; they mainly appear as lines in the longitudinal direction of the coating, which are thinner than the thickness of the adjacent positions, and generally multiple lines will appear; the reason for their occurrence is generally due to the mismatch of the speed ratio; the solution is to increase the roller speed ratio.

[0059] Finally, some vertical defects have only one line. The reason for this type of stripe is different from that of multiple stripes, and is called a vertical line defect. It mainly appears as a single line defect starting from a point and extending upstream. Like a comet, it is a vertical line with decreasing thickness along the coating direction. The reason for its occurrence is generally due to dust or dirt on the substrate. The solution is to establish a dust-free workshop or blow the substrate with high-pressure airflow before coating.

[0060] Based on the above analysis, the types of stripe defects are distinguished by analyzing the image. If it is a horizontal stripe defect, the roller speed ratio is reduced. If it is a continuous vertical stripe defect, the roller speed ratio is increased. The roller speed ratio is controlled by adjusting the parameters of the PID algorithm to reduce the occurrence of stripe defects, thereby improving the quality of aluminum coil coating.

[0061] A CCD industrial camera is installed on the aluminum coil coating production line and an appropriate light source is configured to ensure sufficient lighting and a suitable viewing angle to obtain a clear image. After the aluminum coil enters the coating step, a coating image is taken to obtain the current coating image of the aluminum coil. According to the current coating image of the aluminum coil, the adjustment parameters of the PID algorithm are calculated to obtain the roller speed ratio of the roller coater after adjustment. The coating image is taken again after every t moments to obtain the coating image of the aluminum coil at the next moment.

[0062] Preferably, as an embodiment of the present application, a fluorocarbon coating with a film thickness of 30 μm is used, and the coating is carried out using a two-roll coating process and a reverse coating method; the collected aluminum coil coating image at the current moment is grayed to reduce the complexity of the calculation; a Gaussian filter is applied for denoising to eliminate random noise in the image; the image contrast is improved by histogram equalization to make the boundary between the coating and the substrate more obvious, and finally a grayscale image of the aluminum coil coating at the current moment is obtained, and after obtaining the adjusted roller speed ratio of the roller coater, an image of the coating is taken again every 1 second. Among them, image graying, Gaussian filtering and histogram equalization are all well-known technologies, and the specific process will not be repeated.

[0063] In one embodiment of the present application, the process of two-roll coating process and reverse coating method is introduced as follows: Figure 3 As shown in FIG. A represents the coating roller, and P represents the feeding roller.

[0064] Furthermore, the coating image of the aluminum coil at the current moment is grayed to obtain the grayscale image of the coating of the aluminum coil at the current moment, and the segmented binary image is obtained through image segmentation and recorded as the aluminum coil defect binary image at the current moment, wherein the part with a pixel value of 0 represents the background, i.e., the normal coating area, and the part with a pixel value of 1 represents the defect area; secondly, the edge detection algorithm is used to detect the edge of each defect in the aluminum coil defect binary image at the current moment, and the edge of each defect contained in the aluminum coil defect binary image at the current moment is output; the minimum circumscribed rectangle of each defect edge in the aluminum coil defect binary image at the current moment is used as each defect area in the aluminum coil defect binary image at the current moment.

[0065] Preferably, as an embodiment of the present application, the aluminum coil coating grayscale image at the current moment is segmented using the Otsu threshold method to obtain a binary image of aluminum coil defects at the current moment; secondly, the Canny edge detection algorithm is used to output the edges of each defect in the binary image of aluminum coil defects at the current moment, and the defect edges are subjected to morphological operations such as opening, closing and dilation to make the edges clearer.

[0066] It should be understood that in one embodiment of the present application, the image segmentation algorithm adopts the Otsu threshold method, the edge detection algorithm is the Canny edge detection algorithm, and the morphological operation adopts opening operation, closing operation and dilation operation. In other embodiments of the present application, other technical methods such as region segmentation algorithm, Sobel edge detection, top hat operation, etc. can also be used. Among them, the Otsu threshold method, Canny edge detection algorithm and morphological operation are all well-known technologies, and the specific process will not be repeated.

[0067] At this point, the binary image of aluminum coil defects at the current moment is obtained.

[0068] Step 2, calculating the stripe defect correction factor according to the aspect ratio of each defect area in the aluminum coil defect binary image at the current moment, and determining each stripe defect area in the aluminum coil defect binary image at the current moment.

[0069] Since there are still some noises in the current aluminum coil coating grayscale image that cannot be removed by Gaussian filtering, the Otsu threshold method cannot accurately segment all defect areas. Therefore, the defect areas in the current aluminum coil defect binary image are not necessarily true stripe defects. The true stripe defect shape is a relatively long and thin straight line. The aspect ratio value of each defect area in the current aluminum coil defect binary image is relatively large. By calculating the stripe defect correction factor, it is characterized whether the obtained defect area is a true defect area, and the defect area obtained by noise misjudgment is excluded to avoid affecting the subsequent steps. The specific method is as follows:

[0070] Calculate the average of the aspect ratios of all defective areas in the aluminum coil defect binary image at the current moment as a stripe defect correction factor;

[0071] If the aspect ratio of a defect area in the aluminum coil defect binary image at the current moment is greater than or equal to the stripe defect correction factor, then the defect area is a stripe defect area, and the pixel value of the defect area is kept at 1; if the aspect ratio of a defect area in the aluminum coil defect binary image at the current moment is less than the stripe defect correction factor, then the defect area is not a stripe defect area, but is caused by noise misjudgment, so the pixel value of the defect area is set to 0.

[0072] It should be noted that through the above analysis, the defect areas caused by noise are excluded and the impact of noise is reduced. Because in the aluminum coil defect binary image, the defect areas caused by noise are far less than the real stripe defect areas, and the aspect ratio of the real stripe defect area is much larger than the aspect ratio of the defect area caused by noise, so if the aspect ratio of each defect area is less than the stripe defect correction factor, it can be considered to be caused by noise.

[0073] At this point, each stripe defect area of ​​the aluminum coil defect binary image at the current moment is obtained.

[0074] Step 3, according to the angle between each stripe defect and the horizontal axis in the aluminum coil defect binary image at the current moment, the stripe defect rotation coefficient of the aluminum coil defect binary image at the current moment is calculated to distinguish between horizontal defects and vertical defects.

[0075] The stripe defects in the current aluminum coil defect binary image are not necessarily completely parallel or perpendicular to the horizontal axis. Due to the coating process, each stripe defect often has a small angle with the horizontal axis. The rotation angle of each stripe defect area in the current aluminum coil defect binary image is obtained by OpenCV, where the rotation angle refers to the angle between one edge of the minimum circumscribed rectangle of the stripe defect area and the horizontal axis (X axis). For example, if the stripe defect is a horizontal defect, it is parallel to the horizontal axis, and the rotation angle of each stripe defect area is 0°, indicating that the current aluminum coil defect binary image has a horizontal stripe defect. If the stripe defect is a vertical defect, it is perpendicular to the horizontal axis, and the rotation angle of each stripe defect area is 90°, indicating that the current aluminum coil defect binary image has a vertical stripe defect or a vertical line defect. Therefore, by presetting a deviation value of the rotation angle, the stripe defect rotation coefficient is calculated to reflect the degree of rotation of the stripe defect, so as to distinguish whether it is a horizontal stripe defect caused by an excessive roller speed ratio, or a vertical stripe defect or a vertical line defect caused by an excessively small roller speed ratio. The specific method is as follows:

[0076] A rotation angle deviation value is preset and recorded as the deviation angle;

[0077] Calculate the average value of the rotation angles of all stripe defect areas in the aluminum coil defect binary image at the current moment and record it as the average rotation angle;

[0078] The sum of the average rotation angle and the deviation angle is recorded as the rotation deviation angle;

[0079] Perform sine transform and cosine transform on the rotation deviation angle of the binary image of aluminum coil defects at the current moment respectively. If the sine transform result is greater than 0, the sine value is assigned to 1; if the sine transform result is less than 0, the sine value is assigned to 0; if the cosine transform result is greater than 0, the cosine value is assigned to 1; if the cosine transform result is less than 0, the cosine value is assigned to 0; if the sine transform result is equal to 0, both the sine value and the cosine value are assigned to 0; if the cosine transform result is equal to 0, the sine value is assigned to 1 and the cosine value is assigned to 0;

[0080] If the sine value and the cosine value have the same value, the fringe defect rotation coefficient is 1; if the sine value and the cosine value have different value, the fringe defect rotation coefficient is 0.

[0081] Preferably, as an embodiment of the present application, the preset value of the deviation angle is 5°, that is, when the average rotation angle is 0°~5°, 355°~360° or 175°~185°, it is a horizontal stripe, and when the average rotation angle is 85°~95° or 265°~275°, it is a vertical stripe or vertical line; when the average rotation angle is not within the above range, this stripe defect area is considered to be other defects.

[0082] It should be noted that if the value of the stripe defect rotation coefficient of the aluminum coil defect binary image at the current moment is 1, it means that the stripe defect in the aluminum coil defect binary image at the current moment is a horizontal defect, that is, a horizontal stripe defect occurs; if the value of the stripe defect rotation coefficient of the aluminum coil defect binary image at the current moment is 0, it means that the stripe defect in the aluminum coil defect binary image at the current moment is a vertical defect, that is, a vertical stripe defect or a vertical line defect occurs.

[0083] At this point, the stripe defect rotation coefficient of the aluminum coil defect binary image at the current moment is obtained, and the horizontal defects and vertical defects are distinguished.

[0084] Step 4, according to the distance between each stripe defect area and the distribution of each stripe defect area in the aluminum coil defect binary image at the current moment, calculate the stripe defect distribution index of the aluminum coil defect binary image at the current moment, and distinguish between vertical stripe defects and vertical line defects.

[0085] Based on the above analysis, vertical streak defects are usually continuous and uninterrupted, while vertical line defects are intermittent. Secondly, the spacing between vertical streak defects is smaller than the spacing between vertical line defects, and the distribution density of vertical streak defects is higher than the distribution density between vertical line defects. Therefore, the streak defect distribution index can be obtained by calculating the centroid distance between each streak defect area and the distribution density of each streak defect area in the current aluminum coil defect binary image to characterize the distribution of vertical defects and distinguish between vertical streak defects and vertical line defects. The specific method is as follows:

[0086] The average value of the horizontal coordinate and the average value of the vertical coordinate of the pixel points with a pixel value of 1 in each stripe defect area in the aluminum coil defect binary image at the current moment are used as the centroid coordinates of each stripe defect area;

[0087] Calculate the distance between the centroids of any two stripe defect areas in the aluminum coil defect binary image at the current moment and record it as the centroid distance;

[0088] The average of all the centroid distances in the aluminum coil defect binary image at the current moment is used as the defect area distribution distance of the aluminum coil defect binary image at the current moment;

[0089] The sum of the number of pixels of all stripe defect areas on the aluminum coil defect binary image at the current moment is recorded as the defect area;

[0090] The number of all pixel points of the aluminum coil defect binary image at the current moment is recorded as the binary image area;

[0091] Calculating the ratio of the defect region area to the binary image area as the defect region distribution density of the aluminum coil defect binary image at the current moment;

[0092] The stripe defect distribution index of the aluminum coil defect binary image at the current moment is negatively correlated with the distribution distance of the defect area, and is positively correlated with the distribution density of the defect area.

[0093] It should be understood that a positive correlation means that the dependent variable will increase as the independent variable increases, and the dependent variable will decrease as the independent variable decreases. The specific relationship may be a multiplication relationship, an addition relationship, or the power of an exponential function, which is determined by actual application. A negative correlation means that the dependent variable will decrease as the independent variable increases, and the dependent variable will increase as the independent variable decreases. It may be a subtraction relationship, a division relationship, or a negative exponential power, which is determined by actual application, and this application does not impose any special restrictions on this.

[0094] Preferably, as an embodiment of the present application, the centroid distance can be obtained by calculating the Euclidean distance between the centroids of two stripe defect areas; secondly, the ratio between the defect area distribution density and the defect area distribution distance is used as the stripe defect distribution index of the aluminum coil defect binary image at the current moment.

[0095] It should be understood that this embodiment only provides a method for calculating the distance between two centroids, namely, the Euclidean distance. As other implementation methods, other distance calculation methods in the prior art may be used to determine the distance between two centroids, such as the Manhattan distance, etc. This embodiment does not impose any special restrictions on this.

[0096] Further, in one embodiment of the present application, the step flow chart of the method for obtaining the stripe defect distribution index of the aluminum coil defect binary image at the current moment is as follows: Figure 4 shown.

[0097] It should be noted that the distribution of vertical defects is characterized by the defect area distribution density and the defect area distribution distance. If the defect area distribution distance is smaller and the defect area distribution density is larger, the probability that the vertical defect is a vertical stripe defect is greater. Conversely, if the defect area distribution distance is larger and the defect area distribution density is smaller, the probability that the vertical defect is a vertical line defect is greater. Furthermore, if the stripe defect distribution index of each aluminum coil defect binary map is larger, the probability that it is a vertical stripe defect is greater. Conversely, the probability that it is a vertical line defect is greater.

[0098] At this point, the stripe defect distribution index of the aluminum coil defect binary image at the current moment is obtained to distinguish whether the vertical defect is a vertical stripe defect or a vertical line defect.

[0099] Step 5, based on the stripe defect rotation coefficient and stripe defect distribution index of the aluminum coil defect binary image at the current moment and the angle between each stripe defect area and the horizontal axis, obtain the coating pressure adjustment coefficient of the aluminum coil defect binary image at the current moment.

[0100] Based on the above analysis, the defects in the aluminum coil defect binary map at the current moment are distinguished, and the coating pressure adjustment coefficient is calculated according to the type of defect, which is used to adjust the parameters of the PID algorithm, and then control the roller speed ratio of the roller coater to eliminate stripe defects and improve coating quality. The specific method is as follows:

[0101] Preset an adjustment coefficient, where the value range of the adjustment coefficient is (0,1);

[0102] The difference between the adjustment coefficient and the stripe defect rotation coefficient of the aluminum coil defect binary image at the current moment is used as a factor of the coating pressure adjustment coefficient;

[0103] If the stripe defect rotation coefficient is 1, the stripe defect influence factor is the average rotation angle of the aluminum coil defect binary image at the current moment; if the stripe defect rotation coefficient is 0, the stripe defect influence factor is the stripe defect distribution index of the aluminum coil defect binary image at the current moment;

[0104] The coating pressure adjustment coefficient of the aluminum coil defect binary image at the current moment is positively correlated with the factor of the coating pressure adjustment coefficient and the stripe defect influencing factor.

[0105] Preferably, as an embodiment of the present application, the preset value of the adjustment coefficient is 0.5; the coating pressure adjustment coefficient of the aluminum coil defect binary image at the current moment is the product of the factor of the coating pressure adjustment coefficient and the stripe defect influencing factor.

[0106] It should be noted that when the coating pressure adjustment coefficient factor is a positive value, the stripe defect is a vertical defect, that is, a vertical stripe defect or a vertical line defect occurs, and the roller speed ratio should be increased. When the coating pressure adjustment coefficient factor is a negative value, the stripe defect is a transverse defect, that is, a horizontal stripe defect occurs, and the roller speed ratio should be reduced; if the average value of the rotation angle of all defect areas in the aluminum coil defect binary image at the current moment is larger, the roller speed ratio should be reduced to avoid horizontal stripe defects; if the stripe defect distribution index of the aluminum coil defect binary image at the current moment is larger, the roller speed ratio should be increased to reduce the occurrence of vertical stripe defects. Further, when the coating pressure adjustment coefficient is a negative value, it means that the stripe defect is a horizontal stripe defect, and the roller speed ratio should be reduced. When the coating pressure adjustment coefficient is a positive value, it means that the stripe defect is a vertical defect. The larger the value, the more likely it is a vertical stripe defect, and the roller speed ratio should be increased.

[0107] At this point, the coating pressure adjustment coefficient of the aluminum coil defect binary image at the current moment is obtained.

[0108] Step 6, determine the integral gain parameter in the PID algorithm through the coating pressure adjustment coefficient of the aluminum coil defect binary map at the current moment, and adjust the roller speed ratio of the roller coater during the aluminum coil coating process at the next moment according to the output of the control system to complete the intelligent roller coating of the aluminum coil.

[0109] Furthermore, the PID algorithm is a classic feedback control strategy that can accurately adjust and control the output of the system through the synergy of three parameters: proportional (P), integral (I) and differential (D) to achieve predetermined performance indicators. The advantages of this algorithm are its simplicity, efficiency, ease of implementation and strong adaptability, and it can achieve fast, accurate and stable control in various industrial processes. In controlling the parameters of aluminum coil coating, the PID algorithm can automatically adjust the coating speed and control the uniformity of the coating thickness according to the key indicators monitored in real time, thereby ensuring the consistency of product quality and significantly improving the automation level of the production process and the market competitiveness of the product.

[0110] The PID algorithm has three initial parameters, including proportional gain, integral gain and differential gain. The proportional gain is responsible for responding to the current error and can control the response speed; the integral gain is used to eliminate the steady-state error and adjust the control input by accumulating past errors, which can control the amplitude of the adjustment; the differential gain is responsible for predicting the future behavior of the system and can control the frequency of the adjustment. Therefore, the sum of the coating pressure adjustment coefficient and the initial integral gain coefficient is used as the integral gain parameter of the PID algorithm to reduce the error. The designed control algorithm is deployed in the control system of the roller coater. The controller calculates the control output in real time according to the collected data. According to the output result, the roller speed ratio of the roller coater in the next moment of the aluminum coil coating process is adjusted to complete the intelligent roller coating of the aluminum coil.

[0111] Preferably, as an embodiment of the present application, the proportional gain parameter takes an initial value of 0.02; the initial integral gain coefficient takes a value of 0.02, and the integral gain parameter is the sum of the coating pressure adjustment coefficient and the initial integral gain coefficient; the differential gain parameter takes an initial value of 0.01, wherein the PID algorithm is a well-known technology and the specific process will not be repeated.

[0112] Based on the same inventive concept as the above method, an embodiment of the present application also provides an intelligent coating system for composite coatings on aluminum coils based on computer vision, comprising a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, the steps of any one of the above-mentioned intelligent coating methods for composite coatings on aluminum coils based on computer vision are implemented.

[0113] It should be understood that although Figure 1The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0114] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0115] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the present application. It should be pointed out that, for those of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present application. Therefore, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present application without departing from the content of the technical solution of the present application, shall fall within the protection scope of the technical solution of the present application.

Claims

1. A computer vision-based intelligent coating method for aluminum coil composite coating, characterized in that: The method comprises the following steps: Collect the coating image of the aluminum coil and obtain the binary image of the aluminum coil defect at the current moment; Calculate the stripe defect correction factor according to the aspect ratio of each defect area in the aluminum coil defect binary image at the current moment, and determine the stripe defect area of ​​the aluminum coil defect binary image at the current moment; According to the angle between each stripe defect and the horizontal axis in the aluminum coil defect binary image at the current moment, a stripe defect rotation coefficient of the aluminum coil defect binary image at the current moment is calculated; Calculate the stripe defect distribution index of the aluminum coil defect binary image at the current moment according to the distance between each stripe defect area and the distribution of each stripe defect area in the aluminum coil defect binary image at the current moment; Based on the stripe defect rotation coefficient, the stripe defect distribution index and the angle between each stripe defect area and the horizontal axis of the aluminum coil defect binary image at the current moment, the coating pressure adjustment coefficient of the aluminum coil defect binary image at the current moment is obtained; The integral gain parameter in the PID algorithm is determined by the coating pressure adjustment coefficient of the current aluminum coil defect binary map. According to the output of the control system, the roller speed ratio of the roller coater in the aluminum coil coating process at the next moment is adjusted to complete the intelligent roller coating of the aluminum coil. The calculation method of the stripe defect rotation coefficient of the aluminum coil defect binary image at the current moment is: Calculate the angle between one edge of the minimum circumscribed rectangle of each stripe defect area in the aluminum coil defect binary image at the current moment and the horizontal axis as the rotation angle; The deviation value of the preset rotation angle is recorded as the deviation angle; Calculate the average value of the rotation angles of all stripe defect areas in the aluminum coil defect binary image at the current moment and record it as the average rotation angle; The sum of the average rotation angle and the deviation angle is recorded as the rotation deviation angle; Perform sine transform and cosine transform on the rotation deviation angle in the binary image of aluminum coil defects at the current moment respectively. If the sine transform result is greater than 0, the sine value is assigned to 1; if the sine transform result is less than 0, the sine value is assigned to 0; if the cosine transform result is greater than 0, the cosine value is assigned to 1; if the cosine transform result is less than 0, the cosine value is assigned to 0; if the sine transform result is equal to 0, both the sine value and the cosine value are assigned to 0; if the cosine transform result is equal to 0, the sine value is assigned to 1 and the cosine value is assigned to 0; If the sine value and cosine value have the same value, the fringe defect rotation coefficient is 1; if the sine value and cosine value have different value, the fringe defect rotation coefficient is 0. The step of calculating the stripe defect distribution index of the aluminum coil defect binary image at the current moment includes: According to the distance between each stripe defect area in the aluminum coil defect binary image at the current moment, the defect area distribution distance of the aluminum coil defect binary image at the current moment is calculated; According to the distribution of each stripe defect area in the aluminum coil defect binary image at the current moment, the defect area distribution density of the aluminum coil defect binary image at the current moment is calculated; The stripe defect distribution index of the aluminum coil defect binary image at the current moment is negatively correlated with the distribution distance of the defect area, and positively correlated with the distribution density of the defect area; The calculation method of the defect area distribution distance of the aluminum coil defect binary image at the current moment is: The average value of the horizontal coordinate and the average value of the vertical coordinate of the pixel points with a pixel value of 1 in each stripe defect area in the aluminum coil defect binary image at the current moment are used as the centroid coordinates of each stripe defect area; Calculate the distance between the centroids of any two stripe defect areas in the aluminum coil defect binary image at the current moment and record it as the centroid distance; The average of all the centroid distances in the aluminum coil defect binary image at the current moment is used as the defect area distribution distance of the aluminum coil defect binary image at the current moment; The calculation method of the defect area distribution density of the aluminum coil defect binary image at the current moment is: The sum of the number of pixels of all stripe defect areas on the aluminum coil defect binary image at the current moment is recorded as the defect area; The number of all pixel points of the aluminum coil defect binary image at the current moment is recorded as the binary image area; Calculating the ratio of the defect region area to the binary image area as the defect region distribution density of the aluminum coil defect binary image at the current moment; The method of determining the adjusted integral gain parameter in the PID algorithm, adjusting the roller speed ratio of the roller coater during the coating process of the aluminum coil at the next moment, and completing the intelligent roller coating of the aluminum coil includes: Preset the initial value of the proportional gain parameter of the PID algorithm and the initial value of the differential gain parameter of the PID algorithm; The sum of the coating pressure adjustment coefficient and the preset initial integral gain coefficient is used as the integral gain parameter of the PID algorithm; The PID control algorithm is deployed in the control system of the roller coater to adjust the roller speed ratio of the roller coater to complete the intelligent roller coating of the aluminum coil; The stripe defect correction factor is the average of the aspect ratios of all defect areas in the aluminum coil defect binary image at the current moment; The method for determining each stripe defect area of ​​the aluminum coil defect binary image at the current moment is: If the aspect ratio of the defect region is greater than the stripe defect correction factor, the defect region is a stripe defect region.

2. The intelligent coating method for aluminum coil composite coating based on computer vision according to claim 1, characterized in that: The calculation of the coating pressure adjustment coefficient of the aluminum coil defect binary image at the current moment includes: Preset an adjustment coefficient, where the value range of the adjustment coefficient is (0,1); The difference between the adjustment coefficient and the stripe defect rotation coefficient of the aluminum coil defect binary image at the current moment is used as a factor of the coating pressure adjustment coefficient; If the stripe defect rotation coefficient is 1, the stripe defect influence factor is the average rotation angle of the aluminum coil defect binary image at the current moment; if the stripe defect rotation coefficient is 0, the stripe defect influence factor is the stripe defect distribution index of the aluminum coil defect binary image at the current moment; The coating pressure adjustment coefficient of the aluminum coil defect binary image at the current moment is positively correlated with the factor of the coating pressure adjustment coefficient and the stripe defect influencing factor.

3. A computer vision-based intelligent coating system for aluminum coil composite coating, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the intelligent coating method of aluminum coil composite coating based on computer vision as described in any one of claims 1-2 are implemented.

Images