Raymond mill grinding roller wear degree detection method and system

Abstract

The invention relates to the field of image processing, in particular to a Raymond mill grinding roller wear degree detection method and system. The method comprises the following steps: carrying out edge detection on an obtained gray-scale map of a grinding roller to obtain edges of a plurality of grinding rollers; calculating defect features of the edge and a defect feature mean value; the Euclidean distance between the mass center of any edge and the mass center of the adjacent edge and the Euclidean distance mean value are calculated; calculating the absolute difference between each Euclidean distance and the Euclidean distance mean value to obtain the deviation of the edge, and carrying out negative correlation normalization on the mean value of all the edge deviations to obtain the distribution regularity of the edge; and calculating the wear degree of the grinding roller according to the defect feature mean value and the distribution regularity, wherein the wear degree of the grinding roller is in positive correlation with the defect feature mean value and is in negative correlation with the edge distribution regularity. The method and the device have the effect of improving the wear degree detection accuracy.

Description

Technical Field

[0001] The present application relates to the field of image processing, and in particular, to a method and system for detecting the wear degree of the grinding roller of a Raymond mill. Background Art

[0002] A Raymond mill is a new type of environmental protection grinding equipment that replaces the ball mill for processing ore powder. This equipment is suitable for grinding ore materials such as limestone, granite, quartzite, and marble. The grinding roller is the core grinding component of the Raymond mill and is usually made of high wear-resistant materials (such as high manganese steel and high chromium steel). It realizes revolution and rotation through a device suspended on the plum blossom frame of the main machine, and cooperates with the grinding ring to roll and crush the material. After the material enters the Raymond mill from the feed inlet, the grinding roller clings to the surface of the grinding ring under the action of centrifugal force, and at the same time, the scraper blade sends the material between the grinding roller and the grinding ring. The rotational movement (revolution and rotation) of the grinding roller generates a rolling pressure to crush the material.

[0003] If the grinding roller is worn, the contact area between the grinding roller and the grinding ring decreases, resulting in a decline in the grinding ability; at the same time, the fineness distribution of the powder particles becomes wider and the quality significantly decreases. Therefore, it is necessary to detect the wear degree of the grinding roller. For example, the application document with the publication number CN117540331A discloses a monitoring and analysis system for the wear degree of the grinding roller of a roller medium-speed coal mill. A static image of the grinding roller of the roller medium-speed coal mill is obtained through a camera and recorded as the grinding roller information image. The edge detection algorithm is used to detect the crack edge contour in the grinding roller information image, the crack part in the grinding roller information image is segmented according to the crack edge contour, and feature extraction is performed on the segmented crack part. The prior art provides a method for analyzing the wear degree through the edge detection algorithm. The change in the edge strength can reflect the wear degree of the grinding roller surface, and the edge strength in the severely worn area will decrease. Or compare the edge positions of the new and old grinding rollers. Wear will cause the edge position to shrink inward.

[0004] Traditional edge detection algorithms can only identify the edge of the grinding roller. When manufacturing the grinding roller of a Raymond mill, in order to fully grind the ore, sandblasting and sand pressing treatments are carried out on the surface of the grinding roller to increase the friction force, and there are also edges on the sand grains on the surface of the grinding roller. Traditional edge detection algorithms cannot effectively identify the edges of the sand grains and the wear edges, resulting in inaccurate wear degree detection results. Summary of the Invention

[0005] In order to solve the technical problem of low accuracy in detecting the wear degree, the present application provides a method and system for detecting the wear degree of the grinding roller of a Raymond mill.

[0006] In the first aspect, the present application provides a method for detecting the wear degree of the grinding roller of a Raymond mill, adopting the following technical solution: A method for detecting the wear of a grinding roller of a Raymond mill comprises the following steps: performing edge detection on a grayscale image of a grinding roller to obtain edges of multiple grinding rollers; calculating defect features of the edges and defect feature means; wherein the calculation of defect features comprises: calculating the similarity between any edge and other edges and the similarity mean; calculating the standard deviation of the similarity and the similarity mean; performing negative correlation normalization on the ratio of the similarity mean to the standard deviation to obtain the defect features of the edges; calculating the Euclidean distance and the mean of the Euclidean distance between the centroid of any edge and the centroid of its adjacent edges; calculating the absolute difference between each Euclidean distance and the mean of the Euclidean distance to obtain the deviation of the edge, performing negative correlation normalization on the means of all edge deviations to obtain the distribution regularity of the edge; and calculating the wear degree of the grinding roller according to the defect feature mean and the distribution regularity, wherein the wear degree of the grinding roller is positively correlated with the defect feature mean and negatively correlated with the edge distribution regularity.

[0007] For the grinding roller, the surface texture is regular and rough before wear, and irregular and smooth after wear. By analyzing the edge feature changes before and after wear of the grinding roller, calculating the mean value of the defect feature and the regularity of the edge distribution of the grinding roller, and combining these two factors to quantify the wear degree of the grinding roller, the actual wear condition of the grinding roller can be more accurately evaluated to avoid equipment failures caused by misjudgment or omission. At the same time, this method can provide a scientific basis for the maintenance and replacement of the grinding roller, help to reasonably arrange maintenance plans, and improve production efficiency.

[0008] Optionally, the expression for the degree of roller wear is: ; In the formula, Indicates the degree of roller wear; Indicates Defect characteristics of the edge; represents the total number of edges; Indicates the regularity of the distribution of all edges; Indicates An exponential function with base .

[0009] Provides a way to quantify the degree of roller wear. The exponential function maps the value of roller wear to In the range of The larger the size (i.e. the more defects), the more regular the distribution The smaller the value (i.e. the more irregular the distribution), the greater the wear degree, and vice versa. The curve obtained by this method is smoother and more sensitive to initial wear.

[0010] Optionally, the expression for the degree of roller wear is: ; In the formula, Indicates the degree of roller wear; Indicates Defect characteristics of the edge; represents the total number of edges; Indicates the distribution regularity of all edges; Indicates the hyperbolic tangent function.

[0011] Provides another way to quantify the wear degree of the grinding roller, which is more non-linear and can better reflect the result of increased wear in the later stage.

[0012] Optionally, the similarity calculation method is as follows: Calculate the distance between each pixel point on the edge and the centroid of the edge and the average distance, and use the standard deviation of the distance and the average distance as the distance change between all pixel points on the edge and the centroid; Use the normalized result of the difference in distance change between any two edges as the similarity between the two edges, and the difference in distance change is negatively correlated with the similarity.

[0013] Perform edge detection on the surface of the grinding roller. If the grinding roller has no wear or little wear, the edges on its surface will be highly similar; otherwise, the similarity of the edges will be low.

[0014] Optionally, the similarity expression is: ; In the formula, Indicates the th edge and the th edge similarity; Indicates the th edge all pixel points and the th edge centroid distance change difference, Indicates the th edge all pixel points and the th edge centroid distance change difference, Indicates function.

[0015] Provides a similarity calculation method. When is larger, that is, the difference in distance change between the two edges is significantly different, the output is closer to 0, indicating that the similarity of the two edges is low; when is smaller, that is, the difference in distance change between the two edges is very close, the output is closer to 0.5, indicating that the similarity of the two edges is high.

[0016] Optionally, the similarity expression is: ; In the formula, Indicates the th edge and the th edge similarity; Indicates the th edge all pixel points and the th edge centroid distance change difference, Indicates the th edge all pixel points and the The distance change difference between the centroids of the strip edges.

[0017] Another method for calculating similarity is provided. When is larger, that is, the distance change differences between the two edges are significantly different, the output is closer to 0, indicating that the similarity between the two edges is lower; when is smaller, that is, the distance change differences between the two edges are very close, the output is closer to 1, indicating that the similarity between the two edges is higher.

[0018] Optionally, the calculation formula for the distribution regularity is: ; in the formula, represents the distribution regularity of all edges; represents the total number of edges; represents the total number of adjacent edges; represents the th edge and the th adjacent edge, the Euclidean distance between the centroids; represents the th edge and the mean Euclidean distance of all adjacent edges; represents the exponential function with as the base.

[0019] represents the mean Euclidean distance between all edges and the centroids of their adjacent edges. The smaller its value, the more uniform the distribution, that is, the more regular the distribution of all edges. At this time, the distribution regularity is larger. On the contrary, the distribution regularity is smaller.

[0020] Optionally, the method for determining adjacent edges is: calculate the Euclidean distance between any edge and the centroids of other edges and construct a Euclidean distance sequence in ascending order, and take the edges corresponding to the first Euclidean distances in the Euclidean distance sequence as the adjacent edges of any edge; or calculate the Euclidean distance between any edge and the centroids of other edges and construct a Euclidean distance sequence in descending order, and take the edges corresponding to the last Euclidean distances in the Euclidean distance sequence as the adjacent edges of any edge.

[0021] Optionally, the canny edge detection algorithm is used for edge detection on the obtained grayscale image of the grinding roller to obtain the edges of multiple grinding rollers.

[0022] In a second aspect, the present application provides a Raymond mill grinding roller wear detection system, adopting the following technical solutions: A Raymond mill roller wear detection system comprises: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the Raymond mill roller wear detection method described above is implemented.

[0023] The above-mentioned Raymond mill roller wear detection method is generated into a computer program and stored in a memory so as to be loaded and executed by a processor. Thus, a system is made based on the memory and the processor for easy use.

[0024] This application has the following technical effects: For the grinding roller, the surface texture is regular and rough before wear, and irregular and smooth after wear. By analyzing the edge feature changes before and after wear of the grinding roller, calculating the mean value of the defect feature and the regularity of the edge distribution of the grinding roller, and combining these two factors to quantify the wear degree of the grinding roller, the actual wear condition of the grinding roller can be more accurately evaluated to avoid equipment failures caused by misjudgment or omission. At the same time, this method can provide a scientific basis for the maintenance and replacement of the grinding roller, help to reasonably arrange maintenance plans, and improve production efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is a flow chart of a method for detecting wear of a Raymond mill roller according to an embodiment of the present application.

[0026] Figure 2 This is a method flow chart of step S2 of a method for detecting wear of a Raymond mill roller in an embodiment of the present application. DETAILED DESCRIPTION

[0027] The present application discloses a method for detecting the wear of a Raymond mill roller. Figure 1 , including steps S1 to S4, which are specifically as follows: S1: Perform edge detection on the acquired grayscale image of the grinding roller to obtain edges of multiple grinding rollers.

[0028] The video stream of the idling grinding roller is collected by a camera, and multiple frames of RGB (Red, Green, Blue) images are obtained by extracting the video stream. The grinding roller area in the RGB image is extracted and converted into a grayscale image. The Canny operator is used to perform edge detection on the grayscale image to obtain the edge detection result.

[0029] When the roller is idling (not rotating for material processing), the video stream including the roller area is collected from the operation port of the Raymond mill housing, with a frame rate of 30fps (frames per second)-60fps. The appropriate frame rate can be selected according to the actual situation. If the roller rotates faster, a higher frame rate is required to ensure that clear images of the roller at different positions can be captured, such as 60fps.

[0030] Determine the frame extraction frequency based on the characteristics of the idle rotation of the grinding roller and the frame rate of the video. If the movement of the grinding roller changes slowly, the frame extraction frequency can be appropriately reduced. For example, for a 30fps video, one frame can be extracted every 3 frames, so that an image sequence of about 10 frames per second can be obtained, which can basically reflect the movement of the grinding roller. If the movement of the grinding roller changes rapidly, a higher frame extraction frequency is required. For example, for a 60fps video, one frame can be extracted every 2 frames to obtain an image sequence of 30 frames per second, which can more accurately capture the movement details of the grinding roller.

[0031] The method for extracting the grinding roller area from the RGB image can be: based on the color segmentation method or the object detection method based on deep learning. The color segmentation method is: the color segmentation method can be used to extract the grinding roller area. The grinding roller is black and the background is white. The grinding roller area can be extracted by setting color thresholds. In the RGB color space, threshold segmentation is performed on the pixel values of the R, G, and B channels respectively. For example, set the threshold of the R channel to be less than a certain value (such as 50), and similar settings are made for the G and B channels, so that a binary image can be obtained, where the grinding roller area is white and the background is black. Then, through image morphological operations (such as dilation, erosion, etc.) to remove noise and fill the holes in the grinding roller area, so as to obtain the complete grinding roller area. The prior art will not be elaborated here.

[0032] Object detection method based on deep learning: For example, use the YOLO (You Only Look Once) algorithm. First, a large number of grinding roller images need to be collected as training data, and these images are labeled to mark the position and category of the grinding roller. Then use these labeled images to train the YOLO model. After training, the frame-extracted images are input into the trained model, and the model can automatically detect the position of the grinding roller and extract the grinding roller area. This method has good adaptability to situations such as large changes in the shape of the grinding roller and complex backgrounds, but requires a large amount of training data and computing resources. The prior art will not be elaborated here.

[0033] After extracting the grinding roller area, convert the RGB image into a grayscale image, and use the Canny edge detection algorithm to perform edge detection on the grayscale image to obtain the edge detection result. The edge detection result is the edges of multiple grinding rollers; the conversion of the edge detection and the grayscale image to the prior art will not be elaborated here.

[0034] S2: Calculate the defect features of the edges and the average value of the defect features.

[0035] Refer to Figure 2 , the calculation of the defect features includes step S20 - step S21, specifically as follows: S20: Calculate the similarity between any edge and other edges and the average value of the similarity.

[0036] When crushing ores, since the crushing of ores relies on the frictional force between the grinding rollers and the materials, the surface of the grinding rollers of the Raymond mill will be subjected to controlled roughening treatment by sandblasting and sand pressing. When sandblasting or sand pressing the grinding rollers, the process parameters will be strictly controlled to make the texture distribution on the surface of the grinding rollers uniform. If the sandblasting pressure is uneven, resulting in random distribution of gravel, the surface roughness distribution of the grinding rollers will be uneven, reducing the stability of the frictional force and causing the materials to slip or accumulate in some areas. Therefore, edge detection is performed on the surface of the grinding rollers. If the grinding rollers have no wear or less wear, the edges on their surfaces will be highly similar; otherwise, the similarity of the edges will be low.

[0037] The calculation method of similarity is as follows: Calculate the distance between each pixel point on the edge and the centroid of the edge and the average value of the distances, and use the standard deviation of the distance and the average value of the distances as the distance variation between all pixel points on the edge and the centroid; Take the normalized result of the difference in distance variation between any two edges as the similarity between the two edges, and the difference in distance variation is negatively correlated with the similarity.

[0038] Specifically, for any one edge, take the average value of the index positions of all pixel points on the edge (the average value of the abscissa and the average value of the ordinate) as the centroid of the edge, and obtain the Euclidean distance between any pixel point on the edge and the centroid.

[0039] Calculate the difference in distance variation between the pixel points and the centroid in any two edges. Specifically, the mathematical formula for the difference in distance variation is: ; In the formula, represents the distance variation situation between all pixel points on the th edge and the centroid; represents the total number of pixel points in the th edge; represents the ordinal number of the pixel point in the th edge; represents the Euclidean distance between the th pixel point in the th edge and the centroid of this edge; represents the average value of the Euclidean distances between all pixel points on the th edge and the centroid of this edge.

[0040] The expression for similarity is: ; In the formula, represents the similarity between the th edge and the th edge; represents the difference in distance variation between all pixel points on the th edge and the centroid of the th edge, Indicates the distance change difference between all pixel points on the th edge and the centroid of the th edge, indicating the

[0041] When is larger, that is, the distance change differences between the two edges are significantly different, the output is closer to 0, indicating a lower similarity between the two edges; when is smaller, that is, the distance change differences between the two edges are very close, the output is closer to 0.5, indicating a higher similarity between the two edges.

[0042] The expression for similarity can also be: ; where represents the similarity between the th edge and the th edge; represents the distance change difference between all pixel points on the th edge and the centroid of the th edge, represents the distance change difference between all pixel points on the th edge and the centroid of the th edge.

[0043] When is larger, that is, the distance change differences between the two edges are significantly different, the output is closer to 0, indicating a lower similarity between the two edges; when is smaller, that is, the distance change differences between the two edges are very close, the output is closer to 1, indicating a higher similarity between the two edges.

[0044] S21: Calculate the standard deviation of the similarity and the similarity mean; perform negative correlation normalization on the ratio of the similarity mean to the standard deviation to obtain the defect feature of the edge.

[0045] The defect feature calculation can be expressed by the following mathematical formula: ; where represents the defect feature of the th edge; represents the total number of edges; represents the edge ordinal number, ; represents the similarity between the th edge and the th edge; represents the similarity mean between the th edge and all other edges.

[0046] Among them, It represents the mean similarity between this edge and other edges. The larger the value, the more edges are similar to this edge. Usually, the defective edge is partially detached or worn, that is, the similarity between the defective edge and most edges is low. Therefore, the larger the value, the smaller the defect feature of the edge.

[0047] is the standard deviation of the similarity and the mean of similarity, indicating the difference in similarity between the edge and other edges. The smaller the value, the smaller the difference, which means that the similarity difference between the edge and other edges is more stable.

[0048] S3: Calculate the Euclidean distance and the mean of the Euclidean distances between the centroid of any edge and the centroid of its adjacent edges; calculate the absolute difference between each Euclidean distance and the mean of the Euclidean distance to obtain the edge deviation, and negatively normalize the means of all edge deviations to obtain the distribution regularity of the edge.

[0049] The formula for calculating the distribution regularity is: ; In the formula, Indicates the regularity of the distribution of all edges; represents the total number of edges; represents the total number of adjacent edges; Indicates Edge and The Euclidean distance between the centroids of adjacent edges; Indicates The mean Euclidean distance between an edge and all adjacent edges; Indicates An exponential function with base .

[0050] The method for determining the adjacent edges is to calculate the Euclidean distance between any edge and the centroid of other edges and construct a Euclidean distance sequence in ascending order. The edges corresponding to the Euclidean distance are taken as the neighboring edges of any edge; for example, .

[0051] Or calculate the Euclidean distance between any edge and the centroid of other edges and construct a Euclidean distance sequence in descending order. The edges corresponding to the Euclidean distance are regarded as the neighboring edges of any edge.

[0052] It represents the mean Euclidean distance between the centroids of all edges and their neighboring edges. The smaller its value is, the more uniform the distribution is, that is, the more regular the distribution of all edges is, the greater the regularity of the distribution is. Conversely, the smaller the regularity of the distribution is.

[0053] S4: Calculate the wear degree of the grinding roller according to the defect feature mean value and the distribution regularity. The wear degree of the grinding roller is positively correlated with the defect feature mean value and negatively correlated with the edge distribution regularity.

[0054] In one embodiment, the expression of the wear degree of the grinding roller can be: ; In the formula, represents the wear degree of the grinding roller; represents the defect feature of the th edge; represents the total number of edges; represents the distribution regularity of all edges; represents the exponential function with as the base.

[0055] The exponential function maps the value of the wear degree of the grinding roller to the range of . When the defect feature mean value is larger (i.e., more defects), and the distribution regularity is smaller (i.e., more irregular distribution), the wear degree is larger. Conversely, the wear degree is smaller. The curve obtained in this embodiment is smoother and more sensitive to initial wear.

[0056] In one embodiment, the expression of the wear degree of the grinding roller can also be: ; In the formula, represents the wear degree of the grinding roller; represents the defect feature of the th edge; represents the total number of edges; represents the distribution regularity of all edges; represents the hyperbolic tangent function. This embodiment has stronger nonlinearity and can better reflect the result of increased wear in the later stage. The calculation formula of the wear degree of the grinding roller can be selected according to the actual application scenario.

[0057] The embodiment of the present application also discloses a detection system for the wear degree of the Raymond mill grinding roller, including a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, the detection method for the wear degree of the Raymond mill grinding roller according to the present application is implemented.

[0058] The above system also includes other components well known to those skilled in the art such as a communication bus and a communication interface. Their settings and functions are known in the art, so they will not be described in detail here.

[0059] In the present application, the foregoing memory may be any tangible medium that contains or stores a program, which can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium may be any suitable magnetic storage medium or magneto-optical storage medium, such as, a resistive random access memory, a dynamic random access memory, a static random access memory, etc., or any other medium that can be used to store the required information and can be accessed by an application program, module, or both. Any such computer storage medium may be part of the device or accessible or connectable to the device.

[0060] The above are all preferred embodiments of the present application, and the protection scope of the present application is not limited thereby. Therefore, all equivalent changes made according to the structure, shape, and principle of the present application shall be covered within the protection scope of the present application.

Claims

1. A method for detecting wear of a Raymond mill roller, characterized in that: Includes steps: Performing edge detection on the acquired grayscale image of the grinding roller to obtain edges of multiple grinding rollers; Calculating the defect characteristics and the mean of the defect characteristics of the edge; wherein the calculation of the defect characteristics includes: calculating the similarity and the mean of the similarity between any edge and other edges; calculating the standard deviation of the similarity and the mean of the similarity; performing negative correlation normalization on the ratio of the mean of the similarity to the standard deviation to obtain the defect characteristics of the edge; Calculate the Euclidean distance and the mean of the Euclidean distance between the centroid of any edge and the centroid of its adjacent edge; calculate the absolute difference between each Euclidean distance and the mean of the Euclidean distance to obtain the edge deviation, and negatively normalize the mean of all edge deviations to obtain the distribution regularity of the edge; The wear degree of the grinding roller is calculated based on the mean value of the defect characteristics and the regularity of the distribution. The wear degree of the grinding roller is positively correlated with the mean value of the defect characteristics and negatively correlated with the regularity of the edge distribution.

2. The method for detecting wear of a Raymond mill roller according to claim 1, characterized in that: The expression of roller wear degree is: ; In the formula, Indicates the degree of roller wear; Indicates Defect characteristics of the edge; represents the total number of edges; Indicates the regularity of the distribution of all edges; Indicates An exponential function with base .

3. The method for detecting wear of a Raymond mill roller according to claim 1, characterized in that: The expression of roller wear degree is: ; In the formula, Indicates the degree of roller wear; Indicates Defect characteristics of the edge; represents the total number of edges; Indicates the regularity of the distribution of all edges; represents the hyperbolic tangent function.

4. The method for detecting wear of a Raymond mill roller according to claim 1, characterized in that: The similarity is calculated as: Calculate the distance and mean distance between each pixel on the edge and the edge centroid, and use the standard deviation of the distance and the mean distance as the distance change between all pixels on the edge and the centroid; The normalized result of the distance change difference between any two edges is taken as the similarity of the two edges, and the distance change difference is negatively correlated with the similarity.

5. The method for detecting wear of a Raymond mill roller according to claim 4, characterized in that: The similarity expression is: ; In the formula, Indicates Edge and similarity of strip edges; Indicates All pixels on the edge of the strip are The difference in distance change between the centroids of the strip edges, Indicates All pixels on the edge of the strip are The difference in distance change between the centroids of the strip edges, express function.

6. The method for detecting wear of a Raymond mill roller according to claim 4, characterized in that: The similarity expression is: ; In the formula, Indicates Edge and similarity of strip edges; Indicates All pixels on the edge of the strip are The difference in distance change between the centroids of the strip edges, Indicates All pixels on the edge of the strip are The difference in distance variation between the centroids of the strip edges.

7. The method for detecting wear of a Raymond mill roller according to claim 1, characterized in that: The formula for calculating the distribution regularity is: ; In the formula, Indicates the regularity of the distribution of all edges; represents the total number of edges; represents the total number of adjacent edges; Indicates Edge and The Euclidean distance between the centroids of adjacent edges; Indicates The mean Euclidean distance between an edge and all adjacent edges; Indicates An exponential function with base .

8. The method for detecting wear of a Raymond mill roller according to claim 1, characterized in that: in, The method for determining the adjacent edges is to calculate the Euclidean distance between any edge and the centroid of other edges and construct a Euclidean distance sequence in ascending order. The edges corresponding to the Euclidean distance are taken as the neighboring edges of any edge; Or calculate the Euclidean distance between any edge and the centroid of other edges and construct a Euclidean distance sequence in descending order. The edges corresponding to the Euclidean distance are regarded as the neighboring edges of any edge.

9. The method for detecting wear of a Raymond mill roller according to claim 1, characterized in that: The canny edge detection algorithm is used to detect the edges of multiple grinding rollers by performing edge detection on the grayscale image of the grinding roller.

10. A Raymond mill roller wear detection system, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the method for detecting the wear degree of a Raymond mill roller according to any one of claims 1 to 9 is implemented.

Images