A fully automatic unmanned machining tool breakage detection system

Abstract

This invention discloses a fully automated, unmanned tool breakage detection system. This invention relates to the field of tool inspection technology and solves the problem of insufficient comprehensiveness in detection caused by using single visual or acoustic detection methods. This invention employs a standardized grayscale algorithm to ensure the consistency of image feature extraction, and combines it with the Sobel algorithm to accurately identify gradient pixels and closed-loop contour regions. Through the verification logic of "image point alignment + rotation comparison," it achieves rapid identification of surface damage such as chipping, notches, and scratches on circular tools, avoiding the missed detection problem caused by contour positioning deviations in traditional visual inspection. The surface damage detection accuracy is ≥99.5%, and the surface inspection time for a single tool is ≤1.5 seconds, adapting to the high-speed production rhythm of unmanned machining lines.

Description

Technical Field

[0001] This invention relates to the field of cutting tool inspection technology, specifically to a fully automated, unmanned cutting tool damage detection system. Background Technology

[0002] In the field of machining, circular cutting tools (such as milling cutters and drills) have become core consumables in fully automated unmanned production lines due to their strong cutting stability and wide applicability. However, during high-speed cutting, these tools are prone to defects such as surface chipping, scratches, and internal microcracks. If these defects are not detected and replaced in time, they can lead to workpiece scrap, equipment damage, or even safety accidents, severely restricting production efficiency and quality. Traditional tool inspection relies primarily on manual visual inspection, depending on operator experience to judge surface defects. This is not only inefficient but also fails to identify internal microcracks, resulting in a missed detection rate of over 8%, making it completely unsuitable for the high-speed production pace of unmanned machining lines. Some existing technologies employ single visual inspection or acoustic inspection solutions. Single visual inspection is susceptible to interference from cutting fluid and metal chips, and its accuracy in identifying minute surface damage is insufficient. Single acoustic inspection lacks linkage with surface features, making it difficult to accurately locate defects, and it requires manual parameter adjustment to adapt to different specifications of round tools, resulting in high adaptation costs. Fully automated unmanned manufacturing has become a core trend in the upgrading of the manufacturing industry, and the demand for automation, precision, and comprehensive tool inspection in production lines is becoming increasingly urgent. Existing inspection technologies suffer from problems such as reliance on manual labor, limitations of single-dimensional inspection, and poor adaptability, which have become key bottlenecks restricting the continuity and stability of unmanned manufacturing production lines.

[0003] Therefore, developing a tool breakage detection system that can be adapted to circular tools, realize full-dimensional detection of surface and internal defects, and achieve fully automated operation of the entire process has important practical significance and industrial value. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a fully automated, unmanned tool breakage detection system, which solves the problem of insufficient detection comprehensiveness caused by using a single visual inspection or acoustic inspection scheme.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a fully automated, unmanned tool breakage detection system, comprising: The visual detection end detects the circular cutting tool, acquires its surface image, performs grayscale processing on the acquired surface image, and transmits the processed grayscale image to the image analysis end. The specific method is as follows: The surface image associated with the circular cutting tool is determined, and the pixel values ​​associated with different pixels within the surface image are identified. Each pixel value is then labeled as X. i, where i represents different pixels; From the corresponding pixel value X i Extract the RGB values ​​and denote them as R. i G i And B i And using: HD i =0.114×R i +0.587×G i +0.299×B i Confirm the grayscale value HD associated with the corresponding pixel. i ; Based on the different gray values ​​associated with different pixels, the corresponding surface image is converted to grayscale, the grayscale image associated with the corresponding surface image is confirmed, and the generated grayscale image is transmitted to the image analysis terminal. In image analysis, the Sobel algorithm is used to sequentially mark the gradient pixels within the grayscale image to determine the contour regions. Multiple sets of contour regions are then analyzed to confirm the consistency between the grayscale image and a set standard image. The specific method is as follows: Based on the different grayscale values ​​associated with different pixels within a grayscale image, the Sobel algorithm is used to determine the vertical and horizontal gradients associated with each pixel, and the following is employed: Confirm the comprehensive gradient associated with the corresponding pixel, and mark the pixel that satisfies: comprehensive gradient ≥ Y1 as gradient pixel, where Y1 is a preset value; otherwise, do not mark it. Connect consecutive gradient pixels to identify the gradient contour lines generated between several groups of consecutive gradient pixels, and identify whether the gradient contour lines are in a closed loop state. If they are completely closed, the region included by the gradient contour lines is determined and recorded as the contour region; otherwise, no processing is performed. Determine the total number of contour regions included in the grayscale image. If there is only one set of contour regions, identify the center point of the contour region and record it as the image fixed point. If there are multiple sets of contour regions, identify the center point of each set of contour regions in turn and connect the center points. If only one set of line segments is generated after connecting, record the midpoint of the line segment as the image fixed point. If a set of polygons is generated after connecting, record the center point inside the polygon as the image fixed point. Based on the image points identified within the grayscale image, the grayscale image is compared with a preset standard image, and the standard image has a pre-marked built-in midpoint: the image points are made to coincide with the built-in midpoint, and the grayscale image is rotated. During the rotation process, it is identified whether there is a process where the grayscale image and the standard image completely coincide. If there is, a second-order test is performed; if not, an abnormal signal on the surface of the qualified part is directly output through the signal output terminal. In the second-order test, acoustic wave detection is performed on the circular cutting tool. Using the image points marked on the surface of the circular cutting tool as the center point, two sets of symmetrical detection paths are identified. The detection waveforms generated by these two sets of symmetrical detection paths are then transmitted to the data analysis terminal. The specific method is as follows: Using the image fixed point marked inside the circular cutter as the center point, a set of contour points are randomly selected on the edge contour of the circular cutter. The line connecting this contour point and the center point is confirmed to generate the first set of single line segments. The single line segments are then extended, and the intersection point generated by the extended segment and the edge contour is confirmed. The line connecting the center point and the intersection point is recorded as the second set of single line segments. The first set of single line segments and the second set of single line segments are recorded as two sets of symmetrical detection line segments. Within the two sets of symmetrical detection line segments, the center point is used as the starting point and the contour point is used as the ending point to generate two sets of symmetrical detection routes associated with the two sets of symmetrical detection line segments. The acoustic detection equipment is used to perform acoustic detection processing based on the determined two sets of symmetrical detection routes to generate the detection waveforms associated with the corresponding detection routes. The generated two sets of symmetrical detection waveforms are then transmitted to the data analysis terminal. Then, the same method is used to confirm the symmetrical detection waveform associated with each contour point. Contour points in the repeating state do not need to be confirmed a second time. All confirmed sets of symmetrical detection waveforms are transmitted to the data analysis terminal. The data analysis end compares and verifies the received multiple sets of symmetrical probe waveforms, identifies whether the waveform features associated with the symmetrical probe waveforms are consistent, and confirms the output signal based on the identification results, and outputs it through the signal output terminal. The specific method is as follows: Two sets of symmetrical detection waveforms are placed in the same spectrum diagram, and the two sets of symmetrical detection waveforms are controlled to move up and down. Different overlapping bands associated with different moving processes are identified. The moving process with the maximum value of the overlapping band line length is recorded as the standard process, and the line length ratio of the overlapping band corresponding to the standard process is confirmed. The line length value of the overlapping segment is recorded as SC1, and the line lengths associated with the two sets of detection waveforms are recorded as B1 and B2 respectively. The ratios of the two sets are confirmed by (SC1÷B1) and (SC1÷B2). The maximum value is selected from the two confirmed ratios, and it is identified whether the maximum value exceeds 90%. If it does, no processing is required. If the maximum value does not exceed 90%, an internal abnormal signal of the tool will be generated directly through the signal output terminal for display.

[0006] This invention provides a fully automated, unmanned tool breakage detection system. Compared with existing technologies, it has the following advantages: By selectively acquiring images of the circular tool surface through a visual detection end, and employing a standardized grayscale algorithm (based on RGB value weighted calculation of grayscale values) to ensure the consistency of image feature extraction, combined with the Sobel algorithm to accurately identify gradient pixels and closed-loop contour regions, and through the verification logic of "image fixed-point alignment + rotation comparison", the system can quickly identify damage such as chipping, notches, and scratches on the surface of the circular tool, avoiding the missed detection problem caused by contour positioning deviation in traditional visual inspection. The surface damage detection accuracy is ≥99.5%, and the surface inspection time for a single tool is ≤1.5 seconds, which is suitable for the high-speed production rhythm of unmanned machining lines. For cutting tools that pass surface inspection, a symmetrical detection route is constructed using a second-order testing end with image positioning as the center. High-frequency ultrasonic waves of 10-20MHz are used for internal detection. Combined with the waveform overlap comparison algorithm of the data analysis end (with 90% overlap as the judgment threshold), waveform anomalies caused by internal microcracks are accurately identified, realizing full-dimensional detection of "surface damage + internal microcracks". This avoids problems such as tool breakage and workpiece scrap caused by missing internal defects due to single surface inspection. The microcrack detection accuracy reaches 0.03mm, and the missed detection rate is ≤0.3%. Attached Figure Description

[0007] Figure 1 This is a schematic diagram of the principle framework of the present invention. Detailed Implementation

[0008] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0009] First Embodiment Please see Figure 1 This application provides a fully automated unmanned tool breakage detection system, including a vision detection end, an image analysis end, a second-order test end, a signal output end, and a data analysis end. The vision detection end, the image analysis end, and the signal output end are electrically connected sequentially from the output node to the input node, and the image analysis end, the second-order test end, and the data analysis end are electrically connected sequentially from the output node to the input node. Among them, the vision detection end detects the circular cutting tool, obtains the surface image of the corresponding circular cutting tool, performs grayscale processing on the obtained surface image, and transmits the processed grayscale image to the image analysis end. Specifically, the cutting tools used in the processing machine tool are basically circular cutting tools, and there are corresponding detection images for the circular cutting tools. By performing grayscale processing on the corresponding detection images, the corresponding grayscale image can be obtained. The specific method for confirming grayscale images is as follows: The surface image associated with the circular cutting tool is determined, and the pixel values ​​associated with different pixels within the surface image are identified. Each pixel value is then labeled as X. i , where i represents different pixels; From the corresponding pixel value X i Extract the RGB values ​​and denote them as R. i G i And B i And using: HD i =0.114×R i +0.587×G i +0.299×B i Confirm the grayscale value HD associated with the corresponding pixel. i ; Based on the different gray values ​​associated with different pixels, the corresponding surface image is converted to grayscale, the grayscale image associated with the corresponding surface image is confirmed, and the generated grayscale image is transmitted to the image analysis terminal.

[0010] In the image analysis section, the Sobel algorithm is used to mark the gradient pixels in the grayscale image in sequence to determine the contour regions in the grayscale image. Multiple contour regions are analyzed to confirm whether the grayscale image is consistent with the set standard image. If they are consistent, the second-order test is performed. If they are not consistent, the signal is output directly through the signal output section. The specific method for determining the inner contour region of the grayscale image is as follows: Based on the different grayscale values ​​associated with different pixels within a grayscale image, the Sobel algorithm is used to determine the vertical and horizontal gradients associated with each pixel (combining the set weights and the surrounding associated pixels to determine the gradient data of the middle pixels, thus completing the gradient data confirmation process), and the following is adopted: Confirm the comprehensive gradient associated with the corresponding pixel, and record the pixel that satisfies: comprehensive gradient ≥ Y1 as gradient pixel, where Y1 is a preset value, and its specific value is determined by the operator based on experience; otherwise, no marking is made. Connect consecutive gradient pixels to identify the gradient contour lines generated between several groups of consecutive gradient pixels, and identify whether the gradient contour lines are in a closed loop state. If they are completely closed, the region included by the gradient contour lines is determined and recorded as the contour region; otherwise, no processing is performed. Determine the total number of contour regions included in the grayscale image. If the total number of contour regions is only one set (at least one set, because the corresponding tool contour will also generate a contour region), then identify the center point of the contour region and record it as the image fixed point (combined with the two-dimensional coordinate system, identify the two-dimensional coordinates associated with different points on the contour of the contour region, and then perform average processing on several sets of two-dimensional coordinates to identify the average point, which is the center point of the corresponding contour region, that is, the corresponding image fixed point). If there are multiple sets of contour regions, then identify the center point of each set of contour regions in turn and connect the center points. If only one set of line segments is generated after connecting, then the midpoint of the line segment is recorded as the image fixed point. If a set of polygons is generated after connecting, then the internal center point of the polygon is recorded as the image fixed point. Based on the image points identified within the grayscale image, the grayscale image is compared with a preset standard image, which has a pre-marked built-in midpoint. The image points are aligned with the built-in midpoint, and the grayscale image is rotated. During the rotation process, it is identified whether the grayscale image and the standard image completely overlap. If they do, a second-order test is performed to conduct a secondary test on the tool, indicating that there is no damage. If not, an abnormal signal on the surface of the qualified part is directly output through the signal output terminal.

[0011] The corresponding standard image has a corresponding midpoint set synchronously. By aligning the two midpoints, it can be confirmed whether there is any abnormality in the corresponding grayscale image. If there is an abnormality, it means that there is a surface abnormality area, which means that the corresponding tool surface is damaged or has other abnormal conditions compared with the standard image. The signal can be displayed directly. If it is normal, it is necessary to re-analyze and test the internal structure of the tool to identify whether there is any damage inside the tool.

[0012] Second Embodiment In the specific implementation process of this embodiment, the two-stage testing process is mainly carried out on the cutting tools with normal surface test results. The main execution end is the second-stage testing end, which performs ultrasonic detection processing on the inside of the cutting tool. In the second-order test end, the circular tool is subjected to acoustic detection processing. The image fixed point marked on the surface of the circular tool is used as the center point to confirm two sets of symmetrical detection routes, and the detection waveforms generated by the two sets of symmetrical detection routes are transmitted to the data analysis end. The specific method for confirming the two sets of symmetrical detection routes is as follows: Using the image fixed point marked inside the circular cutter as the center point, a set of contour points are randomly selected on the edge contour of the circular cutter. The line connecting this contour point and the center point is confirmed to generate the first set of single line segments. The single line segments are then extended, and the intersection point generated by the extended segment and the edge contour is confirmed. The line connecting the center point and the intersection point is recorded as the second set of single line segments. The first set of single line segments and the second set of single line segments are recorded as two sets of symmetrical detection line segments. Within the two sets of symmetrical detection line segments, the center point is used as the starting point and the contour point is used as the ending point to generate two sets of symmetrical detection routes associated with the two sets of symmetrical detection line segments. The acoustic detection equipment is used to perform acoustic detection processing based on the determined two sets of symmetrical detection routes to generate the detection waveforms associated with the corresponding detection routes. The generated two sets of symmetrical detection waveforms are then transmitted to the data analysis terminal. Then, the same method is used to confirm the symmetrical detection waveform associated with each contour point. Contour points in the repeating state do not need to be confirmed a second time. All confirmed sets of symmetrical detection waveforms are transmitted to the data analysis terminal. The data analysis end compares and verifies the received multiple sets of symmetrical detection waveforms, identifies whether the waveform features associated with the symmetrical detection waveforms are consistent, and confirms the output signal based on the identification results, and outputs it through the signal output end. The specific method for identifying whether the waveform characteristics of symmetrical probe waveforms are consistent is as follows: Both sets of symmetrical detection waveforms are placed in the same spectrum diagram, and the two sets of symmetrical detection waveforms are controlled to move up and down. Different overlapping bands associated with different movement processes are identified. The movement process with the maximum value of the overlapping band line length is recorded as the standard process, and the line length ratio of the overlapping band corresponding to the standard process is confirmed. The line length value of the overlapping segment is recorded as SC1, and the line lengths associated with the two sets of detection waveforms are recorded as B1 and B2 respectively. The ratios of the two sets are confirmed by (SC1÷B1) and (SC1÷B2). The maximum value is selected from the two confirmed ratios, and it is identified whether the maximum value exceeds 90%. If it exceeds 90%, no processing is required. If it does not exceed 90%, the tool internal abnormal signal is directly generated through the signal output terminal for display. Specifically, when ultrasonic waves (10-20MHz high frequency) are perpendicularly incident on the surface of the tool, they propagate at a uniform speed along the interior of the material in the intact area and form a "standard reflected waveform" after being reflected by the bottom surface. If there are hidden cracks (internal cracks, interlayer delamination, etc.) in the propagation path, the ultrasonic waves will be "reflected, refracted, and scattered" at the crack interface, resulting in a decrease in propagation speed, energy attenuation, and a change in the reflection path, ultimately forming an "abnormal waveform" that is significantly different from the standard waveform. By extracting the key parameters of the waveform through algorithms and comparing them with a standard library, it can be determined whether there are hidden cracks in the corresponding detection area.

[0013] Some of the data in the above formulas are numerical calculations with dimensions removed, and the contents not described in detail in this specification are all prior art known to those skilled in the art.

[0014] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A fully automated, unmanned tool breakage detection system, characterized in that, include: The visual detection end detects the circular cutting tool, acquires the surface image of the corresponding circular cutting tool, performs grayscale processing on the acquired surface image, and transmits the processed grayscale image to the image analysis end. On the image analysis side, the Sobel algorithm is used to mark the gradient pixels in the grayscale image in sequence to determine the contour regions in the grayscale image. Then, the grayscale image is analyzed at fixed points to confirm whether it is consistent with the set standard image. The second-order test end performs acoustic detection processing on the circular tool. Taking the image fixed point marked on the surface of the circular tool as the center point, it confirms two sets of symmetrical detection routes and transmits the detection waveforms generated by the two sets of symmetrical detection routes to the data analysis end. The data analysis end compares and verifies the received multiple sets of symmetrical detection waveforms, identifies whether the waveform characteristics associated with the symmetrical detection waveforms are consistent, and confirms the output signal based on the identification results, and outputs it through the signal output end.

2. The fully automated unmanned tool breakage detection system according to claim 1, characterized in that, The visual detection end confirms the grayscale image in the following specific way: The surface image associated with the circular cutting tool is determined, and the pixel values ​​associated with different pixels within the surface image are identified. Each pixel value is then labeled as X. i , where i represents different pixels; From the corresponding pixel value X i Extract the RGB values ​​and denote them as R. i G i And B i And using: HD i =0.114×R i +0.587×G i +0.299×B i Confirm the grayscale value HD associated with the corresponding pixel. i ; Based on the different gray values ​​associated with different pixels, the corresponding surface image is converted to grayscale, the grayscale image associated with the corresponding surface image is confirmed, and the generated grayscale image is transmitted to the image analysis terminal.

3. The fully automated unmanned tool breakage detection system according to claim 1, characterized in that, The image analysis terminal determines the contour regions existing within the grayscale image in the following specific way: Based on the different grayscale values ​​associated with different pixels within a grayscale image, the Sobel algorithm is used to determine the vertical and horizontal gradients associated with each pixel, and the following is employed: Confirm the comprehensive gradient associated with the corresponding pixel, and mark the pixel that satisfies: comprehensive gradient ≥ Y1 as gradient pixel, where Y1 is a preset value; otherwise, do not mark it. Connect consecutive gradient pixels to identify the gradient contour lines generated between several groups of consecutive gradient pixels, and identify whether the gradient contour lines are in a closed loop state. If they are completely closed, the region included by the gradient contour lines is determined and recorded as the contour region; otherwise, no processing is performed.

4. The fully automated unmanned tool breakage detection system according to claim 3, characterized in that, The image analysis terminal confirms whether the grayscale image is consistent with the standard image in the following specific way: Determine the total number of contour regions included in the grayscale image. If the total number of contour regions is only one set, then identify the center point of the contour region and record it as the image fixed point. Based on the image points identified within the grayscale image, the grayscale image is compared with a preset standard image, which has a pre-defined built-in midpoint. The image points are aligned with the built-in midpoint, and the grayscale image is rotated. During the rotation process, it is determined whether the grayscale image and the standard image completely overlap. If they do, a second-order test is performed; otherwise, an abnormal signal on the surface of the compliant part is directly output through the signal output terminal.

5. The fully automated unmanned tool breakage detection system according to claim 4, characterized in that, If there are multiple sets of contour regions, the center point of each set of contour regions is identified in turn, and the center points are connected. If only one set of line segments is generated after connecting, the midpoint of the line segment is recorded as the image fixed point. If a set of polygons is generated after connecting, the center point inside the polygon is recorded as the image fixed point.

6. The fully automated unmanned tool breakage detection system according to claim 1, characterized in that, The second-order test terminal confirms the specific method for the two sets of symmetrical detection routes as follows: Using the image fixed point marked inside the circular cutter as the center point, a set of contour points are randomly selected on the edge contour of the circular cutter. The line connecting this contour point and the center point is confirmed to generate the first set of single line segments. The single line segments are then extended, and the intersection point generated by the extended segment and the edge contour is confirmed. The line connecting the center point and the intersection point is recorded as the second set of single line segments. The first set of single line segments and the second set of single line segments are recorded as two sets of symmetrical detection line segments. Within the two sets of symmetrical detection line segments, the center point is used as the starting point and the contour point is used as the ending point to generate two sets of symmetrical detection routes associated with the two sets of symmetrical detection line segments. The acoustic detection equipment is used to perform acoustic detection processing based on the determined two sets of symmetrical detection routes to generate the detection waveforms associated with the corresponding detection routes. The generated two sets of symmetrical detection waveforms are then transmitted to the data analysis terminal. Then, the same method is used to confirm the symmetrical detection waveform associated with each contour point. Contour points in the repeating state do not need to be confirmed a second time. All confirmed sets of symmetrical detection waveforms are transmitted to the data analysis terminal.

7. The fully automated unmanned tool breakage detection system according to claim 6, characterized in that, The specific method by which the data analysis terminal identifies whether the waveform characteristics of the symmetrical detection waveforms are consistent is as follows: Both sets of symmetrical detection waveforms are placed in the same spectrum diagram, and the two sets of symmetrical detection waveforms are controlled to move up and down. Different overlapping bands associated with different moving processes are identified. The moving process with the maximum value of the overlapping band line length is recorded as the standard process, and the line length ratio of the overlapping band corresponding to the standard process is confirmed. The line length value of the overlapping segment is recorded as SC1, and the line lengths associated with the two sets of detection waveforms are recorded as B1 and B2 respectively. The ratios of the two sets are confirmed by (SC1÷B1) and (SC1÷B2). The maximum value is selected from the two confirmed ratios, and it is determined whether the maximum value exceeds 90%. If it does, no processing is required.

8. The fully automated unmanned cutting tool breakage detection system according to claim 7, characterized in that, If the maximum value does not exceed 90%, an internal abnormal signal of the tool will be generated directly through the signal output terminal for display.

Images