A tool abnormal state detection method, device, equipment and medium

Abstract

The application discloses a tool abnormal state detection method, device, equipment and medium, relates to the processing detection technical field, and aims to solve the technical problem that the prior art cannot accurately identify the abnormal working condition of an aero part in a processing process. The tool abnormal state detection method comprises the following steps: obtaining an image based on spindle speed data and tool tooth periodic vibration data in a numerical control processing program org ; removing abnormal pixel points in the image org ; identifying a relative distribution position area in the image to obtain an image obj ; performing correction processing and boundary feature image transformation on the image obj ; obtaining an image result ; obtaining multiple images result based on the image part ; quantitatively extracting feature values of the multiple images part ; performing normalization processing on the feature values of the multiple images part ; and obtaining a tool state detection result based on a preset threshold.

Description

Technical Field

[0001] This application relates to the field of machining and inspection technology, and in particular to a method, apparatus, equipment and medium for detecting abnormal states of cutting tools. Background Technology

[0002] Aircraft structural component manufacturing is characterized by small batches and multiple categories. The components are also characterized by complex features, a variety of cutting tools required during processing, variable processing trajectories, and significant influence from complex time-varying conditions. If vibration data signals obtained from sensors are directly used for time-frequency domain feature analysis for monitoring, the accuracy is often low, with many false alarms and missed alarms. This cannot meet the monitoring requirements for abnormal conditions such as tooth breakage and wear during the processing of aircraft structural components.

[0003] Therefore, there is an urgent need for a method to detect abnormal states that occur in cutting tools during the machining of aerospace parts. Summary of the Invention

[0004] The main objective of this application is to provide a method, apparatus, equipment, and medium for detecting abnormal tool conditions, aiming to solve the technical problem that existing technologies cannot accurately identify abnormal working conditions that occur during the machining of aerospace parts.

[0005] To address the aforementioned technical problems, this application proposes a method for detecting abnormal tool states, comprising the following steps:

[0006] Based on the spindle speed data and the periodic vibration data of the cutting teeth in the CNC machining program, an image is obtained. org image;

[0007] The image org After removing abnormal pixels from the image, the relative distribution regions in the image are identified to obtain the image. obj image;

[0008] Regarding the image obj The image undergoes correction processing and boundary feature image transformation to obtain the image. result image;

[0009] Based on the image result Images, obtain multiple images part Images; quantization extraction of multiple images. part The image's feature values; wherein, the feature values ​​include area feature values, length feature values, and width feature values;

[0010] The multiple images partThe feature values ​​of the image are normalized, and the tool condition detection result is obtained based on a preset threshold; wherein, the tool condition detection result includes tool tooth breakage, tool wear, or abnormal condition.

[0011] As some optional embodiments of this application, the image org The image is set with the image center as the initial zero point position, and the periodic vibration data of the cutter teeth are distributed along the radial direction (the rotation speed affects the number of vibration data points corresponding to each cutter tooth along the radial direction per unit time, that is, the faster the rotation speed, the more vibration data points the cutter tooth has, and vice versa).

[0012] As some optional embodiments of this application, the image is... org After removing abnormal pixels from the image, the relative distribution regions in the image are identified to obtain the image. obj Images, including:

[0013] Regarding the image obj The image dimensions are uniformly processed to obtain the first image. org Image; wherein, the image org The length and width of the image are equal.

[0014] For the first image org Abnormal pixels in the image are removed, and relatively distributed regions in the image are identified to obtain the image. obj image.

[0015] As some optional embodiments of this application, the first image org Abnormal pixels in the image are removed, and relatively distributed regions in the image are identified to obtain the image. obj Images, including:

[0016] For the first image org The grayscale value of each pixel in the image is traversed to obtain a first pixel and a second pixel; wherein the first pixel and the second pixel have the same grayscale value, and the chessboard distance between the first pixel and the second pixel is less than the chessboard distance between the first pixel and any other pixel; the other arbitrary pixel refers to any other pixel with the same grayscale value as the first pixel;

[0017] If the first pixel meets the preset threshold, it is identified as an abnormal pixel and removed to obtain the image. dst image;

[0018] Regarding the image dstTo obtain the image, the relative distribution regions in the image are identified. obj image.

[0019] As some optional embodiments of this application, the image is described dst To obtain the image, the relative distribution regions in the image are identified. obj Images, including:

[0020] Based on the cutting tooth value N of the cutting tool t The image dst Image according to 360° / N t Composed of a fan-shaped region, an image is obtained. v image;

[0021] The image v Image and the image dst Images are overlapped and overlaid, and the image is calculated. dst The sum of the number of target pixels in the image v The target pixel refers to a pixel that falls on the line segment that makes up the sector area and whose gray value is equal to the gray value of the abnormal pixel.

[0022] Based on the target pixel count value sum v Get image v-max Image; wherein, the image v-max The number of target pixels in the image meets the preset threshold.

[0023] The image v-max Image and the image dst Image overlay is used to determine the direction of the periodic vibration data of the tool along the principal axis of the vibration data; the image is then used to determine the direction of the periodic vibration data of the tool along the principal axis of the vibration data. dst The center of the image, the image dst Image rotated 360° / (2*N) t After that, the image dst To obtain the image, the relative distribution regions in the image are identified. obj Image; wherein, the relative distribution location region refers to the region where the overlap between the fan-shaped region and the blade vibration data meets a preset threshold.

[0024] As some optional embodiments of this application, the step based on the target pixel count value sum v Get image v-max Images, including:

[0025] Set the tilt angle of the sector region to 0°, and set the size of the array to 360° / N. t Initialize all values ​​in the array array, i.e., int array[360° / N] t ] = {0};

[0026] The image v The image is rotated clockwise around its center by a preset angle, and the image is calculated. v The sum of the number of target pixels in the image v ;Sum the number of target pixels v The index value corresponding to the preset rotation angle is recorded in the array;

[0027] Based on the array, obtain the maximum number of target pixels, sum. v-max ; and the maximum number of target pixels, sum v-max The corresponding preset rotation angle max a Spend;

[0028] The image v Rotate the image clockwise around its center by a maximum of 100%. a Degree, obtain imagev-max image.

[0029] As some optional embodiments of this application, the image is described obj The image undergoes correction processing and boundary feature image transformation to obtain the image. result Images, including:

[0030] The image obj The initial point value of each line segment in the sector region of the image is set to point. first (x,y), and calculate the image respectively. obj The endpoint coordinates of each line segment within the sector region of the image. end (x,y); where the initial point is the geometric center of the image;

[0031] Based on the image obj The endpoint coordinates of each line segment within the sector region of the image. end (x, y), determine the endpoint coordinates of each of the constituent line segments. end (x,y) and the image obj To determine whether the images have intersecting boundaries, a judgment result is obtained.

[0032] Based on the judgment result, an image is obtained.result image.

[0033] As some optional embodiments of this application, the step of obtaining an image based on the judgment result... result Images, including:

[0034] If the determination result is that the endpoints of the constituent line segments are connected to the image obj If the image does not intersect with the boundary, the constituent line segments are extended along the endpoint direction until they intersect with the image boundary;

[0035] If the determination result is that the endpoints of the constituent line segments are connected to the image obj If the images have intersecting boundaries and the endpoints are located exactly on the boundaries of the images, then the line segment remains unchanged;

[0036] If the determination result is that the endpoints of the constituent line segments are connected to the image obj If the images have intersecting boundaries and the endpoints are located outside the images, only the boundaries and the line segments within the boundaries will be retained.

[0037] As some optional embodiments of this application, the image-based result Images, obtain multiple images part Images, including:

[0038] Based on the fan-shaped region, the image result The image is divided into N t A separate area;

[0039] Based on the independent regions, obtain the x-coordinate value of each independent region; based on the x-coordinate value of each independent region, obtain the maximum x-coordinate p. max and the minimum value of the x-axis p min Based on the maximum value p of the horizontal coordinate max and the minimum value of the x-axis p min Obtain the line p max p min ;

[0040] Based on the line p max p min Obtain the first feature point p S1 Second feature point p S2 Wherein, the first feature point p S1 and the second feature point p S2 Located respectively on the line p max p min Both sides of it, and its maximum x-coordinate p max and the minimum value of the x-axis p minThe area formed by the lines connecting the points is greater than any point on the same side; the first feature point p S1 With the maximum value of the horizontal coordinate p max and the minimum value of the x-axis p min The area formed by the lines connecting the points is region S1; the second feature point p S2 With the maximum value of the horizontal coordinate p max and the minimum value of the x-axis p min The area formed by the lines connecting the points is region S2;

[0041] Based on the S1 region and the S2 region, an image is obtained. temp image;

[0042] Based on the image temp Image, obtain image part image.

[0043] As some optional embodiments of this application, the step of obtaining an image based on the S1 region and the S2 region is described. temp Images, including:

[0044] Obtain all lines p in the region S1 min p S1 The first pixel on the left, and all pixels in the S2 region on the straight line p min p S2 The second pixel on the left; based on the first pixel, obtain the vertex of the left half boundary of the S1 region; based on the second pixel, obtain the vertex of the left half boundary of the S2 region; connect the first pixel and the second pixel in each independent region in sequence to obtain a closed region with the largest area characteristic within the fan-shaped region; extract the pixel set corresponding to the outermost contour line of each closed region, and calculate the center coordinates Center(x,y) of the closed region;

[0045] Using the center coordinates Center(x,y) as a reference, search to the left for a set of pixels with the same grayscale value as the center pixel corresponding to the center coordinates; based on the set of pixels with the same grayscale value as the center pixel, obtain the image. temp image.

[0046] As some optional embodiments of this application, the image-based temp Image, obtain image part Images, including:

[0047] Image dst All pixels in the image, based on the image temp Pixels in an image, determine the imagedst Are all pixels in the image related to the image? temp The pixels in the image have corresponding relationships, allowing for the extraction and separation of regions with different features to obtain the image. part image;

[0048] Wherein, if the image dst The pixels to be transformed in the image and the image temp If the pixels in the image have a corresponding relationship, then the grayscale value of the pixel to be transformed remains unchanged; if the image dst The pixels to be transformed in the image and the image temp If there is no corresponding relationship between the pixels in the image, the grayscale value of the pixel to be transformed is transformed into the grayscale value of the background.

[0049] As some optional embodiments of this application, the quantization extraction of multiple images part Image feature values ​​include:

[0050] For multiple images mentioned part The area, length, and width features of the image are extracted to obtain N. t One area characteristic value, one length characteristic value, and one width characteristic value;

[0051] By N t By comparing the area feature value, length feature value, and width feature value, the maximum area feature value, Size, is obtained. max Maximum length characteristic value Length max and the maximum width feature value Width max and the minimum area eigenvalue Size min Minimum length characteristic value Length min and minimum width eigenvalue Width min .

[0052] As some optional embodiments of this application, the multiple images are... part The image feature values ​​are normalized, and based on a preset threshold, the tool state detection results are obtained, including:

[0053] Based on the maximum area feature value Size max The maximum length feature value Length max The maximum width feature value Width max The minimum area feature value Size min The minimum length feature value Length min and the minimum width feature value Widthmin For N t The area feature value, length feature value, and width feature value are normalized.

[0054] Based on preset area threshold, preset length threshold, and preset width threshold, the normalized feature values ​​are extracted to obtain the tool condition detection results.

[0055] To address the aforementioned technical problems, this application also provides: a tool abnormality detection device, comprising:

[0056] The first acquisition module is used to obtain an image based on the spindle speed data and the periodic vibration data of the cutting teeth in the CNC machining program. org image;

[0057] A noise reduction module is used to denoise the image. org After removing abnormal pixels from the image, the relative distribution regions in the image are identified to obtain the image. obj image;

[0058] Transformation module, used for transforming the image obj The image undergoes correction processing and boundary feature image transformation to obtain the image. result image;

[0059] Feature extraction module, used to extract features based on the image result Images, obtain multiple images part Images; quantization extraction of multiple images. part The image's feature values; wherein, the feature values ​​include area feature values, length feature values, and width feature values;

[0060] The detection module is used to detect the multiple images. part The feature values ​​of the image are normalized, and the tool status detection results are obtained based on a preset threshold.

[0061] To address the aforementioned technical problems, this application also proposes an electronic device comprising a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method described above.

[0062] To address the aforementioned technical problems, this application also proposes a computer-readable storage medium storing a computer program, wherein a processor executes the computer program to implement the method described above.

[0063] Vibration signal data generated by cutting tools during aerospace component machining is characterized by high frequency and significant variations in local time domain values. Existing methods cannot adaptively analyze this variable data. Image analysis and recognition technology has been widely applied in various fields, providing a new approach for image-based analysis of vibration data. Images generated from discrete vibration signal data points each correspond to the data of a cutting tool tooth. However, intuitive observation of these discrete points is insufficient to accurately determine their correspondence with the cutting tool teeth; a unified fusion analysis of these data points is necessary. Therefore, to facilitate qualitative analysis of tooth wear and the presence of anomalies such as broken or missing teeth, fusing regional feature data points can effectively reflect the overall condition of the cutting tool, improving recognition accuracy and reducing false alarm rates. This paper compares online detection and recognition algorithms for abnormal signals from various cutting tools, specifically image-based recognition and quantitative analysis algorithms for vibration signals in aerospace structural component machining. This application takes the images generated from the vibration data of the cutting tool as the research object. Based on the transformation, recognition and quantification of the vibration data image features, it judges whether there is any abnormality in the cutting tool. This facilitates the real-time and rapid detection of abnormalities and timely shutdown during the cutting tool processing, so as to avoid damage to the machine tool and mechanical components, and further reduce the economic losses caused by the abnormality of the cutting tool. Attached Figure Description

[0064] Figure 1 This is a schematic flowchart of a tool abnormality detection method according to an embodiment of this application;

[0065] Figure 2 This application relates to embodiments of obtaining vibration data images of the tool machining process based on vibration data;

[0066] Figure 3 This is the fan-shaped area image involved in the embodiments of this application. v Image illustration;

[0067] Figure 4 This is a schematic diagram of the identification result area where the sector region Vs(i) in the embodiments of this application highly overlaps with the vibration data of the cutting teeth;

[0068] Figure 5 The images represent the identification results of each sector region with independent characteristics involved in the embodiments of this application. result Image illustration;

[0069] Figure 6 The embodiments of this application involve imageN1, imageN2, ..., imageN t Image illustration;

[0070] Figure 7This is a schematic diagram of a closed region with the largest area characteristic in a fan-shaped region, as described in an embodiment of this application.

[0071] Figure 8 This is a schematic diagram illustrating the principle and steps of the pixel search method involved in the embodiments of this application;

[0072] Figure 9 This is a schematic diagram of the logic flow of the tool abnormality detection method involved in the embodiments of this application. Detailed Implementation

[0073] It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit this application.

[0074] Aerospace structural components are characterized by small batches, multiple categories, and significant differences in features. They are also generally complex, require a variety of cutting tools to complete the entire machining process, have variable machining trajectories, and are greatly affected by complex time-varying working conditions. Due to these factors, directly analyzing the time-frequency domain features of the tool vibration data signals acquired by sensors to determine whether there are tool abnormalities often results in low accuracy, with many false alarms and missed alarms. This is completely unacceptable for monitoring and identifying tool abnormalities such as broken teeth and wear during the machining of aircraft structural components.

[0075] Vibration signal data is characterized by high frequency and significant variations in values ​​across local time domains. Existing methods cannot adaptively analyze this variable data. Analyzing existing methods for identifying and analyzing tool vibration signals reveals that transforming high-frequency vibration signals, such as normalization, mean, variance, and entropy, is crucial for feature extraction. Based on designed judgment criteria, the entire monitoring process is completed. Image analysis and recognition technology has been widely applied in various fields, providing a new approach for image-based analysis of vibration data. Images generated from discrete vibration signal data points correspond to the data of a tool tooth. However, intuitive observation of these discrete points is insufficient to accurately determine their correspondence with the tool teeth; therefore, a unified fusion analysis of these data points is necessary.

[0076] Therefore, to facilitate qualitative analysis of cutter tooth wear and the presence of anomalies such as broken or missing teeth, integrating regional feature data points can effectively reflect the overall condition of the cutting tool, improve identification accuracy, and reduce false alarm rates. This is comparable to online detection and identification algorithms for various types of cutting tool machining anomaly signals, and image-based identification and quantification analysis algorithms for vibration signals in aerospace structural component machining. This invention uses images generated from cutter vibration data as the research object. Based on the transformation, identification, and quantification of vibration data image features, it determines whether there are anomalies in the cutting tool, enabling real-time and rapid detection of anomalies during machining and timely machine stoppage to avoid damage to machine tools and structural components, further reducing economic losses caused by cutting tool anomalies.

[0077] Specifically, to solve the above-mentioned technical problems, this application proposes a method for detecting abnormal tool states, including the following steps: obtaining an image based on spindle speed data and periodic vibration data of the cutting teeth in a CNC machining program. org Image; the image org After removing abnormal pixels from the image, the relative distribution regions in the image are identified to obtain the image. obj Image; for the image obj The image undergoes correction processing and boundary feature image transformation to obtain the image. result Image; based on the image result Images, obtain multiple images part Images; quantization extraction of multiple images. part The image's feature values; wherein, the feature values ​​include area feature values, length feature values, and width feature values; the multiple images part The feature values ​​of the image are normalized, and the tool condition detection result is obtained based on a preset threshold; wherein, the tool condition detection result includes tool tooth breakage, tool wear, or abnormal condition.

[0078] This application also provides an electronic device for a hardware operating environment, which may include: a processor, such as a central processing unit (CPU), a communication bus, a user interface, a network interface, and memory. The communication bus is used to enable communication between these components. The user interface may include a display screen and an input unit such as a keyboard; optionally, the user interface may also include a standard wired interface or a wireless interface. The network interface may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface). The memory may be high-speed random access memory (RAM) or stable non-volatile memory (NVM), such as a disk drive. The memory may also optionally be a storage device independent of the aforementioned processor.

[0079] Those skilled in the art will understand that the above content does not constitute a limitation on the electronic device, and it may include more or fewer components than illustrated, or combine certain components, or have different component arrangements. For example, a memory, as a storage medium, may include an operating system, a data storage module, a network communication module, a user interface module, and electronic programs. In a specific embodiment, the network interface is mainly used for data communication with a network server; the user interface is mainly used for data interaction with the user; the processor and memory in the electronic device of this application may be located in the electronic device, and the electronic device calls the tool abnormality detection device stored in the memory through the processor and executes the tool abnormality detection method provided in the embodiments of this application.

[0080] Based on this, the present application proposes a method for detecting abnormal tool states, characterized by comprising the following steps:

[0081] Step S10: Based on the spindle speed data and the periodic vibration data of the cutting teeth in the CNC machining program, obtain the image. org image.

[0082] It should be noted that the image org The image uses the image center as the initial zero point and distributes the periodic vibration data of the cutting teeth along the radial direction. For example, Figure 2 As shown, the rotational speed affects the number of vibration data points corresponding to each cutter tooth along the radial direction per unit time. That is, the faster the rotational speed, the more vibration data points the cutter tooth has, and vice versa.

[0083] Step S20: Transfer the image orgAfter removing abnormal pixels from the image, the relative distribution regions in the image are identified to obtain the image. obj image;

[0084] In practical applications, the image is... org After removing abnormal pixels from the image, the relative distribution regions in the image are identified to obtain the image. obj Images, including:

[0085] Step S21: For the image obj The image dimensions are uniformly processed to obtain the first image. org Image; wherein, the image org The length and width of the image are equal.

[0086] It should be noted that the image org The image is a plot containing points drawn from the vibration data of the cutting teeth. The aforementioned unified processing refers to processing all images... obj The image size is set to a preset size, meaning the image length is equal to the image width.

[0087] Specifically, when standardizing the image size, the length and width of the image should be equal. Ncols and Nrows can be used to represent the number of columns and rows of the image, respectively. This set of data can be adjusted according to actual needs. The size of the drawn image data points is represented by the radius R (where R refers to the size of a single pixel, which does not need to be standardized and can be kept the same size). That is, the corresponding R value can also be adjusted according to actual needs. Therefore, the value is not unique and needs to be standardized.

[0088] Furthermore, for ease of understanding, the grayscale value of the corresponding vibration data point in the image below will be represented as 'value'. r value g value b .

[0089] Step S22: For the first image org Abnormal pixels in the image are removed, and relatively distributed regions in the image are identified to obtain the image. obj image.

[0090] Specifically, the above may include:

[0091] Step S221: For the first image orgThe grayscale value of each pixel in the image is traversed to obtain a first pixel and a second pixel; wherein the first pixel and the second pixel have the same grayscale value, and the chessboard distance between the first pixel and the second pixel is less than the chessboard distance between the first pixel and any other pixel; the other arbitrary pixel refers to any other pixel with the same grayscale value as the first pixel;

[0092] Step S222: If the first pixel meets the preset threshold, it is identified as an abnormal pixel and removed to obtain the image. dst image;

[0093] Step S223: For the image dst To obtain the image, the relative distribution regions in the image are identified. obj image.

[0094] Specifically, step S223 describes the image... dst To obtain the image, the relative distribution regions in the image are identified. obj Images, including:

[0095] Step S2231: Based on the cutting tooth value N of the tool t The image dst Image according to 360° / N t Composed of a fan-shaped region, an image is obtained. v image.

[0096] Step S2232: Transfer the image v Image and the image dst Images are overlapped and overlaid, and the image is calculated. dst The sum of the number of target pixels in the image v The target pixel refers to a pixel that falls on the line segment that makes up the sector area and whose gray value is equal to the gray value of the abnormal pixel.

[0097] Step S2233: Based on the target pixel count value sum v Get image v-max Image; wherein, the image v-max The number of target pixels in the image meets the preset threshold.

[0098] In practical applications, step S2233, based on the target pixel count value sum, v Get image v-max Images, including:

[0099] 1) Set the tilt angle of the sector region to 0°, and set the size of the array to 360° / N. t Initialize all values ​​in the array array, i.e., int array[360° / N] t ] = {0};

[0100] 2) The image v The image is rotated clockwise around its center by a preset angle, and the image is calculated. v The sum of the number of target pixels in the image v ;Sum the number of target pixels v The index value corresponding to the preset rotation angle is recorded in the array;

[0101] 3) Based on the array array, obtain the maximum number of target pixels, sum. v-max ; and the maximum number of target pixels, sum v-max The corresponding preset rotation angle max a Spend;

[0102] 4) Transfer the image v Rotate the image clockwise around its center by a maximum of 100%. a Degree, obtain imagev-max image.

[0103] Step S2234: Transfer the image v-max Image and the image dst Image overlay is used to determine the direction of the periodic vibration data of the tool along the principal axis of the vibration data; the image is then used to determine the direction of the periodic vibration data of the tool along the principal axis of the vibration data. dst The center of the image, the image dst Image rotated 360° / (2*N) t After that, the image dst To obtain the image, the relative distribution regions in the image are identified. obj Image; wherein, the relative distribution location region refers to the region where the overlap between the fan-shaped region and the blade vibration data meets a preset threshold.

[0104] It should be noted that the overlap refers to the fact that, visually, the vibration data images of any cutting tooth are symmetrically distributed along the corresponding central axis, and the closer they are to the central axis, the higher the degree of overlap. The periodic vibration data of the cutting tool are distributed along the direction of the main axis of the vibration data, showing significant characteristic differences. This distribution is conducive to the identification of the relative distribution position area, that is, this distribution feature provides data basis for area identification.

[0105] More specifically, the above steps may include:

[0106] For the first image generated org An image, or a single data point, cannot accurately reflect the actual state of the tool during machining. Data points from a region must be fused based on their relative distribution to accurately reflect the tool's condition. This involves removing outlier data from the image. point The (x,y) removal method involves iterating through the image and reading each grayscale value as its value. r value g value b The pixel object_point(x,y) is compared with the nearest grayscale value that satisfies the chessboard distance. r value g value b The distance len between pixels o_point(x,y), if len ≥ the set threshold. len If a point meets the characteristics of an outlier, it should be removed. The corresponding processed image is then obtained. dst image.

[0107] The image dst After removing outlier data points from the image, it is necessary to identify the relative distribution of the blade vibration data corresponding to each region in the image, denoted as Si, i∈[1,N]. t For a tooth with N teeth... t The method for identifying the relative distribution location region Si of the cutting tool's tooth vibration data in the image is as follows:

[0108] 1) Obtain the specific number of teeth N of the cutting tool according to the NC program. t Based on N t The value will be distributed over a range of 2π radians by N. t The sector region Vs(i) composed of line segments is evenly distributed, and the corresponding formula is 2π / N. t It can also be expressed as 360° / N t The fan-shaped area is represented by an image. v Image representation, such as Figure 3 As shown;

[0109] 2) Each sector region Vs(i) can represent the vibration characteristics of the cutting tool, but the initial Vs(i) cannot accurately characterize the corresponding cutting tool vibration data. Therefore, it needs to be corrected. v Image dst Images are overlapped and covered;

[0110] 3) Calculate in image vN in the image falls within the sector region Vs(i) t The grayscale value on the line segment is value r value g value b The number of pixels, denoted as sum v And set the initial sector region Vs(i) tilt angle angle equal to 0;

[0111] 4) Design an intermediate process data recording and storage method based on an integer array (int array[]), where the size of the array is 360° / N. t And taking positive integers, initializing all values ​​to 0, it can be represented as int array[360° / N] t ] = {0};

[0112] 5) Around the image v The angle corresponding to a clockwise rotation of the image from its center position is a unit angle, and sum is calculated. v The numerical values ​​are stored in an array where the index and angle values ​​are equal.

[0113] 6) Process 0-360° / N sequentially t All angles within the range fall on N t The grayscale value on the line segment is value r value g value b The number of pixels, all of which are stored in an array with the same index as the angle value;

[0114] 7) Compare the array array to 360° / N t The size of the values ​​is determined by taking the array index max of the largest value. a value;

[0115] 8) Using max a The value will be a reference value for the original image. v The image is rotated clockwise around its center position. (Image max) a Degree, let the obtained image v-max image;

[0116] 9) Using image v-max Image overlay dst In image dst The part corresponding to the sector-shaped line segment in the image is the part with N blades. t The data generated by the tool vibration signal is in the direction of the vibration data principal axis, in image format. dst Image center will image dstImage rotated 360° / (2*N) t This will yield the recognition result region where the fan-shaped region Vs(i) highly overlaps with the vibration data of the cutting teeth, thus obtaining the image. obj Images, such as Figure 4 As shown.

[0117] Step S30: For the image obj The image undergoes correction processing and boundary feature image transformation to obtain the image. result image.

[0118] Specifically, the above steps may include:

[0119] Step S31: Transfer the image obj The initial point value of each line segment in the sector region of the image is set to point. first (x,y), and calculate the image respectively. obj The endpoint coordinates of each line segment within the sector region of the image. end (x,y); where the initial point is the geometric center of the image;

[0120] Step S32, based on the image obj The endpoint coordinates of each line segment within the sector region of the image. end (x, y), determine the endpoint coordinates of each of the constituent line segments. end (x,y) and the image obj To determine whether the images have intersecting boundaries, a judgment result is obtained.

[0121] Step S33: Based on the judgment result, obtain the image. result image.

[0122] Specifically, step S33 involves obtaining the image based on the judgment result. result Images, including:

[0123] If the determination result is that the endpoints of the constituent line segments are connected to the image obj If the image does not intersect with the boundary, the constituent line segments are extended along the endpoint direction until they intersect with the image boundary;

[0124] If the determination result is that the endpoints of the constituent line segments are connected to the image obj If the images have intersecting boundaries and the endpoints are located exactly on the boundaries of the images, then the line segment remains unchanged;

[0125] If the determination result is that the endpoints of the constituent line segments are connected to the image objIf the images have intersecting boundaries and the endpoints are located outside the images, only the boundaries and the line segments within the boundaries will be retained.

[0126] More specifically, the above steps may include:

[0127] Calculate the image separately obj The coordinates of the endpoints of each line segment in each sector region composed of line segments in the image. end (x, y), all line segments have the same initial point value: point. first (x,y) is the geometric center of the image. obj .cols,image obj .rows).

[0128] In practical applications, in image obj The boundary regions of an image exhibit connectivity between adjacent sectors. Therefore, region recognition and correction, along with transformation of the boundary feature images, are required to determine the endpoints of the line segments that make up the sectors. end (x,y) and image obj Boundary relationships: If the endpoints of the line segments forming the sector are points end (x,y) is not related to the image. obj If image boundaries intersect, extend the corresponding line segments along their endpoints until they are relative to the image boundaries; if the endpoints of the line segments forming the fan shape are points... end (x,y) and image obj If image boundaries intersect and the endpoints are located exactly on the image boundaries, the line segments remain unchanged; if the endpoints of the line segments forming the sector are... end (x,y) and image obj If image boundaries intersect and endpoints are located outside the image, only the line segments within and around the boundaries are retained. Each resulting sector has independent recognition feature characteristics. result Images, such as Figure 5 As shown.

[0129] Step S40: Based on the image result Images, obtain multiple images part Images; quantization extraction of multiple images. part The image's feature values; wherein the feature values ​​include area feature values, length feature values, and width feature values.

[0130] Specifically, the above steps may include:

[0131] Step S41: Based on the fan-shaped region, the image... result The image is divided into Nt Each is an independent region.

[0132] Step S42: Based on the independent regions, obtain the x-coordinate value of each independent region; based on the x-coordinate value of each independent region, obtain the maximum x-coordinate p. max and the minimum value of the x-axis p min Based on the maximum value p of the horizontal coordinate max and the minimum value of the x-axis p min Obtain the line p max p min .

[0133] Step S43, based on the line p max p min Obtain the first feature point p S1 Second feature point p S2 Wherein, the first feature point p S1 and the second feature point p S2 Located respectively on the line p max p min Both sides of it, and its maximum x-coordinate p max and the minimum value of the x-axis p min The area formed by the lines connecting the points is greater than any point on the same side; the first feature point p S1 With the maximum value of the horizontal coordinate p max and the minimum value of the x-axis p min The area formed by the lines connecting the points is region S1; the second feature point p S2 With the maximum value of the horizontal coordinate p max and the minimum value of the x-axis p min The area formed by the lines connecting the points is region S2.

[0134] Step S44: Obtain the image based on the S1 region and the S2 region. temp image.

[0135] Step S44 describes obtaining an image based on the S1 region and the S2 region. temp Images, including:

[0136] 1) Obtain all lines in region S1 on line p min p S1 The first pixel on the left, and all pixels in the S2 region on the straight line p min p S2The second pixel on the left; based on the first pixel, obtain the vertex of the left half boundary of the S1 region; based on the second pixel, obtain the vertex of the left half boundary of the S2 region; it should be noted that the angle between adjacent edges of the vertices connecting the boundary regions of the S1 region is greater than or equal to 90 degrees; use the above recursive method to find points until the line segment with pmax and pmin as endpoints in S2 is found: compare all points in S2 to ensure that the angle between the line segments connecting the endpoints is greater than or equal to 90 degrees, and the area enclosed by the corresponding connecting lines is the largest.

[0137] 2) Connect the first and second pixels in each independent region in sequence to obtain a closed region with the characteristic of maximum area within the fan-shaped region (the closed region is connected by the convex hull rule, and the corresponding area is the largest, that is, the included angle between the closed sides is not less than 90 degrees); extract the set of pixels corresponding to the outermost contour line of each closed region, and calculate the center coordinates Center(x,y) of the closed region.

[0138] 3) Using the center coordinates Center(x,y) as a reference, search to the left for a set of pixels with the same grayscale value as the center pixel corresponding to the center coordinates; based on the set of pixels with the same grayscale value as the center pixel, obtain the image. temp image.

[0139] Step S45: Based on the image temp Image, obtain image part image.

[0140] Specifically, the image-based temp Image, obtain image part Images, including:

[0141] Image dst All pixels in the image, based on the image temp Pixels in an image, determine the image dst Are all pixels in the image related to the image? temp The pixels in the image have corresponding relationships, allowing for the extraction and separation of regions with different features to obtain the image. part image;

[0142] Wherein, if the image dst The pixels to be transformed in the image and the image temp If the pixels in the image have a corresponding relationship, then the grayscale value of the pixel to be transformed remains unchanged; if the image dst The pixels to be transformed in the image and the imagetemp If there is no corresponding relationship between the pixels in the image, the grayscale value of the pixel to be transformed is transformed into the grayscale value of the background.

[0143] Specifically, the quantization extraction of multiple images part Image feature values ​​include:

[0144] For multiple images mentioned part The area, length, and width features of the image are extracted to obtain N. t One area characteristic value, one length characteristic value, and one width characteristic value;

[0145] By N t By comparing the area feature value, length feature value, and width feature value, the maximum area feature value, Size, is obtained. max Maximum length characteristic value Length max and the maximum width feature value Width max and the minimum area eigenvalue Size min Minimum length characteristic value Length min and minimum width eigenvalue Width min .

[0146] More specifically, the above steps may include:

[0147] Based on the fan-shaped region, the image result The image is divided into N t Each region is drawn separately on the same image. result Among images of the same size and number of channels, let the corresponding images be denoted as imageN1, imageN2, ..., imageN. t like Figure 6 As shown.

[0148] For each sub-image imageN1, imageN2, ..., imageN t Find the points where x reaches its maximum and minimum values ​​on the x-axis, and denote them as p. max With p min If it contains multiple maximum or multiple minimum values, randomly select one to construct the line p. max p min .

[0149] On line p max p min Find one side of the line p max p min Find the point furthest from the line, and then each time search outwards for the point furthest from the existing line. If multiple points satisfy the property of being furthest from the line, select that point along with point p.max With p min The point that forms the largest area is the feature point p. S1 Let the region consisting of the three points be S1.

[0150] On line p max p min Find the other side that is away from the line p max p min Find the point furthest from the line, and then each time search outwards for the point furthest from the existing line. If multiple points satisfy the property of being furthest from the line, select that point along with point p. max With p min The point that forms the largest area is the feature point p. S2 Let the region consisting of the three points be S2.

[0151] Find all lines in S1 that lie on line p. min p S1 The points on the left, based on the premise that these points must have vertices forming the left half of the boundary in S1, are searched using the recursive method described above until a point in S1 is found. max and p min The line segment ends at the endpoint.

[0152] Find all lines in S2 that lie on line p. min p S2 The points on the left, based on the premise that these points must have vertices forming the left half of the boundary in S2, are searched using the recursive method described above until a point in S2 is found (p). max and p min The line segment ends at the endpoint.

[0153] Each independent sector region imageN1, imageN2, ..., imageN is connected sequentially. t The points found in the search query can be used to identify the closed region with the largest area within the sector, such as... Figure 7 As shown.

[0154] In the closed images imageN1, imageN2, ..., imageN t Extract each corresponding feature region Ss(i) from the image. The extraction method is as follows: calculate the outermost contour line of each closed region with the largest area characteristic, and save it in the newly created temporary independent region image. temp In the middle, each image temp There is one and only one independent region consisting of a single closed line.

[0155] Image of temporary independent regions tempThe pixels within a closed region are labeled. The geometric center coordinates (Center(x,y)) of the closed region are calculated from the closing lines. These coordinates must lie within the closed region. The angles between the lines forming the closed region are no less than 90°. Based on the design, all pixels within the closed region are located and labeled. Figure 8 As shown, the principle and steps of finding all pixels can include the following:

[0156] 1) Using the geometric center coordinates Center(x,y) as a reference, search for pixels with the same grayscale value as the center point to the left of the center point. The search method is point. next The y value corresponding to (x,y) remains unchanged, while the x value decreases by one pixel unit (px) each time.

[0157] 2) Stop the horizontal search to the left when the x-value decreases to the point where the corresponding gray value is no longer consistent with the gray value of the center coordinates Center(x,y), and update the point. next (x,y) coordinates;

[0158] 3) New search point next For each (x, y) point, the y-value is decreased by one pixel, while the x-value remains unchanged. The corresponding grayscale value is then checked. If the grayscale value matches the Center(x, y) pixel value, the search proceeds horizontally to the right. If the grayscale value does not match the Center(x, y) pixel value, x is increased by one unit pixel. This process continues until a matching grayscale value is found, and the point is updated. next (x,y) coordinates;

[0159] 4) Similarly, use the same method to search for the remaining areas located above the closed area and the corresponding areas below the closed area, and record the coordinate values ​​of all pixels that satisfy the relationship.

[0160] Add grayscale information to all pixels that satisfy the relationship; the corresponding grayscale value is called value. r-p value g-p value b-p The corresponding grayscale value r-p value g-p value b-p If the grayscale value corresponding to the boundary line of the closed region is not equal to the grayscale value, and the corresponding value is not unique or fixed, then a new temporary independent region image with grayscale difference information can be obtained. temp image.

[0161] Determine temporary independent regions in the image. temp The grayscale value is equal to the value.r-p value g-p value b-p All points are based on all points. temp-all (x,y) coordinates of the image dst Perform a transformation on the image. dst If the corresponding pixel is in all points temp-all If the grayscale value of a pixel at a given position (x, y) remains unchanged, then the grayscale value of that region will change to match the grayscale value of the background. Based on this design's method of extracting and separating differential feature regions, the corresponding feature sub-image can be obtained. part .

[0162] The features contained in each feature sub-image are quantized and extracted. The feature values ​​include area feature Size(k), length feature Length(k), and width feature Width(k), where k represents an image. dst The corresponding closed feature region k∈[1,N] in the image t ].

[0163] Size(k) is calculated as follows:

[0164]

[0165] Where rows and cols represent the number of rows and columns of the image, respectively; g represents a temporary variable; h r (x,y),h g (x,y),h b (x, y) represent the grayscale values ​​of the pixel at point (x, y).

[0166] The length (k) and width (k) features are calculated as follows:

[0167] 1) Using an image part Establish a cross-shaped projection coordinate system at the geometric center of the closed region k, with the intersection point coinciding with the center point;

[0168] 2) Project the closed region k along the projection coordinates, and calculate the maximum value num on the projection axes X and Y respectively. m-x ,num m-y With minimum value num n-x ,num n-y ;

[0169] 3) Change the tilt angle of the projection coordinates by rotating one degree at a time, and record the two sets of projection values ​​corresponding to each angle;

[0170] 4) Compare the corresponding (num) values ​​for each angle within the range of 0°-360°. m-x -num n-x )*(num m-y -num n-y ) and take the smallest value among them, and record its num. m-x ,num m-y ,num n-x ,num n-y Corresponding numerical value;

[0171] 5) Compare (num) m-x -num n-x ) and (num m-y -num n-y The size of the number, if (num m-x -num n-x )>(num m-y -num n-y Then the length Length(k) = (num m-x -num n-x Width(k) = (num) m-y -num n-y ), otherwise the length Length(k) = (num m-y -num n-y Width(k) = (num) m-x -num n-x ).

[0172] Step S50: Transfer the multiple images part The feature values ​​of the image are normalized, and the tool condition detection result is obtained based on a preset threshold; wherein, the tool condition detection result includes tool tooth breakage, tool wear, or abnormal condition.

[0173] Specifically, the multiple images part The image feature values ​​are normalized, and based on a preset threshold, the tool state detection results are obtained, including:

[0174] Based on the maximum area feature value Size max The maximum length feature value Length max The maximum width feature value Width max The minimum area feature value Size min The minimum length feature value Length min and the minimum width feature value Width min For N t The area feature value, length feature value, and width feature value are normalized.

[0175] Based on preset area threshold, preset length threshold, and preset width threshold, the normalized feature values ​​are extracted to obtain the tool condition detection results.

[0176] It should be noted that the area, length, and width features are extracted from the normalized feature values ​​and compared with their corresponding thresholds. If the ratio of the minimum to the maximum area of ​​the corresponding tooth is less than or equal to the area threshold, an anomaly is found; otherwise, no anomaly is found. If the length is used as the criterion, if the length of the corresponding tooth feature region is less than or equal to the threshold, an anomaly is found; otherwise, no anomaly is found. If the width is used as the criterion, if the width of the corresponding tooth feature region is less than or equal to the threshold, an anomaly is found; otherwise, no anomaly is found.

[0177] More specifically, in each image org N can be obtained from t Given an area, length, and width value, the maximum value (Size) of each value is obtained by comparing them pairwise. max Length max Width max Minimum value Size min Length min Width min All values ​​are normalized based on their maximum and minimum values, using the formula Size(k) / Size. max Length(k) / Length max Width(k) / Width max .

[0178] Set the area threshold according to the actual testing standards. size Length threshold length Width threshold width And compare individual images respectively. org The extracted normalized values ​​can be used to monitor and identify abnormalities such as tool tooth breakage and wear during the machining of aerospace structural components in real time.

[0179] The logical steps of the above detection method are illustrated in the diagram below. Figure 9 As shown.

[0180] The beneficial effects of this invention are: it facilitates the quantitative analysis of tooth wear and the presence of anomalies such as broken or missing teeth; by integrating regional feature data points, it can effectively reflect the overall condition of the cutting tool, improving the accuracy of identification and reducing false alarm rates. Based on the transformation, identification, and quantification of vibration data image features, this invention determines whether there are anomalies in the cutting tool, enabling real-time and rapid detection of anomalies during machining and timely machine stoppage to avoid damage to the machine tool and structural components, further reducing economic losses caused by cutting tool anomalies. Furthermore, this invention has excellent applicability to cutting tools with different tooth counts; the monitoring algorithm implementation code can be quickly modified and deployed according to the actual tooth count of the cutting tool, reducing development costs.

[0181] To address the aforementioned technical problems, this application also proposes: a tool abnormality detection device, comprising:

[0182] The first acquisition module is used to obtain an image based on the spindle speed data and the periodic vibration data of the cutting teeth in the CNC machining program. org image;

[0183] A noise reduction module is used to denoise the image. org After removing abnormal pixels from the image, the relative distribution regions in the image are identified to obtain the image. obj image;

[0184] Transformation module, used for transforming the image obj The image undergoes correction processing and boundary feature image transformation to obtain the image. result image;

[0185] Feature extraction module, used to extract features based on the image result Images, obtain multiple images part Images; quantization extraction of multiple images. part The image's feature values; wherein, the feature values ​​include area feature values, length feature values, and width feature values;

[0186] The detection module is used to detect the multiple images. part The feature values ​​of the image are normalized, and the tool status detection results are obtained based on a preset threshold.

[0187] The device described in this application can be implemented based on a 64-bit Windows 7 system, an Intel(R) Xeon(R) W-2223 3.60GHz processor, 32GB of RAM, a frequency of 3200MHz, and a 256GB hard disk. The software development platform is based on Visual Studio 2019, the programming language is C++, the corresponding image processing library is OpenCV 4.5.2, it is based on a Debug x64 platform, the deep learning platform is based on the PyTorch framework, Python version 3.7.1, the matrix operation library NumPy version is 1.21.4, the SDK version is 10.0.22000.0, and the plotting tool is Matplotlib.

[0188] It should be noted that each module in the tool abnormality detection device in this embodiment corresponds one-to-one with each step in the tool abnormality detection method in the aforementioned embodiment. Therefore, the specific implementation method and the technical effects achieved in this embodiment can be referred to the implementation method of the aforementioned tool abnormality detection method, and will not be repeated here.

[0189] Furthermore, in one embodiment, this application also provides a computer storage medium storing a computer program, which, when executed by a processor, implements the steps of the methods described in the foregoing embodiments.

[0190] In some embodiments, the computer-readable storage medium may be a memory such as FRAM, ROM, PROM, EPROM, EEPROM, flash memory, magnetic surface memory, optical disk, or CD-ROM; or it may be a device including one or any combination of the above-mentioned memories. The computer may be a variety of computing devices, including smart terminals and servers.

[0191] In some embodiments, executable instructions may take the form of a program, software, software module, script, or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including as a standalone program or as a module, component, subroutine, or other unit suitable for use in a computing environment.

[0192] As an example, executable instructions may, but do not necessarily, correspond to files in a file system. They may be stored as part of a file that holds other programs or data, for example, in one or more scripts in a Hyper Text Markup Language (HTML) document, in a single file dedicated to the program in question, or in multiple collaborating files (e.g., files that store one or more modules, subroutines, or code sections).

[0193] As an example, executable instructions can be deployed to execute on a single computing device, or on multiple computing devices located in one location, or on multiple computing devices distributed across multiple locations and interconnected via a communication network.

[0194] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0195] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0196] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as read-only memory / random access memory, magnetic disk, optical disk) and includes several instructions to cause a multimedia terminal device (which may be a mobile phone, computer, television receiver, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0197] The above-disclosed embodiments are merely partial examples of this application and should not be construed as limiting the scope of this application. Those skilled in the art will understand that implementing all or part of the above embodiments and making equivalent changes in accordance with the claims of this application are still within the scope of the invention.

Claims

1. A method for detecting abnormal states of cutting tools, characterized in that, Includes the following steps: Based on the spindle speed data and the periodic vibration data of the cutting teeth in the CNC machining program, an image is obtained. org image; The image org After removing abnormal pixels from the image, the image is obtained. dst Image; based on the numerical value N of the cutting tool t , will image dst Images according to 360 ° / N t Composed of a fan-shaped region, an image is obtained. v Image; the image v Image and the image dst Images are overlapped and overlaid, and the image is calculated. dst The sum of the number of target pixels in the image v Wherein, the target pixel refers to a pixel that falls on the line segment constituting the sector area and whose grayscale value is equal to the grayscale value of the abnormal pixel; based on the target pixel count value sum v Get image v-max Image; wherein, the image v-max The number of target pixels in the image meets a preset threshold; the image v-max Image and the image dst Image overlay is used to determine the direction of the periodic vibration data of the tool along the principal axis of the vibration data; the image is then used to determine the direction of the periodic vibration data of the tool along the principal axis of the vibration data. dst The center of the image, the image dst Image rotation 360 ° / (2*N t After that, the image dst To obtain the image, the relative distribution regions in the image are identified. obj Image; wherein, the relative distribution location region refers to the region where the overlap between the fan-shaped region and the blade vibration data meets a preset threshold; Regarding the image obj The image undergoes correction processing and boundary feature image transformation to obtain the image. result image; Based on the fan-shaped region, the image result The image is divided into N t Each independent region is defined; based on the independent regions, the x-coordinate value of each independent region is obtained; based on the x-coordinate value of each independent region, the maximum x-coordinate value p is obtained. max and the minimum value of the x-axis p min Based on the maximum value p of the horizontal coordinate max and the minimum value of the x-axis p min Obtain the line p max p min Based on the straight line p max p min Obtain the first feature point p S1 Second feature point p S2 Wherein, the first feature point p S1 and the second feature point p S2 Located respectively on the line p max p min Both sides of it, and its maximum x-coordinate p max and the minimum value of the x-axis p min The area formed by the lines connecting the points is greater than any point on the same side; the first feature point p S1 With the maximum value of the horizontal coordinate p max and the minimum value of the x-axis p min The area formed by the lines connecting the points is region S1; the second feature point p S2 With the maximum value of the horizontal coordinate p max and the minimum value of the x-axis p min The area formed by the lines connecting S1 and S2 is region S2; based on region S1 and region S2, an image is obtained. temp Image; based on the image temp Image, obtain image part Images; quantization extraction of multiple images. part The image's feature values; wherein, the feature values ​​include area feature values, length feature values, and width feature values; The multiple images part The feature values ​​of the image are normalized, and the tool condition detection result is obtained based on a preset threshold; wherein, the tool condition detection result includes tool tooth breakage, tool wear, or abnormal condition.

2. The tool abnormality detection method according to claim 1, characterized in that, The image org The image is set with the image center as the initial zero point, and the periodic vibration data of the cutting teeth are distributed along the radial direction.

3. The tool abnormality detection method according to claim 1, characterized in that, The image org After removing abnormal pixels from the image, the relative distribution regions in the image are identified to obtain the image. obj Images, including: Regarding the image obj The image dimensions are uniformly processed to obtain the first image. org Image; wherein, the image org The length and width of the image are equal. For the first image org Abnormal pixels in the image are removed, and relatively distributed regions in the image are identified to obtain the image. obj image.

4. The tool abnormality detection method according to claim 3, characterized in that, The first image org Abnormal pixels in the image are removed, and relatively distributed regions in the image are identified to obtain the image. obj Images, including: For the first image org The grayscale value of each pixel in the image is traversed to obtain a first pixel and a second pixel; wherein the first pixel and the second pixel have the same grayscale value, and the chessboard distance between the first pixel and the second pixel is less than the chessboard distance between the first pixel and any other pixel; the other arbitrary pixel refers to any other pixel with the same grayscale value as the first pixel; If the first pixel meets the preset threshold, it is identified as an abnormal pixel and removed to obtain the image. dst image; Regarding the image dst To obtain the image, the relative distribution regions in the image are identified. obj image.

5. The tool abnormality detection method according to claim 1, characterized in that, The sum based on the number of target pixels v Get image v-max Images, including: Set the tilt angle of the sector region to 0° and the size of the array to 360. ° / N t Initialize all values ​​in the array array, i.e., int array[360 ° / N t ]={0}; The image v The image is rotated clockwise around its center by a preset angle, and the image is calculated. v The sum of the number of target pixels in the image v ;Sum the number of target pixels v The index value corresponding to the preset rotation angle is recorded in the array; Based on the array, obtain the maximum number of target pixels, sum. v-max ; and the maximum number of target pixels, sum v-max The corresponding preset rotation angle max a Spend; The image v Rotate the image clockwise around its center by a maximum of 100%. a Degree, obtain imagev-max image.

6. The tool abnormality detection method according to claim 1, characterized in that, The image obj The image undergoes correction processing and boundary feature image transformation to obtain the image. result Images, including: The image obj The initial point value of each line segment in the sector region of the image is set to point. first ( x , y ), and calculate the image respectively. obj The endpoint coordinates of each line segment within the sector region of the image. end ( x , y ); where the initial point is the geometric center of the image; Based on the image obj The endpoint coordinates of each line segment within the sector region of the image. end ( x , y ), determine the endpoint coordinates of each of the constituent line segments. end ( x , y ) and the image obj To determine whether the images have intersecting boundaries, a judgment result is obtained. Based on the judgment result, an image is obtained. result image.

7. The tool abnormality detection method according to claim 6, characterized in that, Based on the judgment result, the image is obtained. result Images, including: If the determination result is that the endpoints of the constituent line segments are connected to the image obj If the image does not intersect with the boundary, the constituent line segments are extended along the endpoint direction until they intersect with the image boundary; If the determination result is that the endpoints of the constituent line segments are connected to the image obj If the images have intersecting boundaries and the endpoints are located exactly on the boundaries of the images, then the line segment remains unchanged; If the determination result is that the endpoints of the constituent line segments are connected to the image obj If the images have intersecting boundaries and the endpoints are located outside the images, only the boundaries and the line segments within the boundaries will be retained.

8. The tool abnormality detection method according to claim 1, characterized in that, The image is obtained based on the S1 region and the S2 region. temp Images, including: Obtain all lines p in the region S1 min p S1 The first pixel on the left, and all pixels in the S2 region on the straight line p min p S2 The second pixel on the left; based on the first pixel, obtain the vertex of the left half boundary of the S1 region; Based on the second pixel, obtain the vertex of the left half boundary of the S2 region; The first and second pixels of each independent region are connected sequentially to obtain a closed region with the largest area within the fan-shaped region; the pixel set corresponding to the outermost contour line of each closed region is extracted, and the center coordinates of the closed region (Center) are calculated. x , y ); Center ( x , y Using the center pixel as a reference, search to the left for a set of pixels with the same grayscale value as the center pixel; based on the set of pixels with the same grayscale value as the center pixel, obtain the image. temp image.

9. The tool abnormality detection method according to claim 1, characterized in that, Based on the image temp Image, obtain image part Images, including: Image dst All pixels in the image, based on the image temp Pixels in an image, determine the image dst Are all pixels in the image related to the image? temp The pixels in the image have corresponding relationships, allowing for the extraction and separation of regions with different features to obtain the image. part image; Wherein, if the image dst The pixels to be transformed in the image and the image temp If the pixels in the image have a corresponding relationship, then the grayscale value of the pixel to be transformed remains unchanged; if the image dst The pixels to be transformed in the image and the image temp If there is no corresponding relationship between the pixels in the image, the grayscale value of the pixel to be transformed is transformed into the grayscale value of the background.

10. The tool abnormality detection method according to claim 1, characterized in that, The quantization extraction of multiple images part Image feature values ​​include: For multiple images mentioned part The area, length, and width features of the image are extracted to obtain N. t One area characteristic value, one length characteristic value, and one width characteristic value; By N t By comparing the area feature value, length feature value, and width feature value, the maximum area feature value, Size, is obtained. max Maximum length characteristic value Length max and the maximum width feature value Width max and the minimum area eigenvalue Size min Minimum length characteristic value Length min and minimum width eigenvalue Width min .

11. The tool abnormality detection method according to claim 10, characterized in that, The multiple images part The image feature values ​​are normalized, and based on a preset threshold, the tool state detection results are obtained, including: Based on the maximum area feature value Size max The maximum length feature value Length max The maximum width feature value Width max The minimum area feature value Size min The minimum length feature value Length min and the minimum width feature value Width min For N t The area feature value, length feature value, and width feature value are normalized. Based on preset area threshold, preset length threshold, and preset width threshold, the normalized feature values ​​are extracted to obtain the tool condition detection results.

12. A tool abnormality detection device, characterized in that, The apparatus is used to implement the method as described in any one of claims 1-11, the apparatus comprising: The first acquisition module is used to obtain an image based on the spindle speed data and the periodic vibration data of the cutting teeth in the CNC machining program. org image; A noise reduction module is used to denoise the image. org After removing abnormal pixels from the image, the relative distribution regions in the image are identified to obtain the image. obj image; Transformation module, used for transforming the image obj The image undergoes correction processing and boundary feature image transformation to obtain the image. result image; Feature extraction module, used to extract features based on the image result Images, obtain multiple images part Images; quantization extraction of multiple images. part The image's feature values; wherein, the feature values ​​include area feature values, length feature values, and width feature values; The detection module is used to detect the multiple images. part The feature values ​​of the image are normalized, and the tool status detection results are obtained based on a preset threshold.

13. An electronic device, characterized in that, The electronic device includes a memory and a processor, wherein the memory stores a computer program and the processor executes the computer program to implement the method as described in any one of claims 1-11.

14. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, and the processor executes the computer program to implement the method as described in any one of claims 1-11.

Images