Alloy cutter wear prediction method and system based on machine vision

Abstract

The present application relates to the technical field of computer vision, and particularly relates to an alloy cutter wear prediction method and system based on machine vision. The alloy cutter wear prediction method based on machine vision comprises the following steps: setting two groups of light sources with different illumination angles, defining the two groups of light sources as reference light sources and distinguishing light sources respectively, and making the illumination directions of the two groups of light sources form a preset included angle; before each new alloy cutter is put into use, controlling the reference light source to irradiate the alloy cutter alone, collecting a reference image of the alloy cutter in a non-worn state as an individual initial baseline image of the alloy cutter and storing the individual initial baseline image in association with an alloy cutter identifier. Through the machine vision technology, individual initial baseline comparison and double-source differential imaging, the present application realizes separation of a wear area and an interference area and online closed-loop monitoring, thereby improving the accuracy of alloy cutter wear identification.

Description

Technical Field

[0001] This invention relates to the field of computer vision technology, and in particular to a method and system for predicting wear of alloy cutting tools based on machine vision. Background Technology

[0002] In the field of machining, the wear condition of alloy cutting tools directly affects machining accuracy and production stability. Wear monitoring based on machine vision has become a common method, but existing technologies still have significant shortcomings. Traditional methods do not consider the manufacturing differences of the cutting tools themselves, which easily introduces inherent recognition biases and cannot truly reflect the actual wear changes of the cutting tools. At the same time, single-light source imaging is difficult to distinguish between tool material wear and surface deposits such as chips and oil stains, often resulting in misjudgments and omissions, and the recognition results are not stable enough.

[0003] In existing visual monitoring solutions, image acquisition is independent of the cutting process, often requiring downtime for acquisition, which disrupts production continuity. Even with online acquisition, the lack of coordination between lighting and timing results in poor image quality and data reliability. Furthermore, multi-source image data lacks standardized timing synchronization, coordinate matching, and effective verification mechanisms, leading to messy data, ambiguous correspondences, and difficulty in accurately locating and quantitatively analyzing wear areas. Most systems cannot automatically integrate data, differentiate wear patterns, and generate standardized reports, relying on manual judgment. This results in a fragmented workflow, limited practicality, and an inability to meet the high-precision, online, and automated tool wear prediction requirements of industrial settings. Summary of the Invention

[0004] Therefore, it is necessary to provide a machine vision-based method and system for predicting wear of alloy cutting tools in order to solve at least one of the above-mentioned technical problems.

[0005] To achieve the above objective, a machine vision-based method for predicting wear of alloy cutting tools is provided, comprising the following steps: Step S1: Set up two sets of light sources with different illumination angles, define the two sets of light sources as the reference light source and the distinguishing light source, and make the illumination directions of the two sets of light sources form a preset angle; Step S2: Before each new alloy tool is put into use, the reference light source is controlled to illuminate the alloy tool individually, and a reference image of the alloy tool in its unworn state is acquired as the individual initial baseline image of the alloy tool and associated with the alloy tool identification for storage. Step S3: For the alloy cutting tool to be predicted, control the reference light source and the differentiation light source to alternately irradiate it in a time-sharing manner, and collect the real-time reference image and the real-time differentiation image of the alloy cutting tool to be predicted respectively. The timing of the alternating irradiation is synchronized with the cutting interval cycle of the alloy cutting tool. Step S4: Call up the individual initial baseline image corresponding to the alloy tool to be predicted, compare the real-time reference image with the individual initial baseline image, and combine the differences between the real-time distinguishing image and the real-time reference image to complete the wear prediction of the alloy tool to be predicted.

[0006] The present invention also provides a machine vision-based alloy tool wear prediction system for performing the above-described machine vision-based alloy tool wear prediction method. The machine vision-based alloy tool wear prediction system includes: The light source configuration module is used to set two sets of light sources with different illumination angles. The two sets of light sources are defined as a reference light source and a distinguishing light source, so that the illumination directions of the two sets of light sources form a preset angle. The baseline image acquisition and storage module is used to control the reference light source to illuminate the alloy tool individually before each new alloy tool is put into use, acquire a reference image of the alloy tool in its unworn state, and store it as the individual initial baseline image of the alloy tool and associate it with the alloy tool identification. The real-time image acquisition module is used to control the reference light source and the differentiation light source to alternately irradiate the alloy cutting tool to be predicted in use, and to acquire the real-time reference image and the real-time differentiation image of the alloy cutting tool to be predicted, respectively. The timing of the alternating irradiation is synchronized with the cutting interval cycle of the alloy cutting tool. The wear prediction module is used to call up the individual initial baseline image corresponding to the alloy tool to be predicted, compare the real-time reference image with the individual initial baseline image, and combine the differences between the real-time distinguishing image and the real-time reference image to complete the wear prediction of the alloy tool to be predicted.

[0007] The beneficial effects of the present invention are as follows: On the one hand, this invention uses the unworn state of the tool itself as the sole comparison benchmark. By aligning and comparing the real-time benchmark image with the individual initial baseline image region by region and pixel by pixel, it can truly reflect the surface changes that occur during the use of the tool, avoiding recognition deviations caused by individual differences in tools, environmental changes, or universal templates. At the same time, the real-time distinguishing image is compared with the real-time benchmark image at the same location and scale. The brightness and darkness variation patterns of the same wear area under different lighting conditions are extracted and analyzed, so that the missing material of the tool and the attachment of foreign objects on the surface form a distinguishable and stable difference in image features. Thus, without relying on complex algorithms and human experience, reliable separation of wear areas and interference areas is achieved, ensuring that the wear determination is determined only by the actual changes in the state of the tool, thereby improving the stability and consistency of the recognition results.

[0008] On the other hand, this invention performs time-series synchronization, coordinate matching, deduplication, and validity verification on multi-source image difference data to form a structurally complete and clearly correlated difference integration dataset. This dataset is then standardized according to the regional division of the tool cutting edge, rake face, and flank face, achieving accurate extraction and orderly classification of difference information. Based on this, it completes the location of suspected wear areas, actual size conversion, area labeling, and data statistics, and automatically generates a prediction report containing a unique tool identifier and detailed wear parameters, forming a closed loop in the entire wear analysis process from data acquisition to result output. Simultaneously, the dual-light source alternating illumination sequence is synchronized with the cutting interval cycle, ensuring that image acquisition is fully adapted to the machining rhythm. Status monitoring can be completed without machine downtime, guaranteeing analytical accuracy and data standardization while allowing the entire method to be stably embedded into actual production processes, improving the continuity and practicality of tool status monitoring. Attached Figure Description

[0009] Figure 1 This is a schematic diagram of the steps in a machine vision-based method for predicting wear of alloy cutting tools. Figure 2 This is a schematic diagram of the alloy cutting tool in this embodiment; Figure 3 This is a schematic diagram of the illumination from the two sets of light sources in this embodiment; Figure 4 This is a diagram showing the material wear state at the cutting edge of an alloy cutting tool. The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0010] The following description, in conjunction with the accompanying drawings, clearly and completely describes the machine vision-based alloy tool wear prediction method of the present invention. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0011] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, or in one or more hardware modules or integrated circuits, or in machine vision-based alloy tool wear prediction methods and / or microcontroller-based alloy tool wear prediction methods using different networks and / or processors.

[0012] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0013] To achieve the above objectives, please refer to Figures 1 to 4 A machine vision-based method for predicting wear of alloy cutting tools, comprising the following steps: Preferably, step S1: set two sets of light sources with different illumination angles, define the two sets of light sources as a reference light source and a distinguishing light source, and make the illumination directions of the two sets of light sources form a preset angle; Optionally, in step S1, setting two sets of light sources with different illumination angles specifically involves: Two sets of diffuse reflection cold light sources with the same power were selected as the reference light source and the distinguishing light source, respectively. Adjust the emitting surface of the reference light source to be parallel to the main cutting plane of the carbide tool, and adjust the emitting surface of the distinguishing light source to be parallel to the back face of the carbide tool. Adjust the installation height of the two sets of light sources so that the vertical distance between the light source and the cutting edge of the alloy tool is controlled between 100mm and 150mm. Adjust the illumination direction of the two sets of light sources so that the preset angle between them is 45°-60°, and the illumination area of ​​both sets of light sources covers the tool wear monitoring area.

[0014] In this embodiment, setting two sets of light sources with different illumination angles is specifically as follows: two sets of diffuse reflection cold light sources with a power of 50W are selected. The diffuse reflection cold light source uses LED light-emitting chips, the color temperature is set to 5500K, and the color rendering index is ≥90. One set of diffuse reflection cold light sources is used as the reference light source, and the other set of diffuse reflection cold light sources is used as the distinguishing light source.

[0015] Please see Figure 2 It should be noted that the cutting edge of the carbide tool is located at the very tip of the tool, which is the main working part for cutting the workpiece and also the area most prone to wear. It is a key monitoring point during image acquisition. The rake face of the carbide tool is the inclined surface at the top of the tool, along which chips are discharged. Wear will affect the smoothness of chip removal, making it one of the key areas covered by the two sets of light sources. The flank face of the carbide tool is the vertical surface on the side of the tool, in contact with the workpiece surface. Wear will affect machining accuracy, making it the area where the light source focuses its illumination.

[0016] Two sets of diffuse reflection cold light sources are fixed using an angle adjustment bracket. By adjusting the angle adjustment knob of the bracket, the emitting surface of the reference light source is adjusted to be completely parallel to the main cutting plane of the carbide tool, with the adjustment accuracy controlled within ±0.5°. At the same time, the emitting surface of the distinguishing light source is adjusted to be completely parallel to the back face of the carbide tool. After adjustment, the bracket fixing bolts are tightened. The installation height of the two sets of diffuse reflection cold light sources is adjusted by the height adjustment slide rail. A digital height gauge is used to monitor the vertical distance between the light source and the cutting edge of the carbide tool in real time, and the vertical distance is precisely controlled at 120mm, with a height adjustment accuracy of ±1mm.

[0017] Adjust the illumination direction of the two sets of diffuse cold light sources by adjusting the light source angle knob. Use an angle measuring instrument to measure the angle between the illumination directions of the two sets of light sources in real time. Precisely adjust the preset angle to 50°. After adjustment, lock the angle adjustment knob to ensure that the illumination area of ​​the reference light source completely covers the wear monitoring parts of the cutting edge, rake face, and flank face of the alloy tool. The overlap rate between the illumination area of ​​the distinguishing light source and the illumination area of ​​the reference light source at the wear monitoring part is ≥95%. The illumination spot diameter of both sets of light sources is set to 80mm, and the spot uniformity is ≥90% to ensure uniform illumination of the wear monitoring part and no obvious spot edge distortion.

[0018] Optionally, two sets of diffuse cold light sources with the same power are selected as the reference light source and the distinguishing light source, respectively, including: The diffuse reflection cold light source facing directly above the cutting edge of the alloy tool is defined as the reference light source; Adjust the illumination angle of the reference light source so that the illumination area completely covers all wear monitoring areas of the alloy cutting tool, including the cutting edge, rake face, and flank face. The diffuse reflection cold light source facing the side of the rake face of the alloy tool is defined as the distinguishing light source; Adjust the illumination path of the distinguishing light source so that the illumination path forms a preset angle with the mirror reflection path of the alloy tool cutting edge; Adjust the illumination range of the distinguishing light source so that the overlap rate between the illumination range and the illumination area of ​​the reference light source at the wear monitoring part reaches a preset ratio.

[0019] In this embodiment, two sets of diffuse reflection cold light sources, each with a power of 50W, are selected. These diffuse reflection cold light sources use high-brightness LED light-emitting chips with a fixed color temperature of 5500K, a color rendering index of ≥90, and a light-emitting surface size of 100mm×100mm. The two sets of diffuse reflection cold light sources are used as the reference light source and the distinguishing light source, respectively. They are fixed on both sides of the processing station using adjustable metal brackets. The bottom of the brackets is fixed to the worktable surface with expansion bolts to ensure that the light sources are installed stably without shaking.

[0020] Please see Figure 3Define the diffuse reflection cold light source directly above the cutting edge of the alloy tool as the reference light source. Rotate the bracket angle adjustment knob to adjust the illumination angle of the reference light source. Use a digital angle meter to monitor the illumination angle in real time and control the adjustment accuracy within ±0.5°. At the same time, adjust the illumination focal length through the light source focal length adjustment ring so that the illumination area of ​​the reference light source completely covers all wear monitoring parts of the cutting edge, rake face, and flank face of the alloy tool. Set the illumination spot diameter to 80mm and the spot uniformity to ≥90%. After adjustment, tighten the bracket angle locking bolt.

[0021] The diffuse reflection cold light source facing the side of the flank face of the carbide tool is defined as the distinguishing light source. The horizontal position of the distinguishing light source is adjusted by the sliding rail of the bracket so that the illumination path of the distinguishing light source is aligned with the flank face of the carbide tool. The specular reflection detector is used to locate the specular reflection path of the cutting edge of the carbide tool. The illumination angle of the distinguishing light source is adjusted so that the actual illumination path and the specular reflection path of the cutting edge of the carbide tool form a preset angle of 18°, with an adjustment accuracy of ±0.3°.

[0022] Rotate the beam adjustment knob of the distinguishing light source to adjust its illumination range. Use the area overlap measuring instrument to detect the overlap rate of the irradiated areas of the distinguishing light source and the reference light source in the wear monitoring area in real time. Adjust the overlap rate precisely to the preset ratio of 96%. After adjustment, lock the beam adjustment knob to ensure that there is no obvious omission in the irradiated area of ​​the two light sources in the wear monitoring area and no extra irradiated area to interfere with the monitoring.

[0023] Preferably, in step S2: before each new alloy tool is put into use, the reference light source is controlled to illuminate the alloy tool individually, and a reference image of the alloy tool in its unworn state is acquired as the individual initial baseline image of the alloy tool and associated with the alloy tool identifier for storage. Optionally, step S2, controlling the reference light source to irradiate the alloy tool individually, includes: Adjust the illuminance of the reference light source to a stable output range, and keep the output brightness of the light source stable without fluctuation; Fix the unused alloy cutting tool on the positioning fixture to keep the alloy cutting tool in a horizontal and static state; Adjust the orientation of the alloy cutting tool so that the cutting edge of the alloy cutting tool is directly facing the center of the light-emitting surface of the reference light source; Turn on the reference light source and irradiate continuously for a duration that meets the requirement of uniform light distribution on the surface of the alloy tool.

[0024] In this embodiment, the illuminance adjustment knob of the reference light source is turned to the calibration position, and the illuminance value of the cutting edge surface of the alloy tool is monitored in real time by a digital illuminance meter. The illuminance is stably adjusted to the output range of 700 lux, the constant current drive mode of the light source is turned on, the drive current is set to 1.2A, the output brightness of the light source is kept stable without fluctuation, the monitoring time is not less than 5 minutes, and it is confirmed that the illuminance fluctuation range does not exceed ±5 lux.

[0025] Unused alloy cutting tools are loaded into a special positioning fixture. The fixture uses a three-jaw chuck structure with a chuck clamping force set to 800N. The position is adjusted by adjusting the horizontal adjustment screw of the fixture. The horizontality of the alloy cutting tool is checked with a level and the adjustment accuracy is controlled within ±0.1° to ensure that the alloy cutting tool is in a horizontal and stationary state. The fastening bolts of the fixture are then tightened to prevent the tool from shifting during subsequent operations.

[0026] The pitch angle of the alloy tool is adjusted by the attitude adjustment mechanism of the fixture. A positioning beam is emitted by the laser positioning instrument and projected onto the center of the light-emitting surface of the reference light source. The tool position is finely adjusted until the cutting edge of the alloy tool is completely aligned with the positioning beam. At this time, the cutting edge of the alloy tool is directly facing the center of the light-emitting surface of the reference light source. A dial indicator is used to detect the alignment deviation between the cutting edge and the center of the light-emitting surface. The deviation value is controlled within ±0.2mm.

[0027] Start the timed illumination program of the reference light source, set the illumination duration to 3 seconds, keep the relative position of the positioning fixture and the light source unchanged during the illumination process, and ensure that the illumination duration meets the requirement of uniform light distribution on the surface of the alloy tool. After the illumination is completed, turn off the reference light source to complete the operation process of illuminating the alloy tool with the reference light source alone.

[0028] Of particular importance is that the reference image of the unworn alloy tool acquired in step S2 is specifically as follows: During the middle period of continuous illumination by the reference light source, the image acquisition component is activated; Images of the alloy cutting tool are continuously acquired at a fixed frequency, and the resolution and pixel depth of the acquired images are set as conventional monitoring adaptation parameters. Adjust the acquisition angle so that each frame of the acquired image fully includes the cutting edge, rake face, and flank face of the alloy tool. Multiple frames of images are continuously acquired to form a reference image sequence of the alloy tool in its unworn state.

[0029] In this embodiment, during the middle period from the first second to the second second of continuous illumination by the reference light source, the industrial area scan camera is activated as an image acquisition component. The photosensitive chip of the industrial area scan camera is 1 / 1.8 inches in size, and the pixel array specification is 2592×1944. It is connected to the image storage unit through the camera data cable to ensure stable image data transmission.

[0030] Set the acquisition frequency of the industrial area scan camera to a fixed frequency of 30 frames / second, set the image resolution to 2592×1944, and the pixel depth to 8-bit for standard monitoring and adaptation parameters. Enable the camera's automatic gain control off mode, fix the gain value to 1.0, and set the exposure time to 200μs to avoid overexposure or underexposure of the image.

[0031] Adjust the acquisition angle of the industrial area array camera using the three-dimensional adjustment mechanism of the camera bracket. Adjust the horizontal rotation angle of the camera to 0° and the vertical tilt angle to 30°. Use a calibration plate to calibrate the camera acquisition area to ensure that each frame of the acquired image fully contains the cutting edge, rake face, and flank face of the alloy tool, and that the tool area occupies no less than 40% of the total image area. After adjustment, tighten the locking nut of the camera bracket.

[0032] The industrial area array camera is controlled to continuously acquire 60 frames of images according to the set parameters. During the acquisition process, the illuminance of the reference light source is kept stable at 700 lux. The acquired image data is stored sequentially to the image storage unit according to the acquisition time sequence. The storage format is lossless compressed TIFF format. The image naming rule is "tool number-reference image-frame number", which finally forms a reference image sequence of the alloy tool in the unworn state.

[0033] Of particular importance is that the storage of the initial baseline image of the alloy tool as an individual and associated with the alloy tool identifier in step S2 specifically involves: The acquired reference image sequences are sorted sequentially according to the acquisition time sequence; The image sharpness was checked frame by frame, and images with clear outlines of alloy cutting tools and no lighting distortion were selected. Extract the intermediate frame image from the filtered image sequence and determine the intermediate frame image as the individual initial baseline image of the alloy tool; Read the unique serial number, model specification, and production batch information of the alloy cutting tool as the tool identification information; The initial baseline image of each individual is associated with the tool identification information, and the associated image is stored in a lossless compression format in a dedicated storage directory.

[0034] In this embodiment, the 60 frames of reference image sequence are sorted sequentially from frame 1 to frame 60 according to the acquisition time sequence. The sorting is based on the acquisition timestamp information of the image file, with a timestamp accuracy of 1ms, to ensure that the temporal order of the image sequence is without deviation. After sorting, a temporal sorting list is generated, which includes the frame number, acquisition time, and image size information of each image.

[0035] The sharpness of each frame of the sorted baseline image sequence is detected. The sharpness value of each frame is calculated using an image sharpness evaluation algorithm. The sharpness threshold is set to 85. Images with sharpness values ​​higher than the threshold are selected. At the same time, images with illumination distortion, missing pixels, or blurred edges are manually checked and removed frame by frame. After filtering, 45 valid images are retained to form a valid image subset. The frame number range of the valid images is recorded as frame 5 to frame 49.

[0036] The middle frame image of the effective image subset, namely the 27th frame image, is extracted and determined as the individual initial baseline image of the alloy tool. The resolution of this frame image is 2592×1944, the pixel depth is 8bit, the image format is TIFF, the image size is 14.5MB, and the grayscale difference of the edge of the alloy tool in the image is ≥30 to ensure that the outline is clear and distinguishable.

[0037] The unique serial number, model specification, and production batch information of the alloy tool are read by a barcode scanner. The serial number is a seven-digit code (e.g., 1234567), the model specification is "carbide-edge angle 15°-cutting length 80mm", and the production batch information is a six-character code (e.g., 202401). The above information is integrated into the tool identification information and an identification code is generated.

[0038] The filenames of individual initial baseline images are associated with tool identification information, and the association rule is "factory number-model specification-production batch-baseline image". The associated images are stored in a lossless compression format on a solid-state drive (SSD) with a storage capacity of no less than 512GB and a read / write speed of no less than 300MB / s. At the same time, a dedicated storage directory is created on the SSD, named after the tool's factory number, to ensure that the baseline image of each alloy tool is stored independently. The storage path is "SSD-alloy tool baseline image library-factory number directory". After storage is completed, a storage log is generated to record the storage time, file size and associated identification information.

[0039] Preferably, in step S3: for the alloy cutting tool to be predicted, the reference light source and the distinguishing light source are controlled to alternately irradiate the tool in a time-sharing manner, and the real-time reference image and the real-time distinguishing image of the alloy cutting tool to be predicted are acquired respectively, and the timing of the alternating irradiation is synchronized with the cutting interval cycle of the alloy cutting tool. Optionally, in step S3, controlling the reference light source and the distinguishing light source to alternate illumination at different times specifically involves: Obtain the cutting interval cycle duration of the alloy tool to be predicted; The duration of a single irradiation of both the reference light source and the distinguishing light source is set to 40% of the cutting interval cycle length; The switching interval between the two light sources is set to 20% of the cutting interval cycle length; Set the number of alternating irradiation cycles to ensure that the number of cycles is completely consistent with the number of cutting interval cycles; The two sets of light sources are controlled to work alternately in a time-sharing manner, following the sequence of turning on the reference light source, switching interval, turning on the distinguishing light source, and switching interval.

[0040] In this embodiment, the cutting control unit of the processing equipment is connected through a data interface to obtain the cutting interval cycle length of the alloy tool to be predicted. The cycle timer is used to collect and record the length in real time at a collection frequency of 10 times / second. The average value of the 10 collected data is taken as the final cutting interval cycle length. The measured length is 5 seconds, and the data error is controlled within ±0.1 seconds.

[0041] Based on the measured 5-second cutting interval cycle length, the single irradiation duration of both the reference light source and the distinguishing light source was set to 40% of the cutting interval cycle length, resulting in a single irradiation duration of 2 seconds. The switching interval duration of the two light sources was set to 20% of the cutting interval cycle length, resulting in a switching interval duration of 1 second. All duration parameters were entered and locked through the light source control unit to ensure that the parameters were without deviation.

[0042] The number of cycles for alternating illumination by the two sets of light sources is set. By reading the cutting program parameters of the processing equipment, the total number of cutting interval cycles of the alloy tool to be predicted is obtained as 120. The number of alternating illumination cycles is set to 120 to ensure that the number of cycles is completely consistent with the number of cutting interval cycles. The cycle number parameter is synchronously entered into the light source control unit and linked with the cutting program.

[0043] Control commands are sent by the light source control unit to control the two sets of light sources to work alternately in a time-sharing manner, following the sequence of turning on the reference light source, switching interval, turning on the distinguishing light source, and switching interval. After the reference light source is turned on, it maintains illumination for 2 seconds, and after turning off, it enters a 1-second switching interval. After the switching interval ends, the distinguishing light source is turned on and maintains illumination for 2 seconds, and after turning off, it enters a 1-second switching interval again. This cycle continues until 120 alternating illuminations are completed. Throughout the process, the light source status monitor provides real-time feedback on the on / off status of the light source to ensure precise timing control without any stuttering or delay.

[0044] Optionally, the timing of the alternating irradiation in step S3 is synchronized with the cutting interval cycle of the alloy tool, specifically as follows: The timing of the reference light source's activation is linked to the start of the cutting interval cycle, so that the reference light source is activated simultaneously when the cutting action stops; the timing of the distinguishing light source's activation is linked to the middle of the cutting interval cycle, so that the distinguishing light source is activated simultaneously after the reference light source is deactivated and the switching interval ends. The duration of the cutting interval cycle is monitored in real time. When the cycle fluctuates, the duration of the light source illumination and the switching interval are adjusted synchronously.

[0045] In this embodiment, the cutting control unit and the light source control unit of the processing equipment are connected through a data interface to obtain the start time signal of the cutting interval cycle. The start time signal of the reference light source is bound to the start time signal using a timing binding line, and the binding accuracy is controlled within ±10ms. At the same time as the cutting control unit sends the cutting action stop electrical signal, it simultaneously outputs the reference light source start electrical signal to the light source control unit, realizing the synchronous linkage between the cutting action stop and the reference light source start, and the time difference between the two is controlled within ±5ms.

[0046] It should be noted that, based on the measured 5-second cutting interval cycle, the midpoint of the cutting interval cycle is calculated to be 2.5 seconds after the start time. A timing calibration circuit is used to bind the turn-on signal of the distinguishing light source with this midpoint signal. After the reference light source is turned off for the set 2-second illumination duration, it automatically enters a 1-second switching interval. The end time of the switching interval is precisely aligned with the midpoint of the cutting interval cycle through a timing trigger. At this time, the light source control unit outputs the distinguishing light source turn-on electrical signal to ensure that the distinguishing light source is turned on synchronously with the midpoint, and the timing deviation does not exceed ±5ms.

[0047] A cycle monitor is used to collect the cutting interval cycle duration in real time. The monitoring frequency is set to 20 times / second, and the monitoring accuracy is ±0.05 seconds. The monitoring data is transmitted to the light source control unit in real time through the signal transmission line. The light source control unit has a built-in signal comparison circuit, which compares the currently monitored cycle duration with the initially set 5-second cutting interval cycle duration in real time. The comparison frequency is consistent with the monitoring frequency.

[0048] It should be noted that when fluctuations in the cutting interval cycle duration are detected, such as fluctuations to 5.5 seconds, the parameter adjustment circuit built into the light source control unit automatically triggers an adjustment command to synchronously adjust the single irradiation duration and switching interval duration of the reference light source and the distinguishing light source. After adjustment, the single irradiation duration is 40% of 5.5 seconds, i.e., 2.2 seconds, and the switching interval duration is 20% of 5.5 seconds, i.e., 1.1 seconds. The adjustment process is completed within one cutting interval cycle. After adjustment, the cycle monitor is used for recalibration to ensure that the alternating irradiation sequence is always synchronized with the cutting interval cycle, and the timing deviation after adjustment is still controlled within ±10ms.

[0049] Of particular importance is that step S4 involves calling the individual initial baseline image corresponding to the alloy tool to be predicted, specifically as follows: Read the unique serial number of the alloy cutting tool to be predicted and use it as the search keyword; Activate the search function to find the corresponding storage directory based on the search keywords; Retrieve the individual initial baseline image of the alloy cutting tool from the storage directory and read the model and specification information associated with the image; The retrieved model and specification information is compared with the actual model and specification of the alloy cutting tool to be predicted.

[0050] In this embodiment, a barcode scanner is used to read the unique manufacturing number of the alloy cutting tool to be predicted. The barcode scanner has a scanning accuracy of 0.1mm, a scanning speed of 5 times / second, and a scanning range of 5mm-100mm. The manufacturing number read is a seven-digit code (such as 1234567). After reading, the manufacturing number is used as a search keyword and transmitted to the image retrieval unit. The search keyword transmission rate is 100Mbps to ensure that there is no data loss in the data transmission.

[0051] The image retrieval unit is activated with a retrieval rate of 10 times per second and a retrieval accuracy of 100%. The retrieval unit searches for keywords based on the input serial number and automatically matches the storage directory of the alloy tool baseline image library in the solid-state drive. The naming rules of the storage directory correspond exactly to the serial number. The retrieval progress is fed back in real time during the retrieval process. The retrieval timeout is set to 3 seconds. If the timeout occurs, a new retrieval will be automatically triggered.

[0052] After retrieving the corresponding storage directory (directory name 1234567), the image retrieval unit automatically retrieves the individual initial baseline image of the alloy tool from the directory. The retrieval rate is no less than 20MB / s. The retrieved individual initial baseline image has a resolution of 2592×1944, a pixel depth of 8 bits, and an image format of TIFF. At the same time, the model and specification information associated with the image is read. The model and specification information is "carbide - cutting edge angle 15° - cutting length 80mm", and the reading accuracy is 100%.

[0053] An information comparison circuit is used to compare the retrieved model and specification information with the actual model and specification of the alloy tool to be predicted. The comparison frequency is 5 times / second, and the comparison accuracy is ±0.1mm (for cutting length) and ±0.5° (for cutting edge angle). During the comparison process, the three core parameters of tool material, cutting edge angle and cutting length are checked one by one. After the comparison is completed, the comparison result is generated. If the three parameters are completely consistent, the retrieval is completed. If there is a discrepancy, a re-retrieval command is triggered until a matching individual initial baseline image is retrieved.

[0054] Preferably, step S4: call up the individual initial baseline image corresponding to the alloy tool to be predicted, compare the real-time reference image with the individual initial baseline image, and combine the differences between the real-time distinguishing image and the real-time reference image to complete the wear prediction of the alloy tool to be predicted.

[0055] Optionally, comparing the real-time reference image with the individual's initial baseline image in step S4 includes: Simultaneously retrieve the real-time reference image and the individual initial baseline image of the alloy cutting tool to be predicted, and place them on the same screen; Using the initial baseline image of the individual alloy cutting tool to be predicted as a reference, the display scale of the real-time reference image is adjusted. Move the position of the real-time reference image so that the real-time reference image is completely aligned with the alloy tool profile in the individual initial baseline image and is at the same coordinate position; Following the order of cutting edge, rake face, and flank face, the real-time reference image and the individual initial baseline image are compared region by region and pixel by pixel. Record the visual differences between each region in the real-time reference image and the individual's initial baseline image to form preliminary comparison results.

[0056] In this embodiment, a real-time reference image and an individual initial baseline image of the alloy cutting tool to be predicted are simultaneously retrieved. The two images have the same resolution, pixel depth, and format. The two images are placed side-by-side on the screen through the image display interface, with a display resolution of 1920×1080. The two images are distributed horizontally on the display interface, with the image spacing controlled at 50 pixels to ensure no overlap or distortion in the on-screen display. Using the individual initial baseline image of the alloy cutting tool to be predicted as a reference, the display ratio of the real-time reference image is adjusted using the image scaling knob, with the scaling accuracy controlled at 1%. During the scaling process, the tool outline dimensions of the two images are compared in real time until the display ratio of the real-time reference image is completely consistent with the individual initial baseline image. At this point, the deviation of the displayed dimensions of the cutting edge length and blade width of the alloy cutting tool in the two images is controlled within ±1 pixel.

[0057] Move the position of the real-time reference image using the image displacement adjustment knob, with the adjustment precision set to 1 pixel / step. Use crosshair positioning lines for alignment assistance, aligning the intersection of the crosshair positioning lines with the center point of the cutting edge of the alloy tool in the individual initial baseline image. Then move the real-time reference image to make the center point of the cutting edge completely coincide with the intersection of the crosshair positioning lines, ensuring that the real-time reference image and the outline of the alloy tool in the individual initial baseline image are completely aligned and in the same coordinate position.

[0058] Following the order of cutting edge, front face, and back face, the real-time reference image and the individual initial baseline image are compared region by region and pixel by pixel. The comparison rate is set to 1 million pixels / second, and the comparison accuracy is set to ±1 gray value. The gray values ​​of corresponding positions in the two images are compared pixel by pixel. Pixels with a gray value difference of more than 5 are recorded. The number and distribution coordinates of the difference pixels in each region are counted to form a preliminary comparison result.

[0059] It should be noted that the parameters such as image resolution, scaling accuracy, alignment deviation, and comparison threshold involved in this embodiment can be adjusted accordingly based on the actual tool monitoring accuracy requirements and the performance of the image processing equipment.

[0060] Optionally, in step S4, the difference between the real-time distinguishing image and the real-time reference image is specifically described as follows: Retrieve the real-time distinguished image of the alloy tool to be predicted and place it on the same screen as the real-time reference image of the alloy tool to be predicted. Adjust the display scale and position of the real-time distinguishing image to make it perfectly aligned with the outline of the alloy tool in the real-time reference image of the alloy tool to be predicted. The real-time distinguishing image is compared with the real-time reference image region by region in the order of cutting edge, rake face and flank face. The brightness of each region in the real-time distinguishing image and the real-time reference image is compared pixel by pixel, and the regions in which the brightness values ​​of the real-time distinguishing image and the real-time reference image change are marked. Record the location, range, and brightness difference of the difference regions between the real-time distinguished image and the real-time reference image to form difference analysis data.

[0061] In this embodiment, an image storage unit is connected via an image retrieval interface. The factory serial number of the alloy tool to be predicted is input, triggering an image retrieval command. Priority is given to retrieving a real-time distinguishing image that matches the acquisition sequence of the real-time reference image. This image is acquired under the illumination of a distinguishing light source and maintains the same resolution (2592×1944), pixel depth (8-bit), and TIFF image format as the real-time reference image. The retrieval rate is no less than 20MB / s to ensure that the image data is complete and without loss. After retrieval, the two images are displayed on the same screen through an image display control circuit, using a left-right split-screen layout. The real-time reference image is on the left, and the real-time distinguishing image is on the right. The screen boundary is separated by a solid black line. The display resolution is adjusted to 1920×1080. The screen display accuracy is calibrated using a display calibration tool to avoid image stretching and distortion, ensuring that the display ratio of the two images is consistent with the actual acquisition ratio.

[0062] Using the real-time reference image as a reference, the image adjustment component is activated. The scaling knob is rotated to adjust the display ratio of the real-time distinguishing image. During the adjustment process, the tool contour size of the two images is captured in real time by the image contour comparison sensor. The scaling accuracy is controlled within 1% until the sensor detects that the error between the two contour sizes is less than 1 pixel. Then, the display position of the real-time distinguishing image is moved by the displacement adjustment joystick with an adjustment accuracy of 1 pixel / step. At the same time, the tool contour calibration function is activated. The system automatically generates calibration lines that fit the cutting edge, rake face, and flank face of the alloy tool in the real-time reference image. The displacement joystick is finely adjusted to make the tool contour of the real-time distinguishing image completely fit with the calibration lines. After alignment is completed, the adjustment component is locked to ensure that the alignment state is stable and the alignment deviation is controlled within ±1 pixel.

[0063] The comparison is carried out region by region in the order of cutting edge, front face, and back face. First, the two images are divided into three independent comparison regions simultaneously using a region division tool. Each region is set with an independent comparison channel to avoid interference between regions. In order to distinguish the characteristics of brightness and darkness changes that are easily presented under the illumination of light source, a gray value acquisition probe is used to collect the brightness and darkness values ​​of corresponding positions in the two images pixel by pixel. The acquisition rate is set to 1 million pixels / second, and the gray value acquisition range is 0-255. The acquired data is transmitted to the comparison circuit in real time. The comparison circuit calculates the brightness and darkness difference between the two pixels pixel by pixel. When the difference exceeds the set threshold, the marking command is automatically triggered, and the region is marked with a red pixel, while the boundary contour of the difference region is outlined.

[0064] The pixel coordinates of the top-left and bottom-right corners of all difference regions are collected. The specific location and range of each difference region are calculated using a coordinate calculation circuit. The total number of pixels in each difference region is counted, and the brightness difference of each pixel within each difference region is recorded, accurate to one grayscale unit. The data, including location coordinates, region range, total number of pixels, and brightness difference, are categorized by region and stored in a data cache unit to form complete difference analysis data. The cache unit's read / write speed is no less than 100MB / s to ensure complete data retention. All parameters involved in this embodiment can be adjusted according to the actual tool material, monitoring accuracy, and equipment performance.

[0065] Optionally, step S4, which involves predicting the wear of the alloy cutting tool to be predicted, specifically includes: The preliminary comparison results between the real-time reference image and the individual initial baseline image are integrated with the difference analysis data between the real-time distinguished image and the real-time reference image to form a difference integration dataset; Based on the differential integrated dataset, the suspected wear area of ​​the alloy cutting tool to be predicted is located, and the specific range, area and distribution location of the suspected wear area are measured; By combining the brightness difference features between the real-time distinguishing image and the real-time reference image, the material loss area and the surface adhesion area in the suspected wear area can be distinguished. Data statistics are collected on the material loss areas, and relevant data for all material loss areas are recorded to form a complete wear prediction report.

[0066] In this embodiment, the preliminary comparison results between the real-time reference image and the individual initial baseline image, as well as the difference analysis data between the real-time distinguishing image and the real-time reference image, are synchronously imported into the data processing terminal through a data interface. A time-series synchronization method is used to achieve accurate matching of the two types of data. The matching accuracy is controlled within ±1 pixel, with image pixel coordinates as the matching benchmark. During the integration process, duplicate difference data is eliminated by comparing pixel coordinates, and invalid data with abnormal grayscale differences is eliminated by valid grayscale value verification. The data is classified and integrated according to the regional division standards of the cutting edge, front face, and back face to form a difference integration dataset. The dataset contains core information such as the pixel coordinates of the upper left and lower right corners of the difference region, the grayscale difference of each pixel, and the total number of pixels in the difference region. The storage format is set to XML format, and the data transmission rate is controlled at no less than 100MB / s to ensure that the dataset is complete and without missing data and that the data correspondence is without deviation.

[0067] Based on the differential integrated dataset, pixel coordinate information of all differential regions is extracted. Using a preset conversion relationship between image pixels and actual tool size, a conversion ratio of 1 pixel to 0.01 mm is determined. The pixel coordinates of the differential regions are then converted to the actual tool position coordinates, thereby locating the suspected wear area of ​​the alloy tool to be predicted. By statistically analyzing the total number of effective pixels within the differential regions and combining this with the conversion ratio, the actual area of ​​the suspected wear area is calculated, with measurement accuracy controlled within ±0.001 mm. 2 At the same time, in accordance with the tool area division standard, mark the specific part of the tool corresponding to each suspected wear area, clearly distinguish the cutting edge, rake face or flank face area, and record the actual position coordinate range of each suspected wear area.

[0068] By combining the brightness difference features of the real-time distinguishing image and the real-time reference image, a preset brightness difference threshold is set. The gray values ​​of the suspected wear area in the two images are read pixel by pixel, and the brightness difference between the two is calculated. The two types of areas are distinguished by the difference fluctuation pattern: the material loss area is due to the lack of material in the tool body, and the brightness difference is stable at 10-25 gray units with no obvious fluctuation; the surface attachment area is due to the attachment of foreign objects, and the brightness difference fluctuates between 3-8 gray units, and the difference is unevenly distributed. Based on this, the material loss area and the surface attachment area are accurately distinguished. The boundary of the material loss area is delineated by pixel marking, and the pixel range of all material loss areas is locked to avoid confusion with the surface attachment area.

[0069] Data was statistically analyzed region by region for the locked material loss area, recording the actual area, actual location coordinate range, average value of brightness difference of each pixel in the region, and total number of pixels. The statistical accuracy was controlled within ±1 pixel and ±0.001 mm. 2All statistical data are categorized and organized according to tool location, and imported into the report generation module of the data processing terminal in a preset format. A complete wear prediction report is automatically generated. The report clearly marks the tool's unique factory number, model and specifications, and lists in detail the specific parameters and distribution of wear areas for each material. The report is stored in PDF format in a designated storage unit. All parameters involved can be adjusted according to the actual tool material and monitoring accuracy requirements.

[0070] The present invention also provides a machine vision-based alloy tool wear prediction system for performing the above-described machine vision-based alloy tool wear prediction method. The machine vision-based alloy tool wear prediction system includes: The light source configuration module is used to set two sets of light sources with different illumination angles. The two sets of light sources are defined as a reference light source and a distinguishing light source, so that the illumination directions of the two sets of light sources form a preset angle. The baseline image acquisition and storage module is used to control the reference light source to illuminate the alloy tool individually before each new alloy tool is put into use, acquire a reference image of the alloy tool in its unworn state, and store it as the individual initial baseline image of the alloy tool and associate it with the alloy tool identification. The real-time image acquisition module is used to control the reference light source and the differentiation light source to alternately irradiate the alloy cutting tool to be predicted in use, and to acquire the real-time reference image and the real-time differentiation image of the alloy cutting tool to be predicted, respectively. The timing of the alternating irradiation is synchronized with the cutting interval cycle of the alloy cutting tool. The wear prediction module is used to call up the individual initial baseline image corresponding to the alloy tool to be predicted, compare the real-time reference image with the individual initial baseline image, and combine the differences between the real-time distinguishing image and the real-time reference image to complete the wear prediction of the alloy tool to be predicted.

[0071] Please see Figure 4 The image shows the material wear at the cutting edge of the alloy tool: chipped and blackened damaged areas appear on the edge of the cutting edge, which contrasts sharply with the original metallic luster of the alloy tool, and directly reflects the material loss generated during the cutting process.

[0072] Therefore, the embodiments should be considered as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the application are intended to be included within the invention.

[0073] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A machine vision-based method for predicting wear of alloy cutting tools, characterized in that, Includes the following steps: Step S1: Set up two sets of light sources with different illumination angles, define the two sets of light sources as the reference light source and the distinguishing light source, and make the illumination directions of the two sets of light sources form a preset angle; Step S2: Before each new alloy tool is put into use, the reference light source is controlled to illuminate the alloy tool individually, and a reference image of the alloy tool in its unworn state is acquired as the individual initial baseline image of the alloy tool and associated with the alloy tool identification for storage. Step S3: For the alloy cutting tool to be predicted, control the reference light source and the differentiation light source to alternately irradiate it in a time-sharing manner, and collect the real-time reference image and the real-time differentiation image of the alloy cutting tool to be predicted respectively. The timing of the alternating irradiation is synchronized with the cutting interval cycle of the alloy cutting tool. Step S4: Call up the individual initial baseline image corresponding to the alloy tool to be predicted, compare the real-time reference image with the individual initial baseline image, and combine the differences between the real-time distinguishing image and the real-time reference image to complete the wear prediction of the alloy tool to be predicted.

2. The method for predicting wear of alloy cutting tools based on machine vision according to claim 1, characterized in that, The specific steps for setting up two sets of light sources with different illumination angles in step S1 are as follows: Two sets of diffuse reflection cold light sources with the same power were selected as the reference light source and the distinguishing light source, respectively. Adjust the emitting surface of the reference light source to be parallel to the main cutting plane of the carbide tool, and adjust the emitting surface of the distinguishing light source to be parallel to the back face of the carbide tool. Adjust the installation height of the two sets of light sources so that the vertical distance between the light source and the cutting edge of the alloy tool is controlled between 100mm and 150mm. Adjust the illumination direction of the two sets of light sources so that the preset angle between them is 45°-60°, and the illumination area of ​​both sets of light sources covers the tool wear monitoring area.

3. The machine vision-based alloy tool wear prediction method according to claim 2, characterized in that, Two sets of diffuse cold light sources with the same power were selected, serving as the reference light source and the distinguishing light source, respectively: The diffuse reflection cold light source facing directly above the cutting edge of the alloy tool is defined as the reference light source; Adjust the illumination angle of the reference light source so that the illumination area completely covers all wear monitoring areas of the alloy cutting tool, including the cutting edge, rake face, and flank face. The diffuse reflection cold light source facing the side of the rake face of the alloy tool is defined as the distinguishing light source; Adjust the illumination path of the distinguishing light source so that the illumination path forms a preset angle with the mirror reflection path of the alloy tool cutting edge; Adjust the illumination range of the distinguishing light source so that the overlap rate between the illumination range and the illumination area of ​​the reference light source at the wear monitoring part reaches a preset ratio.

4. The method for predicting wear of alloy cutting tools based on machine vision according to claim 1, characterized in that, Step S2, controlling the reference light source to irradiate the alloy tool individually, includes: Adjust the illuminance of the reference light source to a stable output range, and keep the output brightness of the light source stable without fluctuation; Fix the unused alloy cutting tool on the positioning fixture to keep the alloy cutting tool in a horizontal and static state; Adjust the orientation of the alloy cutting tool so that the cutting edge of the alloy cutting tool is directly facing the center of the light-emitting surface of the reference light source; Turn on the reference light source and irradiate continuously for a duration that meets the requirement of uniform light distribution on the surface of the alloy tool.

5. The method for predicting wear of alloy cutting tools based on machine vision according to claim 1, characterized in that, In step S3, controlling the reference light source and the distinguishing light source to alternate illumination at different times specifically involves: Obtain the cutting interval cycle duration of the alloy tool to be predicted; The duration of a single irradiation of both the reference light source and the distinguishing light source is set to 40% of the cutting interval cycle length; The switching interval between the two light sources is set to 20% of the cutting interval cycle length; Set the number of alternating irradiation cycles to ensure that the number of cycles is completely consistent with the number of cutting interval cycles; The two sets of light sources are controlled to work alternately in a time-sharing manner, following the sequence of turning on the reference light source, switching interval, turning on the distinguishing light source, and switching interval.

6. The machine vision-based alloy tool wear prediction method according to claim 1, characterized in that, The timing of the alternating irradiation in step S3 is synchronized with the cutting interval cycle of the alloy tool, specifically as follows: The timing of the reference light source's activation is linked to the start of the cutting interval cycle, so that the reference light source is activated simultaneously when the cutting action stops; the timing of the distinguishing light source's activation is linked to the middle of the cutting interval cycle, so that the distinguishing light source is activated simultaneously after the reference light source is deactivated and the switching interval ends. The duration of the cutting interval cycle is monitored in real time. When the cycle fluctuates, the duration of the light source illumination and the switching interval are adjusted synchronously.

7. The method for predicting wear of alloy cutting tools based on machine vision according to claim 1, characterized in that, Step S4, comparing the real-time reference image with the individual's initial baseline image, includes: Simultaneously retrieve the real-time reference image and the individual initial baseline image of the alloy cutting tool to be predicted, and place them on the same screen; Using the initial baseline image of the individual alloy cutting tool to be predicted as a reference, the display scale of the real-time reference image is adjusted. Move the position of the real-time reference image so that the real-time reference image is completely aligned with the alloy tool profile in the individual initial baseline image and is at the same coordinate position; Following the order of cutting edge, rake face, and flank face, the real-time reference image and the individual initial baseline image are compared region by region and pixel by pixel. Record the visual differences between each region in the real-time reference image and the individual's initial baseline image to form preliminary comparison results.

8. The machine vision-based alloy tool wear prediction method according to claim 1, characterized in that, Step S4, which combines the differences between the real-time distinguishing image and the real-time reference image, specifically involves: Retrieve the real-time distinguished image of the alloy tool to be predicted and place it on the same screen as the real-time reference image of the alloy tool to be predicted. Adjust the display scale and position of the real-time distinguishing image to make it perfectly aligned with the outline of the alloy tool in the real-time reference image of the alloy tool to be predicted. The real-time distinguishing image is compared with the real-time reference image region by region in the order of cutting edge, rake face and flank face. The brightness of each region in the real-time distinguishing image and the real-time reference image is compared pixel by pixel, and the regions in which the brightness values ​​of the real-time distinguishing image and the real-time reference image change are marked. Record the location, range, and brightness difference of the difference regions between the real-time distinguished image and the real-time reference image to form difference analysis data.

9. The machine vision-based alloy tool wear prediction method according to claim 1, characterized in that, Step S4 involves predicting the wear of the alloy cutting tool to be predicted. The preliminary comparison results between the real-time reference image and the individual initial baseline image are integrated with the difference analysis data between the real-time distinguished image and the real-time reference image to form a difference integration dataset; Based on the differential integrated dataset, the suspected wear area of ​​the alloy cutting tool to be predicted is located, and the specific range, area and distribution location of the suspected wear area are measured; By combining the brightness difference features between the real-time distinguishing image and the real-time reference image, the material loss area and the surface adhesion area in the suspected wear area can be distinguished. Data statistics are collected on the material loss areas, and relevant data for all material loss areas are recorded to form a complete wear prediction report.

10. A machine vision-based alloy tool wear prediction system, characterized in that, For performing the machine vision-based alloy tool wear prediction method as described in claim 1, the machine vision-based alloy tool wear prediction system includes: The light source configuration module is used to set two sets of light sources with different illumination angles. The two sets of light sources are defined as a reference light source and a distinguishing light source, so that the illumination directions of the two sets of light sources form a preset angle. The baseline image acquisition and storage module is used to control the reference light source to illuminate the alloy tool individually before each new alloy tool is put into use, acquire a reference image of the alloy tool in its unworn state, and store it as the individual initial baseline image of the alloy tool and associate it with the alloy tool identification. The real-time image acquisition module is used to control the reference light source and the differentiation light source to alternately irradiate the alloy cutting tool to be predicted in use, and to acquire the real-time reference image and the real-time differentiation image of the alloy cutting tool to be predicted, respectively. The timing of the alternating irradiation is synchronized with the cutting interval cycle of the alloy cutting tool. The wear prediction module is used to call up the individual initial baseline image corresponding to the alloy tool to be predicted, compare the real-time reference image with the individual initial baseline image, and combine the differences between the real-time distinguishing image and the real-time reference image to complete the wear prediction of the alloy tool to be predicted.

Images