Workpiece surface defect intelligent detection system

Abstract

The invention provides a workpiece surface defect intelligent detection system, which comprises an image acquisition unit, a defect identification unit, a rejection agent injection unit, a magnetic bead driving unit, a mobile data acquisition unit and a defect verification unit, and is characterized in that after the image acquisition unit acquires surface image data of a to-be-detected workpiece, the defect identification unit performs identification based on a neural network; a to-be-detected workpiece is arranged on the surface of the to-be-detected workpiece, a suspected defect can be obtained, then the reject injection unit can inject a reject containing soft magnetic particles into the suspected defect and conduct cleaning and flattening treatment, the magnetic bead driving unit can drive the magnetic beads to move on the surface of the to-be-detected workpiece, the magnetic beads pass through the suspected defect, and the soft magnetic particles in the suspected defect can affect movement of the magnetic beads. After the magnetic bead moving path and the magnetic field change rate when the magnetic bead passes through the suspected defect are collected, the defect checking unit can recheck the suspected defect and judge whether the suspected defect is a real defect or not, so that the accuracy of a defect identification and detection result is ensured, and the defect identification and detection efficiency is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of defect detection, in particular to a workpiece surface defect intelligent detection system. BACKGROUND

[0002] In the field of industrial manufacturing, workpiece surface defect detection is a key link to ensure product quality and reliability. Traditional detection methods mainly rely on manual viewing or contact measurement, which has problems such as low efficiency, strong subjectivity, narrow application range, etc., and is difficult to meet the detection needs of micron-level defects in high-end manufacturing fields such as aerospace, semiconductors, and new energy. With the advancement of Industry 4.0 and intelligent manufacturing, intelligent detection technology based on computer vision and deep learning is gradually applied. Through industrial camera image acquisition and convolutional neural network defect recognition, the detection speed is improved to more than 50 pieces per minute. However, current technologies generally rely on a single algorithm or single modal data for one-time defect judgment, lack a secondary verification mechanism for suspected defects, resulting in detection accuracy in complex scenarios that is difficult to meet the needs of high-end manufacturing. SUMMARY

[0003] In view of this, the present application provides a workpiece surface defect intelligent detection system, which can recheck defects to ensure the accuracy of defect detection results.

[0004] The technical scheme of the present application is as follows:

[0005] A workpiece surface defect intelligent detection system, comprising:

[0006] An image acquisition unit for acquiring surface image data of a workpiece to be tested;

[0007] A defect recognition unit for recognizing suspected defects on the surface of the workpiece to be tested according to the surface image data, and obtaining the position, length, depth, shape and direction of the suspected defects;

[0008] A repellent injection unit for injecting a repellent containing soft magnetic particles into the suspected defects;

[0009] A magnetic bead driving unit for driving magnetic beads to move on the surface of the workpiece to be tested and pass through the suspected defects;

[0010] A movement data acquisition unit for acquiring the magnetic bead movement path of the magnetic beads on the surface of the workpiece to be tested and the magnetic field change rate when the magnetic beads pass through the suspected defects;

[0011] A defect verification unit for verifying the suspected defects based on the magnetic bead movement path and the magnetic field change rate;

[0012] The defect identification unit is connected with the image acquisition unit, the repulsive agent injection unit and the magnetic bead driving unit respectively, and the mobile data acquisition unit is connected with the defect identification unit.

[0013] Preferably, the image processing unit is further included for pre-processing the surface image data, the pre-processing including denoising, distortion correction and contrast enhancement, and the image processing unit is connected with the image acquisition unit and the defect identification unit respectively.

[0014] Preferably, the execution steps of the image acquisition unit include:

[0015] Step S11, initializing the industrial camera, and conveying the workpiece to be tested to the center of the field of view of the industrial camera through the conveying belt;

[0016] Step S12, selecting the light source type and adjusting the light angle, intensity and spectrum according to the material and surface characteristics of the workpiece;

[0017] Step S13, controlling the industrial camera to collect the surface image data of the workpiece to be tested at a preset frequency.

[0018] Preferably, the execution steps of the defect identification unit include:

[0019] Step S21, collecting the historical defect data of the workpiece, and dividing the historical defect data into a training set and a test set;

[0020] Step S22, constructing a defect identification model based on a neural network, training the defect identification model through the training set, and testing the accuracy through the test set;

[0021] Step S23, inputting the surface image data into the trained defect identification model, identifying the suspected defects by the defect identification model, and extracting the position, length, depth, shape and trend of the suspected defects.

[0022] Preferably, the execution steps of the repulsive agent injection unit include:

[0023] Step S31, using a high-pressure air gun to blow and clean the position of the suspected defects, and calculate the volume of the suspected defects based on the length, depth and shape of the suspected defects;

[0024] Step S32, mixing the soft magnetic particles and the gel carrier uniformly to form a repulsive agent, starting a micro pump to inject the repulsive agent to the position of the suspected defects according to the volume of the suspected defects, and using an ultraviolet lamp to irradiate and solidify during the injection process;

[0025] Step S33, after the injection is completed, the suspected defect position is scanned by a laser micrometer, the height difference between the suspected defect and the surface of the workpiece to be measured is calculated, the raised area is scraped, and the recessed area is secondarily injected.

[0026] Preferably, the execution steps of the repulsive agent injection unit further include:

[0027] Step S34, the suspected defect is washed around by a gas-liquid spray head, and the area around the suspected defect is cleaned by a fiber brush head.

[0028] Preferably, the execution steps of the magnetic bead driving unit include:

[0029] Step S41, an initial moving path is set according to the trend of the suspected defect, and the initial moving path is perpendicular to the trend of the suspected defect;

[0030] Step S42, an electromagnetic coil array is arranged above the workpiece to be measured, and the magnetic beads are placed at the head of the initial moving path;

[0031] Step S43, the electromagnetic coil array is driven to generate a magnetic field gradient, the magnetic beads are driven to move along the initial moving path, and pass through the suspected defect.

[0032] Preferably, the execution steps of the magnetic bead driving unit further include:

[0033] Step S44, the initial moving path is translated, after the magnetic beads are placed at the head of the initial moving path, the magnetic beads are driven to move along the new initial moving path, and pass through the suspected defect.

[0034] Preferably, the execution steps of the moving data acquisition unit include:

[0035] Step S51, the industrial camera is initialized, and the image acquisition area of the industrial camera is limited to the area near the initial moving path;

[0036] Step S52, video data of the magnetic beads moving along the initial moving path is acquired, the video data is divided into frame image data, and the magnetic beads in the frame image data are connected in time sequence to form a magnetic bead moving path;

[0037] Step S53, the rate of change of the magnetic field when the magnetic beads move through the suspected defect is acquired by the Hall sensor array.

[0038] Preferably, the execution steps of the defect determination unit include:

[0039] Step S61, the magnetic bead moving path is compared with the initial moving path, and a trajectory deviation degree is calculated.

[0040] Step S62, when the trajectory deviation and the magnetic field change rate are both greater than the preset standard threshold, the suspected defect is determined as a real defect.

[0041] Compared with the prior art, the present application has the following advantages:

[0042] ①The image acquisition unit can acquire surface image data of the workpiece to be measured, and the defect recognition unit is provided with a neural network, which can recognize and detect suspected defects based on the surface image data. Compared with manual or contact measurement, the efficiency is higher, and the detection accuracy is more accurate.

[0043] ②After obtaining the suspected defect, the magnetic beads are driven to move on the surface of the workpiece to be measured after injecting the repellant containing soft magnetic particles into the suspected defect. When the magnetic beads move through the repellant, they will deviate due to the action of the soft magnetic particles. According to the movement path of the magnetic beads and the magnetic field change rate, it can be further determined whether the suspected defect really exists, and the coincidence of the suspected defect is realized, and the accuracy of defect detection is improved. BRIEF DESCRIPTION OF DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only preferred embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0045] Figure 1 It is a schematic diagram of a workpiece surface defect intelligent detection system of the present application;

[0046] Figure 2 It is an execution step diagram of the image acquisition unit of the workpiece surface defect intelligent detection system of the present application;

[0047] Figure 3 It is an execution step diagram of the defect recognition unit of the workpiece surface defect intelligent detection system of the present application;

[0048] Figure 4 It is an execution step diagram of the repellant injection unit of the workpiece surface defect intelligent detection system of the present application;

[0049] Figure 5 It is an execution step diagram of the magnetic bead driving unit of the workpiece surface defect intelligent detection system of the present application;

[0050] Figure 6 It is an execution step diagram of the movement data acquisition unit of the workpiece surface defect intelligent detection system of the present application;

[0051] Figure 7An execution step diagram of a defect verification unit of a workpiece surface defect intelligent detection system according to the present application;

[0052] In the figure, 1, image acquisition unit; 2, defect identification unit; 3, repulsive agent injection unit; 4, magnetic bead driving unit; 5, mobile data acquisition unit; 6, defect verification unit; 7, image processing unit. DETAILED DESCRIPTION

[0053] In order to better understand the technical content of the present application, a specific embodiment is provided below, and the present application is further described in combination with the accompanying drawings.

[0054] Referring to Figures 1 to 7 The present application provides a workpiece surface defect intelligent detection system, which comprises:

[0055] The image acquisition unit 1 is used for acquiring surface image data of a workpiece to be measured.

[0056] The defect identification unit 2 is used for identifying suspected defects on the surface of the workpiece to be measured according to the surface image data, and obtaining the position, length, depth, shape and direction of the suspected defects.

[0057] The repulsive agent injection unit 3 is used for injecting a repulsive agent containing soft magnetic particles into the suspected defects.

[0058] The magnetic bead driving unit 4 is used for driving the magnetic beads to move on the surface of the workpiece to be measured and pass through the suspected defects.

[0059] The mobile data acquisition unit 5 is used for acquiring the magnetic bead moving path of the magnetic beads on the surface of the workpiece to be measured and the magnetic field change rate when the magnetic beads pass through the suspected defects.

[0060] The defect verification unit 6 is used for verifying the suspected defects based on the magnetic bead moving path and the magnetic field change rate.

[0061] The defect identification unit 2 is respectively connected with the image acquisition unit 1, the repulsive agent injection unit 3 and the magnetic bead driving unit 4, and the mobile data acquisition unit 5 is connected with the defect verification unit 6.

[0062] The workpiece surface defect intelligent detection system of the application includes two processes of preliminary detection and rechecking. In the preliminary detection process, the image acquisition unit 1 directly acquires the surface image data of the workpiece to be detected, the defect recognition unit 2 recognizes the surface image data after acquiring all possible defects on the surface of the workpiece to be detected, the defect recognition unit 2 is internally provided with a neural network, can intelligently and automatically recognize the surface image data, and obtains suspected defects, and can also obtain basic data such as the position, length, depth, shape and direction of the suspected defects. The defects obtained at this time are only suspected defects obtained by preliminary detection, and the suspected defects need to be rechecked. Since deep learning is used for defect recognition and detection, compared with artificial judgment and contact measurement, the efficiency is higher and the accuracy is higher.

[0063] In the rechecking process, the suspected defects need to be processed. The repulsion agent injection unit 3 needs to inject a repulsion agent containing soft magnetic particles into the suspected defects. The repulsion agent can be fixed in the suspected defects, and then the moving data acquisition unit 5 can drive the magnetic beads to move on the surface of the workpiece to be detected. In the moving process, the magnetic beads pass above the suspected defects, and the suspected defects are filled with the repulsion agent. The soft magnetic particles in the repulsion agent will affect the movement of the magnetic beads, causing the magnetic beads to deviate. The moving data acquisition unit 5 can acquire the path of the magnetic beads when moving on the workpiece to be detected, and form the magnetic bead moving path. Meanwhile, the moving data acquisition unit 5 can also acquire the magnetic field change rate when the magnetic beads move through the suspected defects. Finally, the defect rechecking unit 6 can recheck whether the suspected defects really exist according to the magnetic bead moving path and the magnetic field change rate, so as to ensure the accuracy of the defect recognition and detection result, and facilitate the repair processing of the surface of the workpiece to be detected.

[0064] Preferably, the image processing unit 7 is further included, which is used for pre-processing the surface image data, and the pre-processing includes denoising, distortion correction and contrast enhancement. The image processing unit 7 is respectively connected with the image acquisition unit 1 and the defect recognition unit 2.

[0065] The image processing unit 7 is mainly used for processing the surface image data acquired by the image acquisition unit 1, including median filtering / Gaussian filtering denoising, distortion correction based on perspective transformation of a calibration plate, contrast enhancement and the like, so as to ensure that there is no much irrelevant information in the surface image data, thereby improving the defect recognition and detection accuracy of the defect recognition unit 2.

[0066] Preferably, the execution steps of the image acquisition unit 1 include:

[0067] Step S11, initializing an industrial camera, and conveying the workpiece to be detected to the center of the field of view of the industrial camera through a conveying belt;

[0068] Step S12, according to the workpiece material and surface characteristics, select the light source type and adjust the light angle, intensity and spectrum;

[0069] Step S13, control the industrial camera to collect the surface image data of the workpiece to be tested at a preset frequency.

[0070] Place the workpiece to be tested on the conveyor belt, which can transport the workpiece to be tested. When the workpiece to be tested is transported under the industrial camera, stop moving to ensure that the workpiece to be tested is located at the center of the field of view of the industrial camera, and ensure that the posture of the workpiece to be tested is consistent with the positioning reference. Then, according to the different workpiece materials and surface characteristics, different types of light sources are selected, such as ring light / backlight / structured light, and the light angle, intensity and spectrum are adjusted to ensure that the industrial camera can collect clear surface image data of the workpiece to be tested.

[0071] Preferably, the execution steps of the defect recognition unit 2 include:

[0072] Step S21, collect historical defect data of the workpiece, and divide the historical defect data into a training set and a test set;

[0073] Step S22, construct a defect recognition model based on a neural network, train the defect recognition model through the training set, and test the accuracy through the test set;

[0074] Step S23, input the surface image data into the trained defect recognition model, identify the suspected defects by the defect recognition model, and extract the position, length, depth, shape and trend of the suspected defects.

[0075] The defect recognition unit 2 recognizes defects based on a neural network. After collecting historical defect data of the workpiece, the historical defect data is divided into a training set and a test set according to a ratio of 7:3. The training set is used to train the defect recognition model. After training for a certain period of time, the test set is used for testing. When the test accuracy reaches a set threshold, the training can be stopped. At this time, the defect recognition model can process the surface image data, so that the suspected defects on the surface of the workpiece to be tested can be quickly recognized, and the basic data of the suspected defects can be extracted at the same time.

[0076] Preferably, the execution steps of the repellant injection unit 3 include:

[0077] Step S31, use a high-pressure air gun to blow and clean the position of the suspected defect, and remove surface dust and oil stains, and calculate the volume based on the length, depth and shape of the suspected defect;

[0078] Step S32, after the soft magnetic particles are mixed with the gel carrier to form the repellent, the micro pump is started to inject the repellent into the suspected defect according to the volume of the suspected defect, and ultraviolet light is used for irradiation during the injection process to cure;

[0079] Step S33, after the injection is completed, the suspected defect position is scanned by a laser micrometer, the height difference between the suspected defect and the surface of the workpiece to be measured is calculated, the raised area is scraped, and the recessed area is secondarily injected;

[0080] Step S34, the area around the suspected defect is washed by a gas-liquid spray head, and the area around the suspected defect is cleaned by a fiber brush head.

[0081] After the position of the suspected defect is obtained, the repellent needs to be filled into the suspected defect. In order to reduce external interference as much as possible, the area near the suspected defect position on the surface of the workpiece to be measured needs to be cleaned. A high-pressure air gun is used to blow filtered and dried air around the suspected defect for 1-2 seconds to remove surface dust, oil stains and other impurities, so as to realize the cleaning of the surface. After the soft magnetic particles are mixed with the gel carrier, the repellent can be obtained. Then, according to the length, depth and shape of the suspected defect, the volume can be calculated. The micro pump can inject a corresponding amount of repellent according to the volume of the suspected defect, so that the repellent fills the suspected defect. In the process of injection, ultraviolet light is irradiated by ultraviolet light, which can trigger the crosslinking reaction of the photoinitiator in the gel carrier, so that the soft magnetic particles are fixed in the gel carrier, and the gel carrier is firmly connected with the suspected defect. At this time, the preliminary injection of the repellent can be completed.

[0082] After the repellent is cured, due to deformation, a raised or recessed area may be formed. At this time, the suspected defect area needs to be leveled. A laser micrometer can be used to scan the suspected defect area to obtain the height difference between the suspected defect area and the surface of the workpiece to be measured, so as to obtain the raised area and the recessed area. The raised area needs to be scraped, and the recessed area needs to be secondarily injected. Repeated scanning, detection and leveling can ensure the leveling of the suspected defect area and the surface of the workpiece to be measured, so as to avoid unexpected deviation of the magnetic beads when moving.

[0083] Preferably, the execution steps of the magnetic bead driving unit 4 include:

[0084] Step S41, setting an initial moving path according to the trend of the suspected defect, the initial moving path being perpendicular to the trend of the suspected defect;

[0085] Step S42, setting an electromagnetic coil array above the workpiece to be measured, and placing the magnetic beads at the head of the initial moving path;

[0086] Step S43, driving the electromagnetic coil array to generate a magnetic field gradient, driving the magnetic beads to move along the initial moving path and passing through the suspected defect;

[0087] Step S44, translating the initial moving path, after placing the magnetic beads at the head of the initial moving path, driving the magnetic beads to move along the new initial moving path and passing through the suspected defect.

[0088] According to the trend of the suspected defect, the moving direction of the magnetic beads can be determined, wherein the initial moving path is perpendicular to the trend of the suspected defect, so as to ensure that the magnetic beads can pass through the suspected defect when moving along the initial moving path. The driving of the magnetic beads is realized by the magnetic field gradient. After the electromagnetic coil array is arranged above the workpiece to be tested, the magnetic field gradient can be obtained by controlling the current size and direction of the electromagnetic coil. Under the action of the magnetic field gradient, the magnetic beads can move from the head of the initial moving path. When the magnetic beads move through the suspected defect, the movement of the magnetic beads will change due to the magnetic field of the soft magnetic particles. After the magnetic beads move to the edge of the workpiece to be tested, the driving movement of the magnetic beads can be performed multiple times. Each time, the position of the initial moving path can be changed. After the initial moving path is translated, the magnetic beads are driven to move repeatedly, so that the data of the deviation of the magnetic beads moving along different paths to the suspected defect multiple times can be obtained.

[0089] Preferably, the execution steps of the moving data acquisition unit 5 include:

[0090] Step S51, initializing the industrial camera, limiting the image acquisition area of the industrial camera in the area near the initial moving path;

[0091] Step S52, acquiring video data of the magnetic beads moving along the initial moving path, dividing the video data into frame image data, and connecting the magnetic beads in the frame image data in time sequence into a magnetic bead moving path;

[0092] Step S53, acquiring the magnetic field change rate of the magnetic beads moving through the suspected defect by the Hall sensor array.

[0093] When the magnetic beads move on the surface of the workpiece to be tested, the image data of the movement of the magnetic beads is acquired by the industrial camera. The image acquisition area of the industrial camera can be adjusted so that the image acquisition area is limited in the area near the initial moving path. Then the industrial camera can acquire the video data of the magnetic beads moving along the initial moving path. The video data is divided into frame image data. The magnetic beads in each frame image data can be connected in time sequence. The connection line after connection is used as the magnetic bead moving path. When the magnetic beads pass through the suspected defect, the magnetic field near the suspected defect will change due to the existence of the soft magnetic particles. The magnetic field change rate of the magnetic beads passing through the suspected defect can be acquired by the Hall sensor array for the defect core determination unit 6 to determine.

[0094] Preferably, the execution step of the defect verification unit 6 comprises:

[0095] Step S61, compare the magnetic bead moving path with the initial moving path, and calculate the trajectory deviation degree;

[0096] Step S62, when the trajectory deviation degree and the magnetic field change rate are both greater than the preset standard threshold, verify the suspected defect as a real defect.

[0097] The defect verification unit 6 can verify the defect based on the magnetic bead moving path and the magnetic field change rate, wherein the magnetic bead moving path is compared with the initial path, and the trajectory deviation degree is calculated, when the trajectory deviation degree and the magnetic field change rate are both greater than the preset standard threshold, it can be judged as a real defect, and the initial moving path has multiple, the deviation degree of each initial moving path and the corresponding magnetic field change rate can be averaged, and then compared with the preset standard threshold, to improve the accuracy of the judgment, by verifying the defect, the accuracy of the defect recognition detection can be improved, and the defect recognition efficiency can be improved.

[0098] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. An intelligent detection system for surface defects in a workpiece, characterized in that, include: The image acquisition unit is used to acquire surface image data of the workpiece under test. The defect identification unit is used to identify suspected defects on the surface of the workpiece under test based on surface image data, and to obtain the location, length, depth, shape and orientation of the suspected defects. The repellent injection unit is used to inject a repellent containing soft magnetic particles into the suspected defect; A magnetic bead driving unit is used to drive a magnetic bead to move on the surface of the workpiece under test and pass over suspected defects. The mobile data acquisition unit is used to acquire the magnetic bead's movement path on the surface of the workpiece under test and the rate of change of the magnetic field when the magnetic bead passes through a suspected defect. The defect verification unit is used to verify suspected defects based on the movement path of the magnetic bead and the rate of change of the magnetic field; The defect identification unit is connected to the image acquisition unit, the repellent injection unit, and the magnetic bead driving unit, respectively, and the mobile data acquisition unit is connected to the defect verification unit.

2. The intelligent detection system for workpiece surface defects according to claim 1, characterized in that, It also includes an image processing unit, which is used to preprocess the surface image data. The preprocessing includes noise reduction, distortion correction and contrast enhancement. The image processing unit is connected to the image acquisition unit and the defect recognition unit respectively.

3. The intelligent detection system for workpiece surface defects according to claim 1, characterized in that, The execution steps of the image acquisition unit include: Step S11: Initialize the industrial camera and transport the workpiece to be tested to the center of the industrial camera's field of view via a conveyor belt. Step S12: Select the light source type and adjust the illumination angle, intensity, and spectrum according to the workpiece material and surface characteristics; Step S13: Control the industrial camera to collect surface image data of the workpiece under test at a preset frequency.

4. The intelligent detection system for workpiece surface defects according to claim 1, characterized in that, The execution steps of the defect identification unit include: Step S21: Collect historical defect data of the workpiece and divide the historical defect data into training set and test set; Step S22: Construct a defect recognition model based on a neural network, train the defect recognition model using a training set, and test its accuracy using a test set; Step S23: Input the surface image data into the trained defect recognition model. The defect recognition model identifies suspected defects and extracts the location, length, depth, shape, and orientation of the suspected defects.

5. The intelligent detection system for workpiece surface defects according to claim 1, characterized in that, The execution steps of the repellent injection unit include: Step S31: Use a high-pressure air gun to blow away the suspected defect location and remove surface dust and oil. Calculate the volume based on the length, depth and shape of the suspected defect. Step S32: After the soft magnetic particles and gel carrier are mixed evenly to form a repellent, start the micro pump to inject the repellent into the suspected defect location according to the volume of the suspected defect, and use ultraviolet light to cure during the injection process. Step S33: After injection, scan the suspected defect location with a laser micrometer, calculate the height difference between the suspected defect and the surface of the workpiece to be tested, scrape off the raised area, and perform secondary injection on the recessed area.

6. The intelligent detection system for workpiece surface defects according to claim 5, characterized in that, The execution steps of the repellent injection unit further include: Step S34: Rinse the area around the suspected defect using a gas-liquid nozzle and clean the area around the suspected defect using a fiber brush head.

7. The intelligent detection system for workpiece surface defects according to claim 1, characterized in that, The execution steps of the magnetic bead driving unit include: Step S41: Set an initial movement path based on the direction of the suspected defect, wherein the initial movement path is perpendicular to the direction of the suspected defect; Step S42: Set up an electromagnetic coil array above the workpiece to be tested, and place the magnetic bead at the beginning of the initial moving path; Step S43: Drive the electromagnetic coil array to generate a magnetic field gradient, causing the magnetic bead to move along the initial moving path and pass through the suspected defect.

8. The intelligent detection system for workpiece surface defects according to claim 7, characterized in that, The execution steps of the magnetic bead driving unit further include: Step S44: Translate the initial moving path. After placing the magnetic bead at the beginning of the initial moving path, drive the magnetic bead to move along the new initial moving path and pass through the suspected defect.

9. The intelligent detection system for workpiece surface defects according to claim 7, characterized in that, The execution steps of the mobile data acquisition unit include: Step S51: Initialize the industrial camera and limit the image acquisition area of ​​the industrial camera to the area near the initial movement path; Step S52: Collect video data of the magnetic bead moving along the initial moving path, divide the video data into frame image data, and connect the magnetic beads in the frame image data in time order to form the magnetic bead moving path. Step S53: Collect the rate of change of magnetic field when the magnetic bead moves past the suspected defect using a Hall sensor array.

10. The intelligent detection system for workpiece surface defects according to claim 7, characterized in that, The execution steps of the defect verification unit include: Step S61: Compare the moving path of the magnetic bead with the initial moving path and calculate the trajectory deviation. Step S62: When both the trajectory deviation and the magnetic field change rate are greater than the preset standard threshold, the suspected defect is verified as a real defect.

Images