Visual detection method for bolt fluorescent magnetic powder inspection and detection system thereof

Abstract

The invention relates to a visual detection method for bolt fluorescent magnetic powder inspection and a detection system thereof. The visual detection method is characterized by comprising the following specific steps: S1, feeding bolts; s2, spraying and magnetizing the bolt; s3, carrying out CCD automatic identification on the bolt; s4, sorting and demagnetizing the bolts; a CCD automatic identification system is used for detecting the bolt sprayed with the magnetic suspension, and a circumferential identification camera, a bolt end identification camera and a bolt neck identification camera are used for carrying out all-around detection on the bolt, so that the detection is more comprehensive; in addition, according to the method, magnetization parameters are accurately set according to different magnetization modes of the bolt, and the magnetization stability of the bolt is guaranteed; the convolutional neural network is adopted in the CCD automatic identification system for training and detection, so that the detection accuracy is higher; by processing the light source and arranging the filter lens on the lens, the definition of the collected picture is higher, and the recognition precision of the bolt defect is improved.

Description

Technical Field

[0001] The invention relates to the field of bolt flaw detection, and in particular to a visual detection method and a detection system for bolt fluorescent magnetic particle flaw detection. Background Art

[0002] Bolts are mechanical parts, cylindrical threaded fasteners equipped with nuts. They are divided into standard parts and non-standard parts. Non-standard parts include customized, special parts and parts required for special equipment. The quality of bolts needs to be tested by non-destructive testing during the production process.

[0003] After the production of bolts is completed, they need to be inspected internally, and magnetic particle inspection is generally used for flaw detection. After the ferromagnetic materials and parts are magnetized, the magnetic lines of force on the surface and near the surface of the materials and parts are locally distorted due to the existence of discontinuities. This distortion will cause some magnetic lines of force to escape from the surface of the materials and parts to form a leakage magnetic field. The leakage magnetic field absorbs the magnetic powder or magnetic suspension applied to the surface of the materials and parts, forming magnetic marks that are visually visible under appropriate lighting, showing the location, shape and size of the discontinuity, and then the magnetic mark display is observed, interpreted and evaluated according to relevant standards; The existing magnetic particle inspection equipment is not accurate and perfect enough, and it is difficult to detect tiny cracks on the neck of the bolt surface, which may lead to the problem of unqualified workpieces flowing out. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a visual inspection method for bolt fluorescent magnetic particle flaw detection, which can solve the problems of inaccurate magnetizing current, unstable control and poor precision in general bolt fluorescent magnetic particle detection.

[0005] In order to solve the above technical problems, the technical solution of the present invention is: a visual inspection method for bolt fluorescent magnetic particle flaw detection, the innovation of which is: the specific flaw detection method is as follows: S1: Bolt loading: The bolts placed at the loading station are clamped by a combined cylinder and placed on the bolt fluorescent magnetic particle flaw detection line in sequence. The bolt fluorescent magnetic particle flaw detection line feeds two workpieces each time, and can perform double-station flaw detection at the same time; S2: Spray magnetization of bolts: The bolts on the bolt fluorescent magnetic particle inspection line are transported to the spray magnetization station, the bolts are clamped, lifted and rotated, and magnetic suspension is sprayed on the rotating bolts at the same time; the bolt magnetization method can be selected as circumferential magnetization, longitudinal magnetization and composite magnetization; the direct current method is used to detect longitudinal cracks in the thread; the coil induction method is used to detect circumferential defects of the bolts; S3: CCD automatic identification of bolts: The bolts that have been sprayed and magnetized are transported to the visual inspection box; a black light is used in the visual inspection box to cooperate with a bolt circumferential identification camera, a bolt end identification camera, and a bolt neck identification camera to inspect the bolts; the CCD automatic identification system is used to collect and compare images of the bolts to identify defective bolts; S4: Bolt sorting and demagnetization: After the bolts are automatically detected and identified by CCD, the qualified bolts are grabbed to another production line through the PLC control sorting module, and the bolts are demagnetized and unloaded from the qualified product station; the unqualified and defective bolts are output from the defective product station along the original production line.

[0006] Furthermore, the S1: the flaw detection cycle of the double-station flaw detection in the loading of the bolts is 13-14 seconds / 2 pieces; the S2: the circumferential magnetization current in the spray magnetization of the bolts: AC is 0-3000A effective value, touch screen display, continuously adjustable, with power-off phase control; longitudinal magnetization magnetic potential: AC is 0-18000AT effective value, touch screen display, continuously adjustable, with power-off phase control, residual magnetic intensity ≤0.2mT; the S4: in the sorting of the bolts, the demagnetization magnetic potential is AC15000-0AT effective value, continuously adjustable.

[0007] Furthermore, in the S3: CCD automatic identification of bolts, the illumination of the black light reaches more than 4500μw / cm2 at 380mm from the center of the lamp, and a filter is set on the black light, and a polarizing box UV filter is added to each camera to ensure that only ultraviolet rays of a specific wavelength are emitted, and only green fluorescence enters the camera lens; the flaw detection sensitivity is clearly displayed using a type A 2# test piece.

[0008] Further, in the spray magnetization of the bolt, the circumferential magnetization current is I=8D~15D; wherein I- is the effective value of the current, in ampere (A); D- is the diameter or maximum width of the specimen, in millimeters (mm); the longitudinal magnetization current I=4π×d×H; wherein I- is the effective value of the current, in ampere (A); d- is the distance between the cable and the surface to be tested, in millimeters (mm); H- is the tangential field strength, in kiloamperes per meter (A / m); fluorescent magnetic powder in the magnetic suspension: 2-3 g / L; dispersant: 0.2% of the weight of the magnetic suspension; rust inhibitor: 0.5% of the weight of the magnetic suspension; defoaming agent: 0.5-1 g / L.

[0009] Further, in the S3: CCD automatic identification of bolts, the CCD automatic identification system uses six cameras for simultaneous detection, wherein three cameras detect the surface of the bolt from above, two cameras detect both ends of the bolt from the side, and one camera is tilted and is dedicated to detecting the neck position of the hexagonal bolt; The CCD automatic recognition system has the function of automatic image acquisition: the camera and lens use industrial-grade 2D cameras and 10 million-level high-definition lenses, global shutter, dynamic shooting, frame rate 36fps; the number of pixels is 10 million, and the resolution is not less than 4096 × 2460; the shooting range of a single camera is greater than 300mm*200mm, and the detection accuracy is greater than 20pix / mm; The CCD automatic recognition system has the function of automatic analysis and judgment, and can realize sample library production, learning and training, and deployment. The convolutional neural network is used to train and learn a large number of images collected in the image sample library. After the training, we will get the corresponding neural network model weight file. This file will be added to the forward reasoning prediction software, and it can be used for automatic crack detection. The CCD automatic recognition system has the function of automatic storage and recording; it has the function of saving NG images and automatically clearing images regularly; The main window of the detection software of the CCD automatic recognition system should display the detection screen in real time. The number of windows can be customized and the windows are arranged in a matrix. The software processing window contains the detection time of each beat, the detection target category, and the score / confidence information. The software interface should display the version number, current user, server status, camera status, detection results, and detection operation process information. The detection software of the CCD automatic identification system has three detection modes: Offline detection mode, used to select local image samples for offline detection training and optimize learning algorithms; Continuous detection mode is used for the software to automatically perform online real-time detection. In the continuous detection mode, when the software is in the state of executing detection, when the workpiece being detected moves under the camera's field of view, the software can automatically track and lock the crack; Manual detection mode is used for the software to automatically complete a detection process; the software will be in the following states: Waiting for detection: When entering the manual detection mode, the software is in the waiting for detection state; Capture image: When a workpiece enters the camera's field of view, click the button and the software will capture the image; Execute inspection: After the image is captured, click the button to execute the inspection and analyze whether there are defects; Send results: After the analysis is completed, click the button to send the test results and display them in the status bar; The detection software of the CCD automatic recognition system has a camera debugging module, which is used for parameter configuration and debugging of the camera, including initialization, image acquisition, and parameter setting; The detection software of the CCD automatic identification system has a communication setting module for configuring the parameters for communication with the controller; The detection software of the CCD automatic identification system must have a user login module to facilitate user management and permission setting; The detection software of the CCD automatic recognition system identifies the defects, and the equipment automatically sorts the defective workpieces to the defective product station.

[0010] A visual inspection system for bolt fluorescent magnetic particle flaw detection, the innovation of which is that it includes a conveyor frame, a feeding module, a spray magnetization module, a CCD visual recognition module and a bolt sorting module; The conveyor frame is provided with a loading station, a spray magnetization station, a CCD visual recognition station and a bolt sorting station in sequence along the extension direction; the conveyor frame is provided with a conveyor chain, and a plurality of bolt placement seats are provided on the conveyor chain at equal intervals; The loading module is arranged at the loading station of the conveying frame, and the loading module is a clamping claw structure driven by a cylinder, which is used to grab the bolts and place them on the bolt placing seat on the conveying frame; The spray magnetization module is arranged on the spray magnetization station of the conveyor, and the spray magnetization module includes a support frame, a lifting clamping frame, a rotating cylinder, a magnetic suspension collecting tank and a circulating spray pipeline; the support frame has a pair of and are respectively arranged on the two side surfaces of the extending direction of the conveyor frame and are arranged oppositely; the lifting clamping frame has a pair of and are respectively arranged on the support frame and are arranged oppositely on both sides of the conveyor frame; the rotating cylinder is horizontally installed on the lifting clamping frame and is arranged oppositely on the two lifting clamping frames, and the lifting clamping frame drives the rotating cylinder to move in the horizontal direction and the vertical direction, so as to move the magnetic suspension liquid through the two lifting clamping frames on the conveyor frame. The bolts between the side rotating cylinders are clamped and driven to move, and the rotating cylinders are mutually pressed to clamp the bolts and drive the bolts to rotate; a cover shell is arranged across the conveying frame between the lifting clamping frames; the magnetic suspension liquid collecting tank is arranged below the conveying frame, and the magnetic suspension liquid is arranged in the magnetic suspension liquid collecting tank; the circulating spray pipe is installed on the inner wall of the cover shell, and the circulating spray pipe is provided with a plurality of spray ports aligned with the bolts on the conveying frame; the circulating spray pipe is connected to a circulating pump from above, and the magnetic suspension liquid in the magnetic suspension liquid collecting tank is pumped into the circulating spray pipe by the circulating pump for spraying the screw; The CCD visual recognition module is arranged at the CCD visual recognition station on the conveyor frame, and a visual inspection box is arranged at the CCD recognition station, and the conveyor frame is arranged through the visual inspection box; an inspection table is arranged in the visual inspection box; the CCD visual recognition module includes a camera group and a black light; the camera group includes a top recognition camera, an end recognition camera and a bolt neck recognition camera, which respectively recognize the circumference, end and neck of the bolt; the black light is arranged on the inner wall near the top on both sides of the visual inspection box to illuminate the bolts in the visual inspection box; a bolt clamping and transfer clamp is arranged in the visual inspection box, and the bolts at the input end position of the visual inspection box are grabbed and placed on the inspection table by the bolt clamping and transfer clamp, and the bolts that have completed the inspection are grabbed and placed at the output end position of the visual inspection box; The bolt sorting module is arranged at the bolt sorting station of the conveying frame, and the bolt sorting module is a clamping claw structure driven by a cylinder, which is used to grab the qualified bolts from the bolt placement seat.

[0011] The advantages of the present invention are: 1) In the present invention, the bolts sprayed with magnetic suspension are inspected by a CCD automatic recognition system, and a circumferential recognition camera, a bolt end recognition camera and a bolt neck recognition camera are used to perform all-round detection of the bolts, so that the detection is more comprehensive; in addition, the method accurately sets the magnetization parameters according to different magnetization modes of the bolts, thereby ensuring the stability of the magnetization of the bolts; a convolutional neural network is used for training and detection in the CCD automatic recognition system, so that the detection accuracy is higher; by processing the light source and setting a filter on the lens, the clarity of the collected pictures is ensured to be higher, thereby improving the recognition accuracy of bolt defects. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0013] Figure 1 The present invention is a flow chart of a visual inspection method for fluorescent magnetic particle flaw detection of bolts.

[0014] Figure 2 The present invention is a camera arrangement diagram of a visual inspection method for fluorescent magnetic particle flaw detection of bolts.

[0015] Figure 3 The present invention is a structural diagram of equipment used in a visual inspection method for fluorescent magnetic particle flaw detection of bolts.

[0016] Figure 4 The present invention is a structural diagram of a spray magnetization module of a visual inspection system for fluorescent magnetic particle flaw detection of bolts.

[0017] Figure 5 The present invention is a structural diagram of a CCD visual recognition module of a visual inspection system for fluorescent magnetic particle flaw detection of bolts. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0019] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] like Figure 1 Figure 2 A visual inspection method for bolt fluorescent magnetic particle flaw detection is shown in the figure. The specific flaw detection method is as follows: S1: Bolt loading: The bolts placed at the loading station are clamped by a combined cylinder and placed on the bolt fluorescent magnetic particle flaw detection line in sequence. The bolt fluorescent magnetic particle flaw detection line feeds two workpieces each time, and can perform double-station flaw detection at the same time; S2: Spray magnetization of bolts: The bolts on the bolt fluorescent magnetic particle inspection line are transported to the spray magnetization station, the bolts are clamped, lifted and rotated, and magnetic suspension is sprayed on the rotating bolts at the same time; the bolt magnetization method can be selected as circumferential magnetization, longitudinal magnetization and composite magnetization; the direct current method is used to detect longitudinal cracks in the thread; the coil induction method is used to detect circumferential defects of the bolts; S3: CCD automatic identification of bolts: The bolts that have been sprayed and magnetized are transported to the visual inspection box; a black light is used in the visual inspection box to cooperate with a bolt circumferential identification camera, a bolt end identification camera, and a bolt neck identification camera to inspect the bolts; the CCD automatic identification system is used to collect and compare images of the bolts to identify defective bolts; S4: Bolt sorting and demagnetization: After the bolts are automatically detected and identified by CCD, the qualified bolts are grabbed to another production line through the PLC control sorting module, and the bolts are demagnetized and unloaded from the qualified product station; the unqualified and defective bolts are output from the defective product station along the original production line.

[0021] S1: The flaw detection cycle of double-station flaw detection in bolt feeding is 13-14 seconds / 2 pieces; S2: Circumferential magnetization current in spray magnetization of bolts: AC is 0-3000A effective value, touch screen display, continuously adjustable, with power-off phase control; Longitudinal magnetization magnetic potential: AC is 0-18000AT effective value, touch screen display, continuously adjustable, with power-off phase control, residual magnetic intensity ≤0.2mT; S4: In the sorting of bolts, the demagnetization magnetic potential is AC15000-0AT effective value, continuously adjustable.

[0022] S3: In the CCD automatic identification of bolts, the illumination of the black light reaches more than 4500μw / cm2 at 380mm from the center of the light, and a filter is set on the black light, and a polarizing box UV filter is added to each camera to ensure that only ultraviolet light of a specific wavelength is emitted, and only green fluorescence enters the camera lens; the flaw detection sensitivity is clearly displayed using A-type 2# test piece.

[0023] S2: In the spray magnetization of bolts, the circumferential magnetization current is I=8D~15D; where I is the effective value of the current, in ampere (A); D is the diameter of the specimen or the maximum width, in millimeters (mm); longitudinal magnetization current I=4π×d×H; where I is the effective value of the current, in ampere (A); d is the distance between the cable and the surface to be tested, in millimeters (mm); H is the tangential field strength, in kiloamperes per meter (A / m); fluorescent magnetic powder in magnetic suspension: 2-3 g / L; dispersant: 0.2% of the weight of magnetic suspension; rust inhibitor: 0.5% of the weight of magnetic suspension; defoaming agent: 0.5-1 g / L.

[0024] S3: In the CCD automatic recognition of bolts, the CCD automatic recognition system uses six cameras for simultaneous detection, of which three cameras detect the bolt surface from above, two cameras detect the two ends of the bolt from the side, and one camera is set at an angle to detect the neck position of the hexagonal bolt; The CCD automatic recognition system has the function of automatic image acquisition: the camera and lens use industrial-grade 2D cameras and 10 million-level high-definition lenses, global shutter, dynamic shooting, frame rate 36fps; the number of pixels is 10 million, and the resolution is not less than 4096 × 2460; the shooting range of a single camera is greater than 300mm*200mm, and the detection accuracy is greater than 20pix / mm; The CCD automatic recognition system has the function of automatic analysis and judgment, and can realize sample library production, learning and training, and deployment. The convolutional neural network is used to train and learn a large number of images collected in the image sample library. After the training, we will get the corresponding neural network model weight file. This file will be added to the forward reasoning prediction software, and it can be used for automatic crack detection. The CCD automatic recognition system has the function of automatic storage and recording; it has the function of saving NG images and automatically clearing images regularly; The main window of the detection software of the CCD automatic recognition system should display the detection screen in real time. The number of windows can be customized and the windows are arranged in a matrix. The software processing window contains the detection time of each beat, the detection target category, and the score / confidence information. The software interface should display the version number, current user, server status, camera status, detection results, and detection operation process information. The detection software of the CCD automatic identification system has three detection modes: Offline detection mode, used to select local image samples for offline detection training and optimize learning algorithms; Continuous detection mode is used for the software to automatically perform online real-time detection. In the continuous detection mode, when the software is in the state of executing detection, when the workpiece being detected moves under the camera's field of view, the software can automatically track and lock the crack; Manual detection mode is used for the software to automatically complete a detection process; the software will be in the following states: Waiting for detection: When entering the manual detection mode, the software is in the waiting for detection state; Capture image: When a workpiece enters the camera's field of view, click the button and the software will capture the image; Execute inspection: After the image is captured, click the button to execute the inspection and analyze whether there are defects; Send results: After the analysis is completed, click the button to send the test results and display them in the status bar; The detection software of the CCD automatic recognition system has a camera debugging module, which is used for parameter configuration and debugging of the camera, including initialization, image acquisition, and parameter setting; The detection software of the CCD automatic identification system has a communication setting module for configuring the parameters for communication with the controller; The detection software of the CCD automatic identification system must have a user login module to facilitate user management and permission setting; The detection software of the CCD automatic recognition system identifies the defects, and the equipment automatically sorts the defective workpieces to the defective product station.

[0025] like Figures 3 to 5 As shown: A visual inspection system for bolt fluorescent magnetic particle flaw detection, comprising a conveyor frame 1, a feeding module 2, a spray magnetization module 3, a CCD visual recognition module 4 and a bolt sorting module 5; The conveyor frame 1 is provided with a loading station, a spray magnetization station, a CCD visual recognition station and a bolt sorting station in sequence along the extension direction; the conveyor frame is provided with a conveyor chain, and a plurality of bolt placement seats are provided on the conveyor chain at equal intervals; The loading module 2 is arranged at the loading station of the conveying frame, and the loading module 2 is a clamping claw structure driven by a cylinder, which is used to grab the bolts and place them on the bolt placement seat on the conveying frame 1; The spray magnetization module 3 is arranged on the spray magnetization station of the conveyor, and the spray magnetization module 3 includes a support frame 31, a lifting clamping frame 32, a rotating cylinder 33, a magnetic suspension liquid collecting tank 34 and a circulating spray pipeline 35; the support frame 31 has a pair of and is respectively arranged on the two side surfaces of the extending direction of the conveyor frame 1 and is arranged oppositely; the lifting clamping frame 32 has a pair of and is respectively arranged on the support frame 31 and is arranged oppositely on both sides of the conveyor frame 1; the rotating cylinder 33 is horizontally installed on the lifting clamping frame, and is arranged oppositely on the two lifting clamping frames 32, and the lifting clamping frame 32 drives the rotating cylinder 33 to move in the horizontal direction and the vertical direction, so as to adjust the conveyor frame 1 is clamped by bolts between the rotating cylinders 33 on both sides and drives the bolts to move, and the rotating cylinders 33 mutually clamp the bolts and drive the bolts to rotate; a cover is arranged between the lifting clamping frame 32 across the conveying frame; a magnetic suspension liquid collecting tank 34 is arranged below the conveying frame 1, and a magnetic suspension liquid is arranged in the magnetic suspension liquid collecting tank 34; a circulating spray pipe 35 is installed on the inner wall of the cover, and a plurality of spray ports aligned with the bolts on the conveying frame are arranged on the circulating spray pipe 34; a circulating pump is connected to the circulating spray pipe 34 from above, and the magnetic suspension liquid in the magnetic suspension liquid collecting tank is pumped into the circulating spray pipe 34 by the circulating pump for spraying the screw; The CCD visual recognition module 4 is arranged at the CCD visual recognition station on the conveying frame 1, and a visual inspection box 41 is arranged at the CCD recognition station, and the conveying frame 1 is arranged through the visual inspection box 41; a detection table is arranged in the visual inspection box 41; the CCD visual recognition module 4 includes a camera group 42 and a black light 43; the camera group 42 includes a top recognition camera, an end recognition camera and a bolt neck recognition camera, which respectively recognize the circumference, end and neck of the bolt; the black light 43 is arranged on the inner wall near the top on both sides of the visual inspection box 41 to illuminate the bolts in the visual inspection box 41; a bolt clamping transfer clamp 44 is arranged in the visual inspection box 41, and the bolts at the input end position of the input visual inspection box are grabbed and placed on the detection table through the bolt clamping transfer clamp 44, and the bolts that have completed the inspection are grabbed and placed at the output end position of the visual inspection box 41; The bolt sorting module 5 is arranged at the bolt sorting station of the conveying frame 1, and the bolt sorting module 5 is a clamping claw structure driven by a cylinder, which is used to grab the qualified bolts from the bolt placement seat.

[0026] Those skilled in the art should understand that the present invention is not limited to the above-mentioned embodiments. The above-mentioned embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may be subject to various changes and improvements, and these changes and improvements shall all fall within the scope of the present invention claimed for protection.

Claims

1. A visual inspection method for bolt fluorescent magnetic particle flaw detection, characterized in that: The specific flaw detection methods are as follows: S1: Bolt loading: The bolts placed at the loading station are clamped by a combined cylinder and placed on the bolt fluorescent magnetic particle flaw detection line in sequence. The bolt fluorescent magnetic particle flaw detection line feeds two workpieces each time, and can perform double-station flaw detection at the same time; S2: Spray magnetization of bolts: The bolts on the bolt fluorescent magnetic particle inspection line are transported to the spray magnetization station, the bolts are clamped, lifted and rotated, and magnetic suspension is sprayed on the rotating bolts at the same time; the bolt magnetization method can be selected as circumferential magnetization, longitudinal magnetization and composite magnetization; the direct current method is used to detect longitudinal cracks in the thread; the coil induction method is used to detect circumferential defects of the bolts; S3: CCD automatic identification of bolts: The bolts that have been sprayed and magnetized are transported to the visual inspection box; a black light is used in the visual inspection box to cooperate with a bolt circumferential identification camera, a bolt end identification camera, and a bolt neck identification camera to inspect the bolts; the CCD automatic identification system is used to collect and compare images of the bolts to identify defective bolts; S4: Bolt sorting and demagnetization: After the bolts are automatically detected and identified by CCD, the qualified bolts are grabbed to another production line through the PLC control sorting module, and the bolts are demagnetized and unloaded from the qualified product station; the unqualified and defective bolts are output from the defective product station along the original production line.

2. A visual inspection method for bolt fluorescent magnetic particle flaw detection according to claim 1, characterized in that: The S1: the flaw detection cycle of the double-station flaw detection in the loading of bolts is 13-14 seconds / 2 pieces; the S2: the circumferential magnetization current in the spray magnetization of bolts: AC is 0-3000A effective value, touch screen display, continuously adjustable, with power-off phase control; longitudinal magnetization magnetic potential: AC is 0-18000AT effective value, touch screen display, continuously adjustable, with power-off phase control, residual magnetic intensity ≤0.2mT; the S4: in the sorting of bolts, the demagnetization magnetic potential is AC15000-0AT effective value, continuously adjustable.

3. A visual inspection method for bolt fluorescent magnetic particle flaw detection according to claim 1, characterized in that: S3: In the CCD automatic identification of bolts, the illumination of the black light reaches 4500μw / cm at 380mm from the center of the light. 2 The above, and set the filter on the black light, and add the polarizing box UV filter on each camera to ensure that the emitted light is only ultraviolet light of a specific wavelength, and the light entering the camera lens is only green fluorescence; The flaw detection sensitivity is clearly displayed using A-type 2# test piece.

4. A visual inspection method for bolt fluorescent magnetic particle flaw detection according to claim 1, characterized in that: S2: In the spray magnetization of the bolt, the circumferential magnetization current is I=8D~15D; where I is the effective value of the current, in ampere (A); D is the diameter or maximum width of the test piece, in millimeters (mm); longitudinal magnetization current I=4π×d×H; where I is the effective value of the current, in ampere (A); d is the distance between the cable and the surface to be tested, in millimeters (mm); H is the tangential field strength, in kiloamperes per meter (A / m); fluorescent magnetic powder in magnetic suspension: 2-3 grams / liter; Dispersant: 0.2% by weight of magnetic suspension; Rust inhibitor: 0.5% by weight of magnetic suspension; Defoaming agent: 0.5-1 g / L.

5. A visual inspection method for bolt fluorescent magnetic particle flaw detection according to claim 1, characterized in that: In the S3: CCD automatic identification of bolts, the CCD automatic identification system uses six cameras for simultaneous detection, wherein three cameras detect the bolt surface from above, two cameras detect both ends of the bolt from the side, and one camera is tilted and is dedicated to detecting the neck position of the hexagonal bolt; The CCD automatic recognition system has the function of automatic image acquisition: the camera and lens use industrial-grade 2D cameras and 10 million-level high-definition lenses, global shutter, dynamic shooting, frame rate 36fps; the number of pixels is 10 million, and the resolution is not less than 4096 × 2460; the shooting range of a single camera is greater than 300mm*200mm, and the detection accuracy is greater than 20pix / mm; The CCD automatic recognition system has the function of automatic analysis and judgment, and can realize sample library production, learning and training, and deployment. The convolutional neural network is used to train and learn a large number of images collected in the image sample library. After the training, we will get the corresponding neural network model weight file. This file will be added to the forward reasoning prediction software, and it can be used for automatic crack detection. The CCD automatic recognition system has the function of automatic storage and recording; it has the function of saving NG images and automatically clearing images regularly; The main window of the detection software of the CCD automatic recognition system should display the detection screen in real time. The number of windows can be customized and the windows are arranged in a matrix. The software processing window contains the detection time of each beat, the detection target category, and the score / confidence information. The software interface should display the version number, current user, server status, camera status, detection results, and detection operation process information. The detection software of the CCD automatic identification system has three detection modes: Offline detection mode, used to select local image samples for offline detection training and optimize learning algorithms; Continuous detection mode is used for the software to automatically perform online real-time detection. In the continuous detection mode, when the software is in the state of executing detection, when the workpiece being detected moves under the camera's field of view, the software can automatically track and lock the crack; Manual detection mode is used for the software to automatically complete a detection process; the software will be in the following states: Waiting for detection: When entering the manual detection mode, the software is in the waiting for detection state; Capture image: When a workpiece enters the camera's field of view, click the button and the software will capture the image; Execute inspection: After the image is captured, click the button to execute the inspection and analyze whether there are defects; Send results: After the analysis is completed, click the button to send the test results and display them in the status bar; The detection software of the CCD automatic recognition system has a camera debugging module, which is used for parameter configuration and debugging of the camera, including initialization, image acquisition, and parameter setting; The detection software of the CCD automatic identification system has a communication setting module for configuring the parameters for communication with the controller; The detection software of the CCD automatic identification system must have a user login module to facilitate user management and permission setting; The detection software of the CCD automatic recognition system identifies the defects, and the equipment automatically sorts the defective workpieces to the defective product station.

6. A visual inspection system for bolt fluorescent magnetic particle flaw detection, characterized in that: It includes a conveyor rack, a loading module, a spray magnetization module, a CCD visual recognition module and a bolt sorting module; The conveyor frame is provided with a loading station, a spray magnetization station, a CCD visual recognition station and a bolt sorting station in sequence along the extension direction; the conveyor frame is provided with a conveyor chain, and a plurality of bolt placement seats are provided on the conveyor chain at equal intervals; The loading module is arranged at the loading station of the conveying frame, and the loading module is a clamping claw structure driven by a cylinder, which is used to grab the bolts and place them on the bolt placing seat on the conveying frame; The spray magnetization module is arranged on the spray magnetization station of the conveyor, and the spray magnetization module includes a support frame, a lifting clamping frame, a rotating cylinder, a magnetic suspension collecting tank and a circulating spray pipeline; the support frame has a pair of and are respectively arranged on the two side surfaces of the extending direction of the conveyor frame and are arranged oppositely; the lifting clamping frame has a pair of and are respectively arranged on the support frame and are arranged oppositely on both sides of the conveyor frame; the rotating cylinder is horizontally installed on the lifting clamping frame and is arranged oppositely on the two lifting clamping frames, and the lifting clamping frame drives the rotating cylinder to move in the horizontal direction and the vertical direction, so as to move the magnetic suspension liquid through the two lifting clamping frames on the conveyor frame. The bolts between the side rotating cylinders are clamped and driven to move, and the rotating cylinders are mutually pressed to clamp the bolts and drive the bolts to rotate; a cover shell is arranged across the conveying frame between the lifting clamping frames; the magnetic suspension liquid collecting tank is arranged below the conveying frame, and the magnetic suspension liquid is arranged in the magnetic suspension liquid collecting tank; the circulating spray pipe is installed on the inner wall of the cover shell, and the circulating spray pipe is provided with a plurality of spray ports aligned with the bolts on the conveying frame; the circulating spray pipe is connected to a circulating pump from above, and the magnetic suspension liquid in the magnetic suspension liquid collecting tank is pumped into the circulating spray pipe by the circulating pump for spraying the screw; The CCD visual recognition module is arranged at the CCD visual recognition station on the conveyor frame, and a visual inspection box is arranged at the CCD recognition station, and the conveyor frame is arranged through the visual inspection box; an inspection table is arranged in the visual inspection box; the CCD visual recognition module includes a camera group and a black light; the camera group includes a top recognition camera, an end recognition camera and a bolt neck recognition camera, which respectively recognize the circumference, end and neck of the bolt; the black light is arranged on the inner wall near the top on both sides of the visual inspection box to illuminate the bolts in the visual inspection box; a bolt clamping and transfer clamp is arranged in the visual inspection box, and the bolts at the input end position of the visual inspection box are grabbed and placed on the inspection table by the bolt clamping and transfer clamp, and the bolts that have completed the inspection are grabbed and placed at the output end position of the visual inspection box; The bolt sorting module is arranged at the bolt sorting station of the conveying frame, and the bolt sorting module is a clamping claw structure driven by a cylinder, which is used to grab the qualified bolts from the bolt placement seat.

Images