Glass wool defect detection and sorting device based on machine vision

Abstract

The application provides a glass wool defect detection and sorting device based on machine vision. The glass wool defect detection and sorting device comprises a conveying system, a shearing mechanism, a detection unit and a central control unit. The conveying system comprises a first roller group and a second roller group along the downward direction of the glass wool. The first roller group runs at a speed, and the second roller group has a speed and a roller gap adjusting function. The shearing mechanism is arranged below the second roller group. The detection unit is arranged between the first roller group and the second roller group, and is used for real-time acquisition of a glass wool surface image and identification of a defect area. The central control unit is electrically connected with the shearing mechanism and the detection unit. The glass wool defect detection and sorting device realizes accurate positioning and efficient isolation of the defect section, avoids the missed detection and misdiagnosis problems caused by subjective factors in traditional manual detection, and significantly improves the product quality control level.

Description

Technical Field

[0001] This application relates to the field of detection device technology, and in particular to a machine vision-based glass wool defect detection and sorting device. Background Technology

[0002] In the industrialized continuous production of glass wool, the stability of product quality directly affects its thermal insulation, heat insulation, and sound absorption performance. During the molding and curing process, glass wool products are highly susceptible to various surface or structural defects, such as uneven fiber distribution (flocculation), dimensional deviations, and poor curing, due to fluctuations in raw materials, changes in centrifugal fiber forming process parameters, uneven temperature distribution in the curing oven, or interference from environmental factors. If these defects are not detected and addressed in a timely and accurate manner, they will directly affect the quality grade of the final product and may even pose safety hazards in subsequent applications.

[0003] In existing technologies, the detection of defective sections on glass wool production lines mainly relies on manual visual inspection or manual periodic sampling inspection. Operators typically observe the surface of the glass wool with the naked eye next to the production line, or conduct random checks on the products after they come off the line to determine if defects are present. However, manual inspection is highly susceptible to the subjective factors of the operators (such as the degree of concentration, fatigue, and differences in personal experience), making it difficult to guarantee the accuracy and consistency of defect detection. This leads to missed or false detections, resulting in poor real-time performance and low detection rates. This fails to meet the online full inspection requirements of continuous, high-speed glass wool production processes, causing quality control issues.

[0004] Therefore, it is necessary to propose a machine vision-based glass wool defect detection and sorting device to solve the technical problems mentioned in the background. Summary of the Invention

[0005] This application provides a machine vision-based glass wool defect detection and sorting device to solve the problems of poor real-time performance and low detection rate of manual inspection and separation. It achieves accurate positioning and efficient isolation of defective segments, and significantly improves the level of product quality control.

[0006] This application provides a machine vision-based glass wool defect detection and sorting device, including:

[0007] The conveying system, along the downward direction of the glass wool, includes a first roller group and a second roller group, wherein the first roller group moves at a speed of The second roller group is running at a speed of [speed value missing]. It also has a roller gap adjustment function; among which ;

[0008] The shearing mechanism is located below the second roller group and is at a predetermined distance from the second roller group. ;

[0009] The detection unit is located between the first roller group and the second roller group. It is used to acquire images of the glass wool surface in real time and identify defect areas. When the starting point of the defect is detected to cross the lower end of the second roller group, a first defect signal is immediately generated. When the ending point of the defect is detected to cross the lower end of the second roller group, a second defect signal is immediately generated.

[0010] The central control unit, electrically connected to the shearing mechanism and the detection unit, receives a first defect signal and a second defect signal. Upon receiving the first defect signal, it combines it with a delayed signal to generate a first shearing signal, which is then sent to the shearing mechanism to shear the defect starting point. Upon receiving the second defect signal, it combines it with a delayed signal to generate a second shearing signal, which is then sent to the shearing mechanism to shear the defect ending point. The time difference between receiving the second defect signal and the first defect signal is specified. Greater than the delayed signal .

[0011] In a possible design of a machine vision-based glass wool defect detection and sorting device, the detection unit includes a line scan camera and an image preprocessing module.

[0012] The linear scan camera acquires real-time surface image streams of continuously falling glass wool, and the encoder synchronously obtains the real-time operating speed of the production line. and position pulse signal;

[0013] The image preprocessing module is configured to perform grayscale correction and noise filtering on the acquired raw images, construct a standardized input image sequence, and extract defect feature vectors using a pre-trained lightweight convolutional neural network model, outputting the category label and pixel-level bounding box for each defect region.

[0014] The central control unit is electrically connected to the line scan camera and the image preprocessing module, respectively.

[0015] In a possible design of a machine vision-based glass wool defect detection and sorting device, the detection unit further includes a high-frequency light source. The high-frequency light source adopts a double-sided symmetrical lighting or coaxial diffuse reflection lighting method, closely surrounding both sides of the line scan camera lens or integrated into the periphery of the lens. The high-frequency light source is triggered synchronously with the line scan camera's line frequency signal.

[0016] In a possible design of a machine vision-based glass wool defect detection and sorting device, the algorithm for determining the delayed signal is as follows:

[0017] Based on formula Calculate the time difference between the first shear signal and the first defect signal received by the central control unit. Or the central control unit receives the time difference between the second shear signal and the second defect signal. ,in, The distance between the contact point between the second roller group and the glass wool and the contact point between the shearing mechanism and the glass wool is [missing information]. The correction value time for the shearing mechanism to receive the first shearing signal or the second shearing signal before performing the action.

[0018] In a possible design of a machine vision-based glass wool defect detection and sorting device, the conveying system includes a hopper, a cover, and a rotating drive component;

[0019] The cover is configured in multiple ways, and the cover is evenly fixed below the hopper. The rotating drive is fixed on the hopper. The cover includes the first roller group and the second roller group in sequence along the downward direction of the glass wool.

[0020] The first roller group and the second roller group are connected by gear C; the output end of the rotation drive component is fixed to the same axis as the gear C;

[0021] The central control unit is electrically connected to the rotation drive component.

[0022] In a possible design of a machine vision-based glass wool defect detection and sorting device, the first roller group includes roller A, roller B, gear A, and gear B. Roller A and roller B are horizontally arranged and rotatably connected inside the housing. The shaft end of roller A extending out of the housing is coaxially fixed to gear A, and the shaft end of roller B extending out of the housing is coaxially fixed to gear B. Gear A is meshed with gear B.

[0023] The second roller group includes roller C, roller D, gear C and gear D. Roller C and roller D are horizontally arranged. Roller C is rotatably connected to the inside of the cover. The shaft end of roller C extending out of the cover is coaxially fixed to gear C. The shaft end of roller D extending out of the cover is coaxially fixed to gear D. Gear C is meshed with gear D.

[0024] A gear G is rotatably disposed on the cover between gear A and gear C, and gear G meshes with gear A and gear C respectively.

[0025] In a possible design of a machine vision-based glass wool defect detection and sorting device, the shearing mechanism includes a telescopic component;

[0026] Two telescopic components are configured and symmetrically fixed inside the housing;

[0027] The central control unit is electrically connected to the telescopic component.

[0028] In a possible design of a machine vision-based glass wool defect detection and sorting device, the shearing mechanism further includes a gap adjustment component, a shearing component, and an adapter.

[0029] The gap adjustment assembly includes a limiting groove, a guide plate, a slide rod, and a connecting shaft; two limiting grooves, two guide plates, two slide rods, and two connecting shafts are configured. The limiting grooves are symmetrically opened inside the housing. The slide rod is slidably connected to both the limiting groove and the housing end away from the roller C. The end of the slide rod away from the limiting groove is rotatably connected to the connecting shaft. The guide plate is slidably disposed in the limiting groove. The roller D is rotatably connected between the guide plates. The gear D is coaxially fixed at the shaft end of the roller D extending from the guide plate.

[0030] The shearing assembly includes a positioning element, a moving blade, and a crankshaft. The positioning element is fixed inside the housing directly below the roller C. One end of the positioning element near the roller D is flush with the circumferential contact between the roller C and the glass wool. The moving blade is slidably connected inside the housing directly below the roller D. The cutting edge of the moving blade faces the positioning element. Two crankshafts are symmetrically and rotatably arranged at the end of the moving blade away from the positioning element.

[0031] Both the connecting shaft and the crankshaft are rotatably connected to the circumferential direction of the adapter, and the adapter is fixed to the output end of the telescopic component on the corresponding side.

[0032] In a possible design of a machine vision-based glass wool defect detection and sorting device, a third roller group is also included;

[0033] The third roller group is positioned below the second roller group inside the cover along the downward direction of the glass wool.

[0034] The third roller group includes roller E, roller F, gear E and gear F. Roller E and roller F are horizontally arranged and rotatably connected inside the cover. The shaft end of roller E extending out of the cover is coaxially fixed to gear E. The shaft end of roller F extending out of the cover is coaxially fixed to gear F. Gear E meshes with gear F.

[0035] A gear H is rotatably disposed on the cover between gear C and gear E, and gear H is respectively meshed with gear C and gear E;

[0036] Specifically, the time difference between the glass wool passing the shearing component and the glass wool passing the lower end of the third roller group is less than the time difference in the delayed signal. .

[0037] In a possible design of a machine vision-based glass wool defect detection and sorting device, an end effector is also included;

[0038] The end effector is located below the third roller group and includes an openable clamping plate and a rotary drive component; the openable clamping plate is rotatably connected to the hopper, the rotary drive component is fixed to the hopper below the rotating end of the openable clamping plate, and the output end of the rotary drive component is rotatably connected to the openable clamping plate.

[0039] The central control unit is electrically connected to the rotary drive component.

[0040] Beneficial effects:

[0041] 1. This invention uses a detection unit to capture surface defects in glass wool in real time, and a central control unit precisely controls the timing of the shearing mechanism based on defect and delay signals. After detecting the start and end points of the defects, the system utilizes the dynamic reduction of the gap between the second roller group to both reserve sufficient shearing time for the shearing mechanism and effectively prevent fly wool generated during the shearing process from entering the detection area. This avoids false detections and missed detections caused by fly wool obstructing or contaminating the detection unit, ensuring the continuous stability of defect detection. At the same time, the mechanism of quickly cutting off defective sections reduces the residue and waste of defective products. While improving the quality of finished glass wool and production efficiency, it avoids the problems of missed detections and false detections caused by subjective factors in traditional manual inspection, and achieves accurate positioning and efficient isolation of defective sections, significantly improving the level of product quality control.

[0042] 2. The present invention utilizes the differential speed coordination and timing control between the second and third roller groups. When a defect is detected, the gap between the second roller group is reduced to form a clamping force on the glass wool, providing sufficient time for the shearing action and preventing the upstream glass wool from accumulating. At the same time, the third roller group maintains its original speed and applies tensile stress to the glass wool by utilizing the speed difference between the two roller groups, which can significantly accelerate the shearing and separation process and ensure that the defective section is quickly cut off.

[0043] The above description is merely an overview of the technical solutions of the embodiments of this application. In order to better understand the technical means of the embodiments of this application and to implement them in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the embodiments of this application more obvious and understandable, specific implementation methods of this application are described below. Attached Figure Description

[0044] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0045] Figure 1 This is a schematic diagram of the overall workflow of a machine vision-based glass wool defect detection and sorting device.

[0046] Figure 2 This is a schematic diagram of the overall structure of a machine vision-based glass wool defect detection and sorting device.

[0047] Figure 3 This is a schematic diagram of a multi-stage roller conveyor system in a machine vision-based glass wool defect detection and sorting device.

[0048] Figure 4 This is a schematic diagram of the structure of the first roller group, the second roller group, and the third roller group in a machine vision-based glass wool defect detection and sorting device;

[0049] Figure 5 This is a schematic diagram of the shearing mechanism in a machine vision-based glass wool defect detection and sorting device.

[0050] Figure 6 This is a schematic diagram of the end effector in a machine vision-based glass wool defect detection and sorting device.

[0051] Explanation of reference numerals in the attached figures:

[0052] 1. Hopper; 2. Cover; 3. Roller A; 4. Roller B; 5. Gear A; 6. Gear B; 7. Roller C; 8. Roller D; 9. Gear C; 10. Gear D; 11. Roller E; 12. Roller F; 13. Gear E; 14. Gear F; 15. Gear G; 16. Gear H; 17. Rotary drive component; 18. Limiting groove; 19. Guide plate; 20. Slide rod; 21. Connecting shaft; 22. Telescopic component; 23. Positioning component; 24. Moving blade; 25. Crankshaft; 26. Adapter component; 27. Openable clamping plate; 28. Rotary drive component. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0054] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein in the specification of the application is for the purpose of describing particular embodiments only and is not intended to limit the application; the terms “comprising” and “having”, and any variations thereof, in the specification, claims and drawings of this application are intended to cover non-exclusive inclusion.

[0055] The term "embodiment" as used herein means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of the phrase "embodiment" in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0056] The directional terms appearing in the following description refer to the directions shown in the figures and are not intended to limit the specific structure of this application. For example, in the description of this application, terms such as "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," and "circumferential" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the figures. They are used only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.

[0057] Furthermore, the terms "first," "second," etc., in the specification and claims of this application or in the aforementioned drawings are used to distinguish different objects rather than to describe a specific order, and may explicitly or implicitly include one or more of the features.

[0058] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "joining" should be interpreted broadly. For example, "connection" or "joining" in mechanical structures can refer to a physical connection, such as a fixed connection, for example, a connection fixed by fasteners, such as a connection fixed by screws, bolts, or other fasteners; a physical connection can also be a detachable connection, such as a snap-fit ​​or interlocking connection; a physical connection can also be an integral connection, such as a connection formed by welding, bonding, or integral molding. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0059] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0060] Figures 1-6 A machine vision-based glass wool defect detection and sorting device provided in this application includes a conveying system, a shearing mechanism, a detection unit, and a central control unit. The conveying system includes a first roller group and a second roller group along the downward direction of the glass wool. The first roller group moves at a speed... Running, the speed of the second roller group is It also has a roller gap adjustment function; among which The shearing mechanism is located below the second roller group and at a predetermined distance from it. The detection unit, located between the first and second roller groups, is used to acquire real-time images of the glass wool surface and identify defect areas. When a defect starting point is detected crossing the lower end of the second roller group, a first defect signal is immediately generated; when a defect ending point is detected crossing the lower end of the second roller group, a second defect signal is immediately generated. The central control unit, electrically connected to the shearing mechanism and the detection unit, receives the first and second defect signals. Upon receiving the first defect signal, it combines it with a delayed signal to generate a first shearing signal, which is then sent to the shearing mechanism to shear the defect starting point. Upon receiving the second defect signal, it combines it with a delayed signal to generate a second shearing signal, which is then sent to the shearing mechanism to shear the defect ending point. The time difference between receiving the second and first defect signals is specified. Greater than the delayed signal .

[0061] Using the above technical solution, after the glass wool undergoes preliminary shaping by the first roller group, it enters the conveying process. At this time, the detection unit acquires real-time images of the glass wool surface. When the starting point of a defect is detected to have crossed the lower end of the second roller group, a first defect signal is immediately generated and transmitted to the central control unit. Upon receiving the first defect signal, the central control unit combines it with a delay signal to control the activation of the shearing mechanism. After the shearing mechanism is activated, the gap between the second roller group dynamically narrows, clamping the defective section. Simultaneously, the shearing mechanism shears and resets the starting end of the defective section, restoring the gap between the second roller group.

[0062] The central control unit continues to track the length of the defective segment. Once the defective segment has completely passed the second roller group, a second defect signal is immediately generated and transmitted to the central control unit. Upon receiving the second defect signal, the central control unit combines it with a delay signal and determines the time difference between receiving the second defect signal and the first defect signal. Greater than the delayed signal That is, the length of the defective section must be greater than the set distance between the second roller group and the shearing mechanism. Only then can the glass wool be allowed to perform the shearing mechanism after passing the second roller group. At this time, the shearing mechanism is activated and the above clamping and shearing actions are repeated to cut off the end of the defective section and completely isolate the defective section.

[0063] The aforementioned technical solution overcomes the shortcomings of traditional manual visual inspection or sampling inspection, which relies on subjective judgment and suffers from poor real-time performance. By utilizing the collaboration between the detection unit and the central control unit, it achieves automatic defect identification and precise location, avoiding missed or false detections caused by operator fatigue or lack of concentration. Simultaneously, the deceleration clamping function of the second roller group not only allows sufficient shearing time for the shearing mechanism but also effectively prevents fly wool generated during the shearing process from entering the detection area, avoiding false or missed detections caused by fly wool obstructing or contaminating the detection unit, thus ensuring the continuous stability of defect detection. It also effectively prevents the accumulation of glass wool between the first and second roller groups due to speed differences, ensuring the continuous and stable operation of the production line.

[0064] Secondly, the two precise shearing cuts of the shearing mechanism ensure that defective sections are completely removed, significantly improving the defect detection rate and sorting accuracy. This meets the online full inspection requirements of continuous and high-speed glass wool production, effectively controls product quality, and reduces the risk of quality control failure due to defect omissions.

[0065] In one possible implementation, the detection unit includes a line scan camera and an image preprocessing module. The line scan camera acquires a real-time image stream of the surface of the continuously falling glass wool, and the encoder synchronously acquires the real-time operating speed of the production line. The image preprocessing module is configured to perform grayscale correction and noise filtering on the acquired raw images, construct a standardized input image sequence, and extract defect feature vectors using a pre-trained lightweight convolutional neural network model, outputting the category label and pixel-level bounding box for each defect region. The central control unit is electrically connected to the line scan camera and the image preprocessing module.

[0066] Using the above technical solution, during the continuous glass wool feeding process, a line scan camera scans the glass wool surface at a high frame rate, acquiring a continuous stream of ribbon-like surface images in real time. Simultaneously, an encoder monitors the production line's operating status in real time, obtaining the current operating speed. The system transmits position pulse signals and these motion data synchronously to the central control unit to ensure a one-to-one correspondence between image pixels and the actual physical position of the glass wool, eliminating image stretching or compression caused by production line speed fluctuations.

[0067] After the central control unit transmits the acquired raw image stream to the image preprocessing module, the module first performs grayscale correction on the raw image to eliminate the influence of uneven ambient lighting on imaging. Then, it performs noise filtering to remove sensor noise or environmental interference signals, thereby constructing a clear and standardized input image sequence, providing a high-quality data foundation for subsequent intelligent recognition.

[0068] The preprocessed image sequence is then input into a pre-trained lightweight convolutional neural network model, which automatically extracts deep defect feature vectors from the image and accurately identifies the type of defect (such as impurities, broken wires, holes, etc.) through classification and regression algorithms, and calculates the pixel-level bounding box coordinates of each defect region.

[0069] Simultaneously, after receiving the defect category label and location information output by the image preprocessing module, the central control unit combines the position pulse signal of the encoder to convert the pixel coordinates into actual physical coordinates on the production line, thereby accurately locking the specific location of the defect on the glass wool and providing a precise trigger signal for subsequent shearing or rejection actions.

[0070] In the above scheme, the use of a linear scan camera combined with encoder synchronous triggering technology can adapt to the continuous material feeding production environment of glass wool, effectively avoiding the motion blur problem caused by traditional area scan cameras during high-speed movement. This achieves full coverage and no omissions in scanning the glass wool surface, significantly improving detection efficiency. Furthermore, through grayscale correction and noise filtering in the image preprocessing module, the imaging quality problems caused by the fluffiness, reflection, or changes in lighting on the glass wool surface are effectively solved, reducing the false detection rate. Combined with a lightweight convolutional neural network model, it can accurately extract the features of minute defects while ensuring inference speed, achieving high-precision identification of subtle defects against complex texture backgrounds.

[0071] It should be added that using pre-trained lightweight convolutional neural network models to extract defect feature vectors is a common technique in the current field of industrial visual inspection, such as the common application of MobileNet, etc. This is an existing technology and will not be elaborated on in detail.

[0072] In one possible implementation, the detection unit also includes a high-frequency light source, which employs a bilateral symmetrical lighting or coaxial diffuse reflection lighting method, closely surrounding both sides of the line scan camera lens or integrated into the periphery of the lens, and the high-frequency light source is triggered synchronously with the line scan camera's line frequency signal.

[0073] Based on the above technical solutions, the high-speed motion blur can be effectively eliminated by the synchronous triggering mechanism of high-frequency light source and line scan camera. Combined with the bilateral symmetrical or coaxial diffuse reflection lighting method, the image signal-to-noise ratio and the visibility of surface defects are significantly improved.

[0074] Simultaneously, the central control unit, based on the logical judgment of the defect length, only triggers a response when a significant defect is confirmed and there is a sufficient cutting window. This avoids interference with the normal production process caused by erroneous actions, improves the stability and intelligence level of the overall device, effectively prevents quality control failure caused by the omission of defect segments, and significantly improves product quality consistency and production automation.

[0075] It should be added that the specific logic for defect segment tracking and endpoint determination is as follows:

[0076] S1. Establish a real-time mapping relationship between the image pixel coordinate system and the production line encoder pulse count. Physically truncate the normal cotton and defective cotton upstream of the defect section to form an isolation zone. Define the position of the first defect signal as the defect starting point. ,by Based on the benchmark, combined with the real-time speed of the production line And encoder pulses, tracking the movement trajectory of the defective segment downwards;

[0077] S2. Calculate the displacement of the defect region's endpoint edge in physical space in real time. Use image processing algorithms to monitor the endpoint characteristics of the defect segment in real time. When it is determined that the endpoint of the defect segment has completely passed through the lower exit of the second roller group, the position of the second defect signal is defined as the defect endpoint. This allows for the identification of the complete defect segment to be removed. , ];

[0078] S3. When the coordinate value of the endpoint of the defect area exceeds the preset critical threshold of the lower exit of the second roller group, an endpoint locking signal is generated, and the time difference between the starting point and the endpoint locking is recorded. Based on the formula Calculate the actual physical length of the defective segment to be sorted. It should be greater than the set distance between the second roller group and the shearing mechanism. .

[0079] In one possible implementation, the algorithm for determining the delayed signal is based on the formula... Calculate the time difference between the first shear signal and the first defect signal received by the central control unit. Or the time difference between the second shear signal and the second defect signal received by the central control unit. ,in, The distance between the contact point between the second roller group and the glass wool and the contact point between the shearing mechanism and the glass wool. The correction time for the shearing mechanism to receive the first shearing signal or the second shearing signal before performing the action.

[0080] It should be noted that in the actual operation of the glass wool production line, the first roller group is mainly responsible for pulling the glass wool forward at the standard speed of the production line, while the second roller group is located upstream of the shearing mechanism, and its core function is to clamp the glass wool to cooperate with the shearing action. Because the second roller group generates a certain frictional resistance and deformation constraint on the glass wool during the clamping process, the actual running speed of the glass wool in the contact area between the second roller group and the shearing mechanism is lower than the theoretical speed of the production line.

[0081] If the time it takes for a defect to reach the shearing cutter is calculated solely based on the theoretical speed of the production line, the speed difference will cause deviations in the shearing timing, resulting in inaccurate defect removal positions. Furthermore, there is an inherent physical delay between the shearing mechanism receiving the electrical signal and the mechanical components completing their response. Therefore, calculations based on the actual distance and speed, along with added correction values ​​for the time, must be introduced to ensure precise closure of the shearing mechanism upon defect arrival.

[0082] In the above technical solution, the delayed signal determination algorithm introduces a correction value time. Furthermore, compensation is made based on the actual operating conditions of the second roller group. This fully considers the deceleration effect of the glass wool caused by the clamping of the second roller group. Through corrective calculations, the positional deviation caused by the inconsistency between the actual operating speed and the theoretical speed of the production line is compensated. This ensures that regardless of fluctuations in the production line speed, the system can accurately predict the exact moment when the defect reaches the shearing point, avoiding premature or delayed cutting due to speed estimation errors.

[0083] In one possible implementation, the conveying system includes a hopper 1, a housing 2, and a rotary drive 17. Multiple housings 2 are configured and uniformly fixed below the hopper 1. The rotary drive 17 is fixed to the hopper 1. Each housing 2 contains a first roller group and a second roller group sequentially along the downward direction of the glass wool. The first roller group and the second roller group are connected by a gear C9. The output end of the rotary drive 17 is coaxially fixed with the gear C9. A central control unit is electrically connected to the rotary drive 17.

[0084] Using the above scheme, the central control unit sends a control command to the rotation drive 17 based on the feedback signal from the detection unit. After the rotation drive 17 starts, it outputs rotational torque, and the output end of the rotation drive 17 drives the gear C9, which is fixed coaxially with it, to rotate. Since the first roller group and the second roller group are both connected by gear C9, the power is synchronously transmitted to the first roller group and the second roller group inside the cover 2, driving the two roller groups to rotate relative to each other at a set speed ratio.

[0085] After the glass wool raw material falls from the hopper 1, it enters several evenly distributed covers 2 below. Driven by the first roller group, the glass wool is initially shaped and conveyed in the downward direction; then it enters the area of ​​the second roller group. Due to the rigid connection of the gear transmission, the two roller groups maintain a strict phase and speed relationship to ensure the stability of the glass wool during the transmission process.

[0086] The above scheme uses gear C9 to connect the first and second roller groups, achieving a hard-connected synchronous drive. Compared to belt drives or independent multi-motor drives, gear drives eliminate accumulated errors and slippage, ensuring strict matching of the linear velocities of the two roller groups. This is especially important for fluffy and easily deformable glass wool, effectively preventing excessive stretching and breakage or wrinkling of the glass wool due to speed differences between roller groups, thus ensuring product integrity.

[0087] In one possible implementation, the first roller group includes roller A3, roller B4, gear A5, and gear B6. Roller A3 and roller B4 are horizontally arranged and rotatably connected to the housing 2. Roller A3 extends out of the shaft end of the housing 2 and coaxially fixes gear A5. Roller B4 extends out of the shaft end of the housing 2 and coaxially fixes gear B6. Gear A5 meshes with gear B6. The second roller group includes roller C7, roller D8, gear C9, and gear D10. Roller C7 and roller D8 are horizontally arranged. Roller C7 is rotatably connected to the housing 2. Roller C7 extends out of the shaft end of the housing 2 and coaxially fixes gear C9. Roller D8 extends out of the shaft end of the housing 2 and coaxially fixes gear D10. Gear C9 meshes with gear D10. Gear G15 is rotatably arranged on the housing 2 between gears A5 and C9, and gear G15 meshes with gears A5 and C9 respectively.

[0088] Based on the above scheme, the rotation drive 17 receives the command from the central control unit and starts, and its output end drives the gear C9 to rotate; the gear C9 transmits power synchronously to the gear A5 of the first roller group through the meshing gear G15; the rollers A3 and B4 of the first roller group achieve equal speed and opposite rotation through the meshing of gear A5 and gear B6, and perform preliminary shaping and conveying of glass wool; the rollers C7 and D8 of the second roller group also achieve equal speed and opposite rotation through the meshing of gear C9 and gear D10. After setting the transmission ratio of gear C9 and gear A5, the second roller group and the first roller group form a speed difference, which provides the power basis for clamping and shearing the defect section.

[0089] When a defective section is detected, the speed difference between the first and second roller sets creates tension between the rollers. Through the cooperation of the central control unit and the shearing mechanism, the defective section is precisely cut off and isolated, while the normal cotton continues to be conveyed smoothly.

[0090] In one possible implementation, the shearing mechanism includes telescopic members 22. Two telescopic members 22 are configured and symmetrically fixed within the housing 2. A central control unit is electrically connected to the telescopic members 22.

[0091] In the above technical solution, the central control unit receives the precise shearing signal from the detection unit and the delay judgment algorithm. After internal logic operation, it sends a synchronous action command to the two telescopic components 22 (such as pneumatic cylinders or electric push rods) installed in the housing 2. After receiving the electrical signal, the two telescopic components 22 simultaneously generate linear reciprocating motion.

[0092] Based on this, two symmetrically arranged telescopic components 22 are used to drive the shearing action, ensuring that the force on both ends of the shearing mechanism is completely balanced when it contacts the glass wool. This balanced shearing force effectively avoids the phenomena of tool tilting, jamming, or tearing of glass wool that may be caused by unilateral drive, thus ensuring a neat cut without burrs and improving the appearance quality and dimensional accuracy of the finished product.

[0093] In one possible implementation, the shearing mechanism further includes a gap adjustment assembly, a shearing assembly, and a transition piece 26. The gap adjustment assembly includes a limiting groove 18, a guide plate 19, a slide rod 20, and a connecting shaft 21; two limiting grooves 18, two guide plates 19, two slide rods 20, and two connecting shafts 21 are configured. The limiting grooves 18 are symmetrically opened in the housing 2. The slide rods 20 are slidably connected to both the limiting grooves 18 and the housing 2 at the end away from the roller C7. The ends of the slide rods 20 away from the limiting grooves 18 are rotatably connected to the connecting shafts 21. The guide plates 19 are slidably disposed in the limiting grooves 18. A roller D8 is rotatably connected between the guide plates 19. The roller D8 extends from the shaft end of the guide plate 19 and is coaxially fixed with a gear D10. The shearing assembly includes a positioning element 23, a moving blade 24, and a crankshaft 25. The positioning element 23 is fixed inside the housing 2 directly below the roller C7. The end of the positioning element 23 near the roller D8 is flush with the circumferential contact between the roller C7 and the glass wool. The moving blade 24 is slidably connected inside the housing 2 directly below the roller D8, with its cutting edge facing the positioning element 23. Two crankshafts 25 are symmetrically and rotatably mounted on the end of the moving blade 24 away from the positioning element 23. Both the connecting shaft 21 and the crankshafts 25 are rotatably connected to the circumferentially of the adapter 26, which is fixed to the output end of the corresponding telescopic component 22.

[0094] With the above scheme, when the detection unit identifies the defect and confirms that the defect starting point crosses the lower end of the second roller group, the central control unit sends a first shearing signal (a first defect signal and a delay signal) to the telescopic component 22. The telescopic component 22 extends, and the output end of the telescopic component 22 drives the adapter 26 to move synchronously. When the adapter 26 moves, it pulls the slide rod 20 through the circumferentially connected connecting shaft 21. The slide rod 20 drives the guide plate 19 to slide along the limiting groove 18, so that the roller D8 moves closer to the roller C7, reducing the roller gap of the second roller group and clamping the glass wool. At the same time, the adapter 26 drives the moving blade 24 to slide towards the positioning component 23 through the crankshaft 25. The cutting edge of the moving blade 24 cooperates with the positioning component 23 to shear the starting end of the defect segment. After completion, the telescopic component 22 resets, driving the roller D8 and the moving blade 24 to return to their initial positions. The central control unit continuously tracks the end point of the defective section. When it determines that the end point of the defective section has completely passed the lower exit of the second roller group, the central control unit sends a second shearing signal (a second defect signal and a delay signal) to the telescopic member 22. The telescopic member 22 extends again, and the output end of the telescopic member 22 drives the adapter 26 to move synchronously, controlling the telescopic member 22 to start again, repeating the above gap adjustment and shearing actions to cut off the end of the defective section and complete the complete isolation of the defective section. The time difference between the central control unit receiving the second defect signal and the first defect signal is considered. It needs to be greater than the delay signal. .

[0095] In the above solution, on the one hand, by linking the adapter 26, connecting shaft 21 and crankshaft 25, the telescopic component 22 can simultaneously complete the gap adjustment between roller D8 and roller C7 and the shearing action of the moving blade 24 with a single drive, without the need for an additional independent drive mechanism, which simplifies the equipment structure. At the same time, it ensures the precise synchronization of gap adjustment and shearing action, avoids incomplete shearing or glass wool stretching and deformation caused by asynchronous actions, and improves the reliability of defect section isolation.

[0096] On the other hand, the positioning component 23 is fixed directly below the roller C7, and its end near the roller D8 is flush with the contact circumference of the roller C7 and the glass wool, providing a stable shearing reference surface for the moving blade 24, ensuring accurate contact position between the shearing edge and the glass wool, and avoiding burrs or uneven breaks caused by shearing deviation; while the sliding fit design of the guide plate 19 and the limiting groove 18 makes the movement trajectory of the roller D8 stable and there is no shaking during the gap adjustment process, further ensuring the clamping stability of the glass wool during the shearing process, and solving the problem of inconsistent product quality caused by unstable operation in traditional manual shearing.

[0097] It is important to note that the positioning element 23 in this design serves only as a wear-resistant support reference surface (anvil) and does not require a cutting edge. Traditional fixed-blade + moving-blade structures require two precision-ground blades, with specific technical standards governing the material, hardness, and edge retention of each blade, resulting in high tool costs. In contrast, the structure of positioning element 23 with moving-blade 24, after long-term wear, only requires replacement or re-grinding of the low-cost moving-blade 24, eliminating the need for replacing or re-grinding both blades in pairs as in traditional double-blade structures. This significantly reduces consumable costs and maintenance workload.

[0098] Preferably, the housing 2 can be provided with a through hole, and a protective door is installed in the through hole. The protective door can effectively prevent external dust, impurities or foreign objects from entering the housing 2, which not only avoids interference or damage to the internal operating parts (such as the moving knife 24), but also ensures the cleanliness of the glass wool conveying process. Secondly, when it is necessary to replace or maintain the moving knife 24, the protective door can be opened conveniently, and the operation can be carried out directly through the through hole without disassembling the entire housing 2 or other components, which greatly simplifies the maintenance process.

[0099] In one possible implementation, a machine vision-based glass wool defect detection and sorting device further includes a third roller group. The third roller group is positioned below the second roller group within the housing 2, along the downward direction of the glass wool. The third roller group includes rollers E11 and F12, gear E13, and gear F14. Rollers E11 and F12 are horizontally positioned and rotatably connected to the housing 2. Roller E11 extends from the shaft end of the housing 2 and coaxially fixes gear E13. Roller F12 extends from the shaft end of the housing 2 and coaxially fixes gear F14. Gear E13 meshes with gear F14. Gear H16 is rotatably positioned on the housing 2 between gears C9 and E13, meshing with gears C9 and E13 respectively. The time difference between the glass wool passing the shearing component and the glass wool passing the lower end of the third roller group is less than the time difference in the delayed signal. .

[0100] Using the above technical solution, after the rotating drive component 17 drives the gear C9 to rotate, the gear C9 transmits power to the gear E13 of the third roller group through the gear H16. The gear E13 meshes with the gear F14 to drive the roller F12 to rotate. By setting the transmission ratio of gear C9 and gear E13 to 1:1, the third roller group and the first roller group maintain the same speed, forming a stable downward conveying reference.

[0101] Because the time difference between the glass wool passing the shear assembly and the glass wool passing the lower end of the third roller group is less than the time difference in the delayed signal. That is, the set distance between the second roller group and the shearing mechanism must be greater than the set distance between the shearing mechanism and the third roller group, so as to ensure that after the glass wool starting point is sheared, its lower end is always conveyed by the third roller group. At this time, when the gap of the second roller group is reduced due to the defect detection, the third roller group maintains the original speed and forms a speed difference with the second roller group, applying directional tensile stress to the defect section, assisting the moving knife 24 and the positioning part 23 to quickly complete the shearing action, while ensuring that the normal cotton section is smoothly transitioned to the downstream conveyor belt.

[0102] After shearing, the third roller group quickly sends the isolated defective section away at the original speed to avoid the defective section from staying and affecting the conveying rhythm of the subsequent normal cotton, thus achieving efficient separation and connection between the defective section and the normal cotton.

[0103] In the above scheme, the differential speed design between the third roller group and the second roller group generates a directional tensile force on the defective section during the shearing process, making the glass wool fibers easier to break at the shearing point, significantly shortening the shearing time, avoiding the residue of defective sections or damage to normal cotton caused by incomplete shearing, solving the problems of traditional manual shearing relying on experience and being inefficient, and improving the reliability and speed of online sorting.

[0104] The design of the third roller group with the same speed as the first roller group ensures that the normal cotton section can be seamlessly connected to the downstream process at a stable speed after the defective section is sheared and isolated. This avoids the accumulation or stretching deformation of normal cotton caused by the slowdown of the second roller group, maintains the continuous operation rhythm of the production line, makes up for the inefficiency of manual inspection which requires stopping the machine for spot checks, and meets the stringent requirements of high-speed production for conveying stability.

[0105] In one possible implementation, a machine vision-based glass wool defect detection and sorting device further includes an end effector. The end effector is located below the third roller group and includes an openable clamping plate 27 and a rotary drive 28. The openable clamping plate 27 is rotatably connected to the hopper 1, and the rotary drive 28 is fixed to the hopper 1 below the rotating end of the openable clamping plate 27. The output end of the rotary drive 28 is rotatably connected to the openable clamping plate 27. The central control unit is electrically connected to the rotary drive 28.

[0106] Based on the above scheme, under normal production conditions, the openable clamp 27 remains stationary to the side of the hopper 1, without interfering with the normal downward path of the glass wool. After the central control unit completes the end-point locking and end-point cutting of the defective section, it confirms that the defective section has completely fallen within the rotation range of the openable clamp 27, and that the subsequent normal cotton has left this area. The central control unit sends a command to the rotary drive 28 to drive the openable clamp 27 to rotate. Subsequently, the rotary drive 28 drives the openable clamp 27 to rotate to an inclined state, and the defective section moves out along the conveying path on the surface of the openable clamp 27, causing it to fall into the waste collection area.

[0107] After the defective segment is removed, the rotary drive 28 drives the openable clamping plate 27 to rotate in the opposite direction to reset and return to the standby state, waiting for the processing of the next defective segment, thus realizing continuous automated sorting cycle.

[0108] In addition, the clamping, rotating, and resetting actions of the end effector are only activated when the defective section passes through, without interfering with the continuous downward movement of normal cotton. Compared with the method in the background technology that requires manual stopping for sampling or manual cleaning of defects, online non-stop sorting is achieved, which greatly improves the operating efficiency and automation level of the production line.

[0109] In practice, after the glass wool is initially shaped by rollers A3 and B4, it enters the conveying process. The detection unit acquires images of the glass wool surface in real time. When the starting point of a defect is detected to have crossed the lower ends of rollers C7 and D8, a first defect signal is immediately generated and transmitted to the central control unit. After receiving the first defect signal, the central control unit calculates the signal by combining it with a delay signal, controls the extension member 22 to extend, and reduces the gap between rollers C7 and D8 to clamp the defect segment, allowing sufficient time for subsequent shearing. Simultaneously, the moving blade 24 moves towards the positioning member 23 to shear the starting end of the defect segment. After completion, the moving blade 24 resets, and rollers C7 and D8 reset. The detection unit continues to track the defect segment. When the defect segment completely passes the lower ends of rollers C7 and D8, a second defect signal is immediately generated and transmitted to the central control unit. After receiving the second defect signal, the central control unit calculates the signal by combining the second defect signal with a delay signal, and then restarts the shearing mechanism to repeat the clamping and shearing actions to cut off the end of the defective segment and completely isolate it.

[0110] Once the central control unit has completed locking and cutting off the end of the defective segment, the normal cotton crosses the rotation range of the openable clamp 27. The central control unit sends a command to the rotary drive 28 to drive the openable clamp 27 to rotate. Subsequently, the rotary drive 28 drives the openable clamp 27 to rotate to an inclined state, and the defective segment moves out along the conveying path on the surface of the openable clamp 27, falling into the waste collection area. After the defective segment is removed, the rotary drive 28 drives the openable clamp 27 to rotate in the opposite direction to reset and return to the standby state, waiting for the processing of the next defective segment.

[0111] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A machine vision-based defect detection and sorting device for glass wool, characterized in that, include: The conveying system, along the downward direction of the glass wool, includes a first roller group and a second roller group, wherein the first roller group moves at a speed of The second roller group is running at a speed of [speed value missing]. It also has a roller gap adjustment function; among which ; The shearing mechanism is located below the second roller group and is at a predetermined distance from the second roller group. ; The detection unit is located between the first roller group and the second roller group. It is used to acquire images of the glass wool surface in real time and identify defect areas. When the starting point of the defect is detected to cross the lower end of the second roller group, a first defect signal is immediately generated. When the ending point of the defect is detected to cross the lower end of the second roller group, a second defect signal is immediately generated. The central control unit, electrically connected to the shearing mechanism and the detection unit, receives a first defect signal and a second defect signal. Upon receiving the first defect signal, it combines it with a delayed signal to generate a first shearing signal, which is then sent to the shearing mechanism to shear the defect starting point. Upon receiving the second defect signal, it combines it with a delayed signal to generate a second shearing signal, which is then sent to the shearing mechanism to shear the defect ending point. The time difference between receiving the second defect signal and the first defect signal is specified. Greater than the delayed signal ; The detection unit includes a linear array camera and an image preprocessing module; The linear scan camera acquires real-time surface image streams of continuously falling glass wool, and the encoder synchronously obtains the real-time operating speed of the production line. and position pulse signal; The algorithm for determining the delayed signal is as follows: Based on formula Calculate the time difference between the first shear signal and the first defect signal received by the central control unit. Or the central control unit receives the time difference between the second shear signal and the second defect signal. ,in, The distance between the contact point between the second roller group and the glass wool and the contact point between the shearing mechanism and the glass wool is [missing information]. The correction value time for the shearing mechanism to receive the first shearing signal or the second shearing signal and then perform the action; The conveying system includes a hopper, a housing, and a rotating drive component; The cover is configured in multiple ways, and the cover is evenly fixed below the hopper. The rotating drive is fixed on the hopper. The cover includes the first roller group and the second roller group in sequence along the downward direction of the glass wool. The first roller group and the second roller group are connected by gear C; the output end of the rotation drive component is fixed to the same axis as the gear C; The second roller group includes roller C, roller D, gear C and gear D. Roller C and roller D are horizontally arranged. Roller C is rotatably connected to the inside of the cover. The shaft end of roller C extending out of the cover is coaxially fixed to gear C. The shaft end of roller D extending out of the cover is coaxially fixed to gear D. Gear C is meshed with gear D. The shearing mechanism includes a telescopic component; Two telescopic components are configured and symmetrically fixed inside the housing; The shearing mechanism also includes a gap adjustment component, a shearing component, and an adapter; The gap adjustment assembly includes a limiting groove, a guide plate, a slide rod, and a connecting shaft; two limiting grooves, two guide plates, two slide rods, and two connecting shafts are configured. The limiting grooves are symmetrically opened inside the housing. The slide rod is slidably connected to both the limiting groove and the housing end away from the roller C. The end of the slide rod away from the limiting groove is rotatably connected to the connecting shaft. The guide plate is slidably disposed in the limiting groove. The roller D is rotatably connected between the guide plates. The gear D is coaxially fixed at the shaft end of the roller D extending from the guide plate. The shearing assembly includes a positioning element, a moving blade, and a crankshaft. The positioning element is fixed inside the housing directly below the roller C. One end of the positioning element near the roller D is flush with the circumferential contact between the roller C and the glass wool. The moving blade is slidably connected inside the housing directly below the roller D. The cutting edge of the moving blade faces the positioning element. Two crankshafts are symmetrically and rotatably arranged at the end of the moving blade away from the positioning element. Both the connecting shaft and the crankshaft are rotatably connected to the circumferential direction of the adapter, and the adapter is fixed to the output end of the telescopic component on the corresponding side.

2. The machine vision-based glass wool defect detection and sorting device according to claim 1, characterized in that, The image preprocessing module is configured to perform grayscale correction and noise filtering on the acquired raw images, construct a standardized input image sequence, and extract defect feature vectors using a pre-trained lightweight convolutional neural network model, outputting the category label and pixel-level bounding box for each defect region. The central control unit is electrically connected to the line scan camera and the image preprocessing module, respectively.

3. The machine vision-based glass wool defect detection and sorting device according to claim 1, characterized in that, The detection unit also includes a high-frequency light source, which adopts a dual-sided symmetrical lighting or coaxial diffuse reflection lighting method, closely surrounding both sides of the line scan camera lens or integrated into the periphery of the lens. The high-frequency light source is triggered synchronously with the line scan camera's line frequency signal.

4. The machine vision-based glass wool defect detection and sorting device according to claim 1, characterized in that, The central control unit is electrically connected to the rotation drive component.

5. The machine vision-based glass wool defect detection and sorting device according to claim 1, characterized in that, The first roller group includes roller A, roller B, gear A and gear B. Roller A and roller B are horizontally arranged and rotatably connected inside the cover. The shaft end of roller A extending out of the cover is coaxially fixed to gear A, and the shaft end of roller B extending out of the cover is coaxially fixed to gear B. Gear A is meshed with gear B. A gear G is rotatably disposed on the cover between gear A and gear C, and gear G meshes with gear A and gear C respectively.

6. The machine vision-based glass wool defect detection and sorting device according to claim 1, characterized in that, The central control unit is electrically connected to the telescopic component.

7. The machine vision-based glass wool defect detection and sorting device according to claim 1, characterized in that, It also includes the third roller set; The third roller group is positioned below the second roller group inside the cover along the downward direction of the glass wool. The third roller group includes roller E, roller F, gear E and gear F. Roller E and roller F are horizontally arranged and rotatably connected inside the cover. The shaft end of roller E extending out of the cover is coaxially fixed to gear E. The shaft end of roller F extending out of the cover is coaxially fixed to gear F. Gear E meshes with gear F. A gear H is rotatably disposed on the cover between gear C and gear E, and gear H is respectively meshed with gear C and gear E; Specifically, the time difference between the glass wool passing the shearing component and the glass wool passing the lower end of the third roller group is less than the time difference in the delayed signal. .

8. The machine vision-based glass wool defect detection and sorting device according to claim 7, characterized in that, It also includes end effectors; The end effector is located below the third roller group and includes an openable clamping plate and a rotary drive component; the openable clamping plate is rotatably connected to the hopper, the rotary drive component is fixed to the hopper below the rotating end of the openable clamping plate, and the output end of the rotary drive component is rotatably connected to the openable clamping plate. The central control unit is electrically connected to the rotary drive component.

Images