Filter stick melting hole automatic detection system
The automatic filter rod cavity detection system uses a camera system and machine learning model to identify cavities in real time, solving the problem of cavities being difficult to detect in traditional detection methods and improving the quality control and efficiency of cigarette production.
Patent Information
- Application Number
- CN202423223419.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Utility models(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-26
- Publication Date
- 2026-01-23
- Estimated Expiration
- 2034-12-26
AI Technical Summary
Traditional testing methods struggle to detect melting defects in cigarette filter rods during production, leading to substandard products entering the cigarette production and packaging stages and affecting the final product quality.
An automatic filter rod cavity detection system is adopted, which utilizes advanced sensing technology and machine learning methods to capture images of the filter rod end face through a camera group. Combined with machine vision and a filter rod cavity detection model, the system can identify cavity defects in real time.
It enables accurate identification and timely detection of filter rod cavities, improving production efficiency, ensuring product quality, and reducing the inflow of defective products.
Smart Images
Figure CN223827584U_ABST
Abstract
Description
Technical Field
[0001] This utility model relates to the field of automatic detection technology, and in particular to an automatic detection system for filter rod cavities. Background Technology
[0002] In the production of cigarette filter rods, the formation of melt holes is usually closely related to the spraying of glycerol esters during the fiber bundle forming stage. Glycerol esters are a key substance in the forming process, and their main function is to improve the quality of fiber forming. However, during the spraying process, glycerol esters may form abnormal droplets due to backflow. Once these droplets mix into the fiber bundle, they may gradually lead to the formation of melt holes during the filter rod curing process. It is worth noting that melt holes usually do not appear immediately after the filter rod is produced, but rather gradually emerge during the 12 to 24-hour curing process.
[0003] Because fused voids may be confined to the interior of the filter rod and not extend throughout its length, traditional visual inspections struggle to detect these latent defects. Their presence can only be detected indirectly after the filter rod enters the slitting process in cigarette production. Therefore, the randomness and concealment of fused voids pose a significant challenge to quality control in filter rod production. Filter rods containing fused voids are considered substandard products, and if not detected promptly and allowed to enter the cigarette production and packaging stages, they can negatively impact the quality of the final product.
[0004] Therefore, in order to improve the quality control level of cigarette production, it is particularly necessary to develop an efficient and accurate automatic detection system for filter rod defects. This system can accurately identify filter rod defects when defects occur, avoid potential quality risks, improve production efficiency, and provide technical support for achieving high-standard cigarette production. Utility Model Content
[0005] To address the above shortcomings, this utility model provides an automatic filter rod cavity detection system, which can acquire real-time images of the filter rod end face of the cigarette after the slitting process through advanced sensing technology and machine learning methods, and automatically identify filter rod cavities using machine vision.
[0006] To achieve the above objectives, the present invention adopts the following technical solution: an automatic detection system for filter rod cavities, comprising a power supply module, a routing module, and a wireless module, characterized in that: it further comprises a U-shaped basket, a first camera group and a second camera group, a micro host, a cloud server, and a remote client;
[0007] The U-shaped basket is used to place several slit filter rods and is placed on a preset detection area;
[0008] The first camera group and the second camera group are respectively placed at both ends of the U-shaped basket. The first camera group and the second camera group can be aligned with the end faces of the cut filter rods to capture the end faces of the cut filter rods.
[0009] The micro-host is used to acquire real-time image data of the end face of the cut filter rod collected by the first and second camera groups, and transmit the real-time image to the cloud server through a wireless module.
[0010] The remote client is used to acquire and display real-time image data sent by the cloud server, and to perform real-time filter rod cavity detection using the filter rod cavity detection model.
[0011] Specifically, the U-shaped basket is equipped with a handle.
[0012] Specifically, the wireless module is connected to the mini-host via a Mini-PCIe interface; the first and second camera groups are connected to the LAN port of the router module via network cables, while the WAN interface of the router module is connected to the LAN_A1 interface of the mini-host.
[0013] Specifically, the first camera group consists of a first camera and a first ring light, and the second camera group consists of a second camera and a second ring light, with the camera located at the center of the ring light; the camera is an FXH-8028CP camera with a 4mm wide-angle lens and an image resolution of 1080P; the ring light provides illumination to the split filter rod, and the ring light has uniformly distributed light.
[0014] Specifically, the routing module uses a VAR1200-H industrial router, which has three adaptive gigabit WAN / LAN ports and an integrated cooling fan.
[0015] Specifically, the microcontroller is an ELSKY M700SE-UA microcontroller, operating at 12V or 19V, with a full-load power consumption of approximately 40 watts.
[0016] Specifically, the wireless module uses a CF-AX200 SE wireless network card, which supports the WIFI-6 protocol.
[0017] Specifically, the cloud server and the wireless module communicate using the TCP / IP communication protocol; the cloud server establishes a connection with the remote client via wired or wireless means.
[0018] Specifically, the filter rod cavity detection model is trained based on the YOLOv5s, DeepStop, and SlowFast algorithms.
[0019] Specifically, the power supply module uses a D-350 dual-output switching power supply. The D-350 supports dual voltage outputs of 5V and 12V, with a maximum output power of 350W. The power supply module supplies power to the micro host, the first camera group, the second camera group, and the routing module via a DC charging interface.
[0020] Compared with the prior art, the beneficial effects of this utility model are:
[0021] (1) Detection area: The system uses a U-shaped basket to load the cut filter rod, which improves the detection speed.
[0022] (2) Real-time video monitoring: The system is equipped with high-precision video monitoring function and ring light, which can reduce the contrast between the highlight and shadow areas of the filter rod cutting area and reduce overexposure. The system obtains real-time video images of the cigarette filter rod end face through the camera to ensure full visualization of key links and facilitate operators to keep track of production dynamics at any time.
[0023] (3) Precise molten hole detection: Utilizing existing advanced image processing and algorithm technologies, the system can accurately identify the location, size and number of molten holes through the filter rod molten hole detection model, ensuring that defective filter rods are detected in a timely manner and preventing defective products from entering subsequent production stages;
[0024] (4) Filter rod count: The system has a filter rod counting function, which can count the total number of filter rods and the number of defects in real time, providing a data basis for production efficiency analysis and quality management. Attached Figure Description
[0025] To more clearly illustrate the technical solutions of the embodiments of this utility model, the accompanying drawings used in the description of the embodiments will be briefly introduced below.
[0026] Figure 1 This is a structural diagram of an automatic detection system for filter rod cavities according to the present invention;
[0027] Figure 2 This is a diagram of the identification interface of an automatic detection system for filter rod cavities according to this utility model.
[0028] Figure label:
[0029] 1. U-shaped basket; 2-1. First camera; 2-2. First ring light; 3-1. Second camera; 3-2. Second ring light. Detailed Implementation
[0030] The technical solutions of the present utility model will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present utility model, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present utility model without creative effort are within the scope of protection of the present utility model.
[0031] In the description of this utility model, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the accompanying drawings and are only for the convenience of describing this utility model 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, and therefore should not be construed as a limitation of this utility model. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.
[0032] In the description of this utility model, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "joining" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this utility model based on the specific circumstances. Furthermore, the technical features involved in the different embodiments of this utility model described below can be combined with each other as long as they do not conflict with each other.
[0033] Example 1: As shown in the attached document Figure 1 As shown, this utility model provides an automatic detection system for filter rod cavities, including a power supply module, a routing module, a wireless module, a U-shaped basket (1), a first camera group and a second camera group, a micro host, a cloud server, and a remote client.
[0034] The U-shaped basket is used to place several slit filter rods. The U-shaped basket is also equipped with a handle to facilitate the transfer of the U-shaped basket (1). The U-shaped basket is placed on the preset detection area.
[0035] The camera group is detachably and fixedly mounted on the bracket. The first camera group and the second camera group are placed at both ends of the U-shaped basket (1) and at a certain distance from the U-shaped basket (1). The first camera group and the second camera group can be aligned with the end face of the cut filter rod to photograph the end face of the cut filter rod. The first camera group consists of a first camera (2-1) and a first ring light (2-2). The second camera group consists of a second camera (3-1) and a second ring light (3-2). The camera is located at the center of the ring light. The camera is an FXH-8028CP camera with a 4mm wide-angle lens and an image resolution of 1080P. The ring light provides supplementary lighting for the cut filter rod. The ring light has a relatively uniform light distribution, making the image captured by the camera clearer.
[0036] The first and second camera groups are connected to the LAN port of the router module via network cables, while the WAN port of the router module is connected to the LAN_A1 port of the micro-host.
[0037] The power supply module uses a D-350 dual-output switching power supply. The D-350 supports dual voltage output of 5V and 12V, with a maximum output power of 350W. The power supply module supplies power to the micro host, the first camera group, the second camera group, and the routing module through a DC charging interface. The power supply module also has overload, overcurrent, and short circuit protection functions to ensure the safety and stability of the system under extreme conditions.
[0038] The routing module uses the VAR1200-H industrial router, which has three adaptive gigabit WAN / LAN ports to ensure high-speed data transmission. The router also integrates a cooling fan and has overvoltage, overcurrent and reverse connection protection functions to ensure the long-term stable operation of the equipment.
[0039] The miniature host is used to acquire real-time image data of the end face of the slicing filter rods collected by the first and second camera groups, and transmits the real-time images to the cloud server via a wireless module. The miniature host is an ELSKY M700SE-UA, operating at 12V or 19V. It is equipped with an Intel Core i7-7660U chip, possessing dual-core 2.5GHz computing power, supporting real-time processing and transmission of multi-channel sensor data. The miniature host has a full-load power consumption of approximately 40W and can operate stably within a temperature range of -10℃ to 60℃ and relative humidity of 5% to 95% (non-condensing), remaining reliable even in environments with poor heat dissipation. The wireless module uses a CF-AX200 SE wireless network card, supporting the WIFI-6 protocol, which significantly reduces data transmission latency. The wireless module connects to the miniature host via a Mini-PCIe interface.
[0040] The remote client is used to acquire and display real-time video images sent by the cloud server, and performs real-time filter rod cavity detection using a filter rod cavity detection model. The cloud server and the wireless module communicate using the TCP / IP communication protocol; the cloud server establishes a connection with the remote client via wired or wireless means.
[0041] The filter rod cavity detection model was trained based on existing YOLOv5s, DeepStop and SlowFast algorithms.
[0042] The training method for the filter rod cavitation detection model is as follows:
[0043] Step 1: Preprocessing. The real-time image data of the end faces of the slicing filter rods acquired by the first and second camera groups are processed to ensure that the positions of the slicing filter rods in the acquired images correspond one-to-one, and information is marked for each slicing filter rod. The information marking includes slicing filter rod position information, slicing filter rod time information, etc.
[0044] Step 2: Frame extraction. The acquired images are processed by extracting frames within a specific time period. Target detection is then performed on each frame within that time period. When a target—a lava hole—is detected in the image, the lava hole is shown in the attached image. Figure 2 As shown, mark the lava hole and save the image.
[0045] Step 3: Target matching. Based on the preprocessing in Step 1, merge the molten hole marks on both ends of the cut filter rod together and remove duplicate molten hole marks.
[0046] Step 4: Cavity Count. Based on the number of cavities marked in Step 3, count the number of cut filter rods with cavities and the number of cut filter rods without cavities.
[0047] Repeat steps 2-4 until the end time of this test, at which point the test is complete.
[0048] The image segments are captured at 10-minute intervals, with each segment lasting 5 seconds, and the detection period is 24 hours. The parameters of the filter rod melt hole detection model can be adjusted according to actual needs.
[0049] In the specific implementation of this utility model: after the filter rod is cut, a certain number of cut filter rods are placed into the U-shaped basket (1), and the U-shaped basket (1) containing the cut filter rods is placed in the detection area. The first camera group and the second camera group are respectively placed at both ends of the U-shaped basket (2), and the camera groups are all aimed at the end face of the cut filter rods in the U-shaped basket (1).
[0050] The first and second camera groups record real-time images of the end faces of the slit filter rods inside the U-shaped basket. The captured images 1 and 2, respectively, are uploaded to the micro-host via a router. The micro-host then communicates the captured images to the cloud server using the TCP / IP protocol.
[0051] The remote client establishes a connection with the cloud server via wired or wireless means, and can receive and display the captured images sent by the cloud server in real time.
[0052] The remote client uses the filter rod cavity detection model to perform real-time filter rod cavity detection and outputs the detection results.
[0053] After the test is completed, the U-shaped basket is transferred to the next process, and the slit filter rods with molten holes are removed based on the test results.
[0054] Example 2: An improvement on Example 1, wherein the U-shaped basket is marked with a supplementary mark, which is located at the end of the U-shaped basket (1) away from the blue background.
[0055] A guide rail is set at a certain distance from and parallel to the U-shaped basket (1). A movable fixed platform is installed on the guide rail, and a mini robotic arm is fixedly installed on the platform.
[0056] The mini robotic arm is a six-axis robotic arm that can operate at multiple angles, and a three-jaw actuator is installed at the end of the robotic arm.
[0057] The guide rail has a rejection basket and a qualified basket at each end. The rejection basket is used to load the slitting filter rod with molten holes, and the qualified basket is used to load the slitting filter rod without molten holes.
[0058] The training method for the filter rod cavity detection model also includes the following steps:
[0059] Step 1: Preprocessing. The real-time image data of the end faces of the slicing filter rods acquired by the first and second camera groups are processed to ensure that the positions of the slicing filter rods in the acquired images correspond one-to-one, and information is marked for each slicing filter rod. The information marking includes slicing filter rod position information, slicing filter rod time information, etc.
[0060] Step 2: Frame extraction. The acquired image is processed by extracting frames, capturing image segments within a specific time period. Then, target detection is performed on each frame within that time period. When a target appears in the image as shown in the attached image... Figure 2 When a lava hole is shown, mark the lava hole and save the image.
[0061] Step 3: Cavity Rejection. Based on the cavity markings from Step 2, issue a rejection command to the mini robotic arm and place the rejected slit filter rods into the rejection basket. If no cavity has formed on a slit filter rod by the end of the inspection time, issue a pass command to the mini robotic arm and place the pass slit filter rods into the pass basket.
[0062] Step 4: Cavity Count. Based on the rejection and acceptance commands in Step 3, count the number of cut filter bars with cavities and the number of cut filter bars without cavities.
[0063] Step 5: Replenish the slitting filter rods. When the slitting filter rods in the U-shaped basket descend below the replenishment mark, a replenishment instruction is issued to the staff to replenish the slitting filter rods in the U-shaped basket.
[0064] The technical effects of this embodiment are as follows:
[0065] A sophisticated mini robotic arm is mounted on one side of the U-shaped basket, allowing it to quickly process any cavities that form during the filter rod slitting inspection process, without waiting for the entire inspection process to finish. This instant response mechanism significantly reduces the inspection time, thereby greatly improving the efficiency of the inspection work.
[0066] The above description is merely a specific embodiment of this utility model, but the protection scope of this utility model is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this utility model should be included within the protection scope of this utility model. Therefore, the protection scope of this utility model should be determined by the protection scope of the claims.
Claims
1. An automatic detection system for filter rod cavities, comprising a power supply module, a routing module, and a wireless module, characterized in that: It also includes a U-shaped basket, a first camera group and a second camera group, a micro-host, a cloud server, and a remote client; The U-shaped basket is used to place several slit filter rods and is placed on a preset detection area; The first camera group and the second camera group are respectively placed at both ends of the U-shaped basket. The first camera group and the second camera group can be aligned with the end faces of the cut filter rods to capture the end faces of the cut filter rods. The micro-host is used to acquire real-time image data of the end face of the cut filter rod collected by the first and second camera groups, and transmit the real-time image to the cloud server through a wireless module. The remote client is used to acquire and display real-time image data sent by the cloud server, and to perform real-time filter rod cavity detection using the filter rod cavity detection model.
2. The automatic detection system for filter rod cavities according to claim 1, characterized in that: The U-shaped basket is equipped with a handle.
3. The automatic detection system for filter rod cavities according to claim 1, characterized in that: The wireless module is connected to the mini-host via a Mini-PCIe interface; the first and second camera groups are connected to the LAN port of the router module via network cables, while the WAN interface of the router module is connected to the LAN_A1 interface of the mini-host.
4. The automatic detection system for filter rod cavities according to claim 1, characterized in that: The first camera group consists of a first camera and a first ring light, and the second camera group consists of a second camera and a second ring light. The camera is located at the center of the ring light. The camera is an FXH-8028CP camera with a 4mm wide-angle lens and an image resolution of 1080P. The ring light provides illumination to the split filter rod and has uniformly distributed light.
5. The automatic detection system for filter rod cavities according to claim 1, characterized in that: The routing module uses a VAR1200-H industrial router, which has three adaptive gigabit WAN / LAN ports and an integrated cooling fan.
6. The automatic detection system for filter rod cavities according to claim 1, characterized in that: The microcomputer is an ELSKY M700SE-UA microcomputer, with an operating voltage of 12V or 19V and a full-load power consumption of approximately 40 watts.
7. The automatic detection system for filter rod cavities according to claim 1, characterized in that: The wireless module uses a CF-AX200 SE wireless network card, which supports the WIFI-6 protocol.
8. The automatic detection system for filter rod cavities according to claim 1, characterized in that: The cloud server and the wireless module communicate using the TCP / IP communication protocol; the cloud server establishes a connection with the remote client via wired or wireless means.
9. The automatic detection system for filter rod cavities according to claim 1, characterized in that: The filter rod cavity detection model was trained based on the YOLOv5s, DeepStop, and SlowFast algorithms.
10. The automatic detection system for filter rod cavities according to claim 1, characterized in that: The power supply module uses a D-350 dual-output switching power supply. The D-350 supports dual voltage output of 5V and 12V, with a maximum output power of 350W. The power supply module supplies power to the micro host, the first camera group, the second camera group, and the routing module through a DC charging interface.