Foreign Fiber Removal System

Abstract

The present invention provides a foreign fiber removal system, which belongs to the field of textile machinery, and specifically includes a loosening device; a fiber conveying device, which conveys natural fibers that are evenly grasped, wherein the vertical height of the first conveying mechanism is higher than the vertical height of the second conveying mechanism, and the flipping mechanism flips at least part of the natural fibers conveyed by the first conveying mechanism; a foreign fiber detection device; a foreign fiber removal device, wherein the foreign fiber detection device includes: a shooting mechanism, which is used to shoot the natural fibers conveyed by the fiber conveying device; a computing mechanism, which is used to receive the photographed photos and perform computing analysis to generate the current foreign fiber coordinates corresponding to the foreign fiber removal, and the computing mechanism uses at least two deep learning algorithms to train the foreign fiber recognition model; a removal control unit, which is used to control the foreign fiber removal device to remove foreign fibers according to the current foreign fiber coordinates. Through the processing scheme of the present application, small foreign fibers in natural fibers can be quickly identified and removed.

Description

Technical Field

[0001] The invention relates to the field of textile machinery, and in particular to a foreign fiber removing system. Background Art

[0002] In current actual production, there are devices for removing foreign fibers in the cotton field, called foreign fiber sorting machines. However, there are currently no devices or technologies that can achieve similar functions in the wool, cashmere, and linen industries.

[0003] At present, the foreign fiber detection machine used in the cotton spinning industry adopts the photoelectric detection method. The photoelectric foreign fiber detection machine passes the loosened cotton flow through the detection area equipped with a CCD camera, and the two sides of the cotton are illuminated by the light source and photographed by the CCD camera; the light shining on the cotton is strong, and the cotton fiber is white in the image. If there are foreign fibers or other colored objects (such as dirty cotton, etc.) in the cotton, the central processing unit will calculate the deviation of its color from the standard cotton color. If the color deviation exceeds a certain range, the central processing unit will send a clearing signal to the cleaning mechanism to execute the removal of foreign fibers.

[0004] However, the main shortcomings of traditional foreign fiber sorting machines are as follows:

[0005] 1. The detection effect of white or transparent foreign fibers is poor. Since the recognition principle of traditional foreign fiber sorting machines is to compare the contrast between foreign fibers and cotton colors, the recognition rate of traditional foreign fiber sorting machines for white and transparent foreign fibers (such as transparent ground film, white polypropylene yarn and other common foreign fibers) is very low;

[0006] 2. The recognition rate of tiny foreign fibers is low. For example, the recognition accuracy of hair by traditional foreign fiber machines is low. Summary of the invention

[0007] Therefore, in order to overcome the above-mentioned shortcomings of the prior art, the present invention provides a foreign fiber removal system for quickly identifying and removing fine foreign fibers in natural fibers.

[0008] In order to achieve the above-mentioned purpose, the present invention provides a system for removing foreign fibers from natural fibers, comprising: a loosening device for moderately loosening the natural fibers and making the natural fibers fall evenly; a fiber conveying device for conveying the natural fibers that have been dropped and evenly spread, wherein the fiber conveying device is provided with a plurality of positioning marks, and has at least two conveying mechanisms connected by a flipping mechanism, wherein the vertical height of the first conveying mechanism is higher than the vertical height of the second conveying mechanism, and the flipping mechanism flips at least part of the natural fibers conveyed by the first conveying mechanism; and a foreign fiber detection device for detecting the natural fibers conveyed by the fiber conveying device and identifying foreign fibers in the natural fibers. foreign fibers and generates current foreign fiber coordinates corresponding to the removal of foreign fibers; a foreign fiber removal device removes the foreign fibers according to the current foreign fiber coordinates corresponding to the removal of the foreign fibers identified by the foreign fiber detection device, wherein the foreign fiber detection device includes: a shooting mechanism, arranged above the fiber conveying device, for photographing the natural fibers conveyed by the fiber conveying device; a computing mechanism, for receiving the photographed photos, performing computing analysis, generating current foreign fiber coordinates corresponding to the foreign fibers, and calculating and generating the removal time and the current foreign fiber coordinates corresponding to the removal of the foreign fibers, the computing mechanism uses at least two deep learning algorithms selected from the group consisting of the YOLO neural network algorithm, the R-CNN neural network algorithm, the Fast R-CNN neural network algorithm, the Faster R-CNN neural network algorithm, the RetinaNet neural network algorithm, the EfficientDet neural network algorithm, the ATSS neural network algorithm, and the SSD neural network algorithm to train a foreign fiber recognition model; a removal control unit, for controlling the foreign fiber removal device to remove the foreign fibers according to the current foreign fiber coordinates.

[0009] In one embodiment, the computing mechanism includes a model recognition unit for identifying foreign fibers and a coordinate calculation unit for determining coordinates of the foreign fibers.

[0010] In one of the embodiments, the model recognition unit uses at least two deep learning algorithms among the YOLO neural network algorithm, the R-CNN neural network algorithm, the Fast R-CNN neural network algorithm, the Faster R-CNN neural network algorithm, the RetinaNet neural network algorithm, the EfficientDet neural network algorithm, the ATSS neural network algorithm, and the SSD neural network algorithm to train sample images of foreign fibers to obtain different initial recognition models; the model recognition unit uses different initial recognition models to recognize test images of foreign fibers, and corrects the current initial recognition model according to the recognition result of another initial recognition model.

[0011] In one of the embodiments, the coordinate calculation unit receives the historical foreign fiber coordinates corresponding to the foreign fiber represented by the positioning mark of the model recognition unit, and obtains the conveying speed of the fiber conveying device. The coordinate calculation unit generates the execution clearing time corresponding to the foreign fiber and the current foreign fiber coordinates corresponding to the foreign fiber clearing according to the historical foreign fiber coordinates and the conveying speed.

[0012] In one embodiment, the flipping mechanism is arranged between the tail end of the first conveying mechanism and the head end of the second conveying mechanism, and includes a motor, a roller body connected to the motor output shaft, and a plurality of flipping paddles arranged on the roller body.

[0013] In one embodiment, the height difference between the vertical height of the first conveying mechanism and the vertical height of the second conveying mechanism is in the range of 40-60 cm.

[0014] In one of the embodiments, the foreign fiber removal device includes a linear module and a gripper module, the linear module is used to control the movement of the gripper module, and the gripper module includes a cylinder and a multi-link gripping mechanism that releases and closes multiple claws through the cylinder.

[0015] In one embodiment, the multi-link gripping mechanism includes an L-shaped gripper connecting seat, a gripper drive assembly and a gripper base, the vertical end face of the gripper connecting seat is connected to the linear module, and the gripper connecting seat is provided with a triangular fixed shaft for fixing the gripper base at the L angle; the gripper drive assembly is arranged on the horizontal end face of the gripper connecting seat and passes through the horizontal end face, including a gripper drive motor with an output shaft downwardly passing through the gripper connecting seat and a rocker arm support seat connected to the output shaft; the gripper base is fixedly connected to the triangular fixed shaft by bolts, and the gripper base includes a rotating shaft, a connecting arm and a rocker arm gripper, the rotating shaft is provided with the rocker arm support seat and is driven to rotate, the connecting arm is fixed to the rotating shaft, and the rocker arm gripper is fixed to the connecting arm.

[0016] In one embodiment, the foreign fiber removal device is a signal-controlled jet mechanism for spraying foreign fibers in the fiber at a specific time and sending them into a foreign fiber collection port. The jet mechanism includes a solenoid valve controlled by the removal control unit and a jet tank electrically connected to the solenoid valve.

[0017] Compared with the prior art, the advantages of the present invention are: using more than two deep learning models to identify foreign fibers can effectively improve the recognition rate of foreign fibers such as transparent films and white polypropylene fibers, and can also continuously optimize the recognition model by comparing with each other to improve the recognition accuracy; through the modular two-axis mechanical grasping mechanism for removal, this method can effectively reduce the cotton drop rate and reduce the normal raw materials that are removed at the same time as the foreign fibers are removed; by combining a double-layer conveying and flipping mechanism, the cotton / wool raw materials can be flipped and re-identified, which can effectively increase the exposure probability of foreign fibers. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0019] Figure 1 Schematic diagram of the overall structure of the foreign fiber removal system in an embodiment of the present invention;

[0020] Figure 2 is a flow chart of model training in an embodiment of the present invention;

[0021] Figure 3 is a schematic structural diagram of the foreign fiber removal device after the outer shell is removed in an embodiment of the present invention;

[0022] Figure 4 Schematic diagram of the structure of a multi-link grabbing mechanism in an embodiment of the present invention. DETAILED DESCRIPTION

[0023] The embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0024] The following describes the implementation methods of the present application through specific examples, and those skilled in the art can easily understand other advantages and effects of the present application from the contents disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The present application can also be implemented or applied through other different specific implementation methods, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present application. It should be noted that, in the absence of conflict, the following embodiments and the features in the embodiments can be combined with each other. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work belong to the scope of protection of the present application.

[0025] It should be noted that various aspects of the embodiments within the scope of the appended claims are described below. It should be apparent that the aspects described herein can be embodied in a wide variety of forms, and any specific structure and / or function described herein is merely illustrative. Based on the present application, it should be understood by those skilled in the art that an aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number and aspect described herein can be used to implement the device and / or practice the method. In addition, other structures and / or functionalities other than one or more of the aspects described herein can be used to implement this device and / or practice this method.

[0026] It should also be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present application. The drawings only show components related to the present application rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0027] Additionally, in the following description, specific details are provided to facilitate a thorough understanding of the examples. However, it will be understood by those skilled in the art that the aspects described may be practiced without these specific details.

[0028] like Figure 1 As shown, an embodiment of the present application provides a foreign fiber removal system 100 in natural fibers, including an opening device 10 , a fiber conveying device 20 , a foreign fiber detection device 30 and a foreign fiber removal device 40 .

[0029] The opening device 10 is used to moderately open the natural fibers and allow the natural fibers to fall evenly to the fiber conveying device 20. Moderate opening means that the height of the fiber layer is 6-10 cm when the natural fibers are naturally spread out. The opening device 10 includes a cotton storage basket and two baffles. The cotton storage basket is provided with at least one gripper and a gripper driving member corresponding to the gripper, the gripper is connected to the corresponding gripper driving member, and the gripper driving member is used to drive the gripper to move back and forth in the cotton storage basket; the bottom opening of the cotton storage basket is arranged and faces the conveyor belt 211 on the fiber conveying device 20, and the two baffles are both located at the bottom of the cotton storage basket and can block the bottom opening of the cotton storage basket; the two baffles are located on the same horizontal plane, and the opposite side panels of the two baffles are parallel to each other and can abut against each other; each baffle can move back and forth horizontally relative to the other baffle, and the moving directions of the two baffles are opposite. The opening device in the present application only needs to grab the natural fibers and drop the natural fibers directly and evenly on the conveyor belt, and does not need to break the natural fibers. The natural fibers can be natural cotton, wool / cashmere, etc.

[0030] The fiber conveying device 20 is used to convey the natural fibers that have been dropped and evenly spread. The fiber conveying device is provided with a number of positioning marks. The fiber conveying device 20 has at least two sets of conveying mechanisms 21 connected by a flipping mechanism. The vertical height of the first conveying mechanism is higher than the vertical height of the second conveying mechanism. The flipping mechanism flips at least part of the natural fibers conveyed by the first conveying mechanism. When there are multiple sets of conveying mechanisms, the vertical height of the conveying mechanism at the front end is higher than the vertical height of the conveying mechanism at the rear end, and a flipping mechanism is provided between two different conveying mechanisms. The flipping mechanism 22 can flip all the natural fibers conveyed by the conveying mechanism. For example, the flipping mechanism 22 can be a shovel-shaped structure to shovel the natural fibers on the high conveyor belt onto the low conveyor belt.

[0031] Each set of conveying mechanism 21 includes a conveyor belt 211 and a conveying motor. A number of positioning marks are arranged on the conveyor belt 211. The number of positioning marks can be 2*N, wherein N positioning marks can be arranged at each corner of the upper surface of the conveyor belt 211, and the remaining N positioning marks can be arranged at the center of each side of the conveyor belt 211 which is not easily blocked by natural fibers, and N can be not less than 2. Such a design can not only make the positional relationship between the center coordinates of each positioning mark and the foreign fiber coordinates more accurate, so as to improve the accuracy of the coordinate system fitting unification, but also can use the foreign fiber detection device 30 to use wide-angle shooting to perform wide-angle identification and positioning of the measured part, so as to further improve the positioning accuracy.

[0032] The positioning mark of this embodiment can be a sticker, which includes a center area located in the middle of the sticker and a peripheral area located around the center area. In order to improve the quality of the point cloud data, the color of the center area is preferably white, and the color of the peripheral area is preferably black, that is, the black peripheral area is used as the background, and such a background will not be scanned into the point cloud data, reducing the recognition error of the photo; the acquisition position of the white center area as the coordinate point can be clearly scanned into the point cloud data to achieve effective recognition.

[0033] The foreign fiber detection device 30 is used to detect the natural fibers transported by the fiber transport device, identify the foreign fibers in the natural fibers and generate the execution time and the current foreign fiber coordinates corresponding to the foreign fiber removal. The foreign fiber detection device includes a shooting mechanism 31, a computing mechanism and a removal control unit.

[0034] The photographing mechanism 31 is arranged above the conveying mechanism and at the rear end of the opening device 10, and is used to photograph the natural fiber conveyed by the fiber conveying device. The photographing mechanism includes a camera and a lighting lamp for supplementary light for photographing. The number of sets of the photographing mechanism 31 corresponds to the number of the conveying mechanism, and all are arranged at the front end of the conveyor belt of the conveying mechanism.

[0035] The operation mechanism is used to receive the photographed photos, perform operation analysis, generate the current foreign fiber coordinates corresponding to the foreign fibers, and calculate and generate the execution time and the current foreign fiber coordinates corresponding to the foreign fiber removal. The operation mechanism uses at least two deep learning algorithms from the YOLO neural network algorithm, R-CNN neural network algorithm, Fast R-CNN neural network algorithm, Faster R-CNN neural network algorithm, RetinaNet neural network algorithm, EfficientDet neural network algorithm, ATSS neural network algorithm, and SSD neural network algorithm to train the foreign fiber recognition model. The operation mechanism is a chip that can run the model. The chip can be a GPU, CPU, etc. Foreign fibers are visually different from natural fibers, or in color, or in shape, or in thickness. The machine can identify foreign fibers in natural fibers through a large amount of data learning.

[0036] The cleaning control unit is used to control the foreign fiber cleaning device to clean the foreign fibers according to the cleaning time and the current foreign fiber coordinates corresponding to the cleaning. The cleaning control unit can be a PLC, a separate control chip, a single chip microcomputer, etc., or it can share the same computing chip with the computing mechanism.

[0037] The foreign fiber removing device 40 removes the foreign fibers according to the current foreign fiber coordinates corresponding to the foreign fiber removal identified by the foreign fiber detecting device.

[0038] The above-mentioned foreign fiber removal system adopts more than two deep learning models to identify foreign fibers, which can effectively improve the recognition rate of foreign fibers such as transparent films and white polypropylene fibers, and can also continuously optimize the recognition model by comparing with each other to improve the recognition accuracy; it is removed by a modular two-axis mechanical grasping mechanism, which can effectively reduce the cotton drop rate and reduce the normal raw materials that are removed at the same time as the foreign fibers are removed; by combining a double-layer conveying and flipping mechanism, the cotton / wool raw materials can be flipped and re-identified, which can effectively increase the exposure probability of foreign fibers.

[0039] In one embodiment, the computing mechanism includes a model recognition unit for identifying foreign fibers and a coordinate calculation unit for determining coordinates of foreign fibers.

[0040] In one of the embodiments, the model recognition unit uses at least two deep learning algorithms among the YOLO neural network algorithm, the R-CNN neural network algorithm, the Fast R-CNN neural network algorithm, the Faster R-CNN neural network algorithm, the RetinaNet neural network algorithm, the EfficientDet neural network algorithm, the ATSS neural network algorithm, and the SSD neural network algorithm to train sample images of foreign fibers to obtain different initial recognition models; the model recognition unit uses different initial recognition models to recognize test images of foreign fibers, and corrects the current initial recognition model according to the recognition result of another initial recognition model.

[0041] like Figure 2 As shown, the model recognition unit uses at least two deep learning algorithms among the YOLO neural network algorithm, the R-CNN neural network algorithm, the Fast R-CNN neural network algorithm, the Faster R-CNN neural network algorithm, the RetinaNet neural network algorithm, the EfficientDet neural network algorithm, the ATSS neural network algorithm, and the SSD neural network algorithm to train the sample pictures of foreign fibers respectively to obtain different initial recognition models. The model recognition unit can learn and train the sample photos in advance, and the sample photos are photos with foreign fiber annotations. The positions and types of foreign fibers in different sample photos may be different. Before model training, the model recognition unit divides the sample photos into a training set and a test set. The model can not only identify foreign fibers, but also identify various types of targets such as dirty cotton, cotton stalks, colored wool, colored velvet, grass stalks, branches, etc. In actual use, the corresponding sample pictures can be selected for learning and training according to the different cleaning needs of different types of customers for subsequent selective cleaning.

[0042] In step 101, when the model recognition unit adopts the YOLO network, the model recognition unit can perform cluster analysis on the training set to determine the target anchor value (anchor point value), select 12 initial cluster centers of the training set according to the principle that the initial cluster center is as far as possible, and then use the k-mean++ algorithm to perform cluster analysis until the final cluster center is obtained, and measure the IOU and the measurement value d for this cluster analysis. The standard is: d(box, centroid) = 1-IOU, where IOU is the intersection and union ratio of the predicted box and the true box, and the average IOU is used as the measure of this cluster analysis, so as to obtain 12 better Anchor values; improve the main anchor value in the YOLO network model The trunk feature extraction network Darknet-53 and the multi-scale feature fusion network FPN are used to generate the optimized YOLO target detection model. The SPP spatial pyramid network structure is added to the tail of the Darknet-53 trunk extraction network of the YOLO model to solve the problems of partial information loss and size of the image. The network structure includes four parallel pooling layers, and the pooling is 1×1, 5×5, 9×9, and 13×13 respectively. A 104×104 feature scale is added to the multi-scale feature fusion network FPN to increase the detection accuracy of small objects. The training set after clustering analysis is input into the optimized YOLO network model for training and generating the model, and the model is tested by the test set.

[0043] When the model recognition unit uses neural networks such as the R-CNN neural network algorithm, the Fast R-CNN neural network algorithm, the Faster R-CNN neural network algorithm, the RetinaNet neural network algorithm, the EfficientDet neural network algorithm, the ATSS neural network algorithm, and the SSD neural network algorithm, the model recognition unit can use sample photos of the training set to train the detection network until overfitting occurs; then perform prediction operations on the training set, and use data enhancement methods to obtain more unlabeled foreign fibers, merge the predicted unlabeled foreign fiber detection frames with the original detection frames, and generate new training data for the next stage of detection network training; perform iterative training using the above method until it is found that the recall rate no longer increases, and stop training. The training sets used by different models can be the same or different.

[0044] In step 102, after the model recognition unit has trained at least two models, the model recognition unit uses sample images in the same test set to test different models, and performs model correction on its own model according to the test results output by another model, until the result deviation of the two models does not exceed the preset range value. The preset range value can be determined according to the degree to which the foreign fibers in the natural fibers need to be cleaned and the time when the foreign fibers can be identified. The higher the degree to which the foreign fibers in the natural fibers need to be cleaned, and / or the longer the time when the foreign fibers can be identified, the higher the preset range value; conversely, the smaller the preset range value.

[0045] In one of the embodiments, the coordinate calculation unit receives the historical foreign fiber coordinates corresponding to the foreign fibers represented by the positioning mark used by the model recognition unit, and obtains the conveying speed of the fiber conveying device. The coordinate calculation unit generates the execution clearing time corresponding to the foreign fibers and the current foreign fiber coordinates corresponding to the clearing based on the historical foreign fiber coordinates and the conveying speed.

[0046] The model recognition unit uses the horizontal plane of the conveyor belt as the coordinate system, and the coordinate system uses the positioning mark as the origin. When there are multiple positioning marks, there are differences between different positioning marks, so the same conveyor belt can have multiple coordinate systems. The model recognition unit can analyze the photographs taken to determine the positioning mark closest to the foreign fiber, and determine the historical foreign fiber coordinates of the foreign fiber according to the coordinate system corresponding to the positioning mark. The coordinate calculation unit generates the execution removal time corresponding to the foreign fiber and the current foreign fiber coordinates corresponding to the foreign fiber removal according to the historical foreign fiber coordinates, conveying speed, and analysis and calculation time.

[0047] In one embodiment, the flipping mechanism 22 is disposed between the tail end of the first conveying mechanism and the head end of the second conveying mechanism, and includes a motor, a roller body connected to the motor output shaft, and a plurality of flipping paddles disposed on the roller body.

[0048] The flipping paddle pulls and flips the natural fibers being transported, and can flip the wrapped foreign fibers out of the natural fibers so that the natural fibers can be detected and removed again later. The flipping paddle can be replaced according to different natural fibers to increase the detection rate of foreign fibers.

[0049] In one embodiment, the height difference between the vertical height of the first conveying mechanism and the vertical height of the second conveying mechanism is in the range of 40-60 cm. This height difference range can facilitate the turning mechanism to pull and turn the conveyed natural fibers, thereby improving the efficiency of removing foreign fibers.

[0050] In one embodiment, the foreign fiber removal device includes a linear module and a gripper module, the linear module is used to control the movement of the gripper module, and the gripper module includes a cylinder and a multi-link gripping mechanism that loosens and closes multiple claws through the cylinder.

[0051] like Figure 3 As shown, the foreign fiber removal device 40 includes a linear module 41 and a gripper module 42 with modular settings. The linear module 41 is horizontally arranged above the conveyor belt, and has a horizontal slide, a slider that can move on the horizontal slide, a horizontal drive assembly that drives the slider to move horizontally, a vertical screw connected to the slider, and a screw motor. The horizontal drive assembly can include a drive motor, a drive chain, etc. The vertical screw can drive the gripper module to move vertically. The gripper module 42 is fixedly connected to the slider. The gripper module 42 includes a cylinder and a multi-link gripping mechanism that releases and closes multiple claws through the cylinder. Both the linear module 41 and the gripper module 42 adopt a modular design, which is convenient for loading and unloading and increasing and decreasing the number, and the synchronous gripping of two axes improves the gripping efficiency. The multi-link gripping mechanism with multiple grips (three claws and four claws) improves the gripping accuracy and reduces the cotton waste rate.

[0052] like Figure 4 As shown, in one of the embodiments, the multi-link gripping mechanism includes an L-shaped gripper connecting seat, a gripper drive assembly and a gripper base. The vertical end face of the gripper connecting seat is connected to the slider of the linear module, and the gripper connecting seat is provided with a triangular fixed shaft for fixing the gripper base at the L angle; the gripper drive assembly is arranged on the horizontal end face of the gripper connecting seat and passes through the horizontal end face, including a gripper drive motor with an output shaft downwardly passing through the gripper connecting seat and a rocker arm support seat connected to the output shaft; the gripper base is fixedly connected to the triangular fixed shaft by bolts, and the gripper base includes a rotating shaft, a connecting arm and a rocker arm gripper. The rotating shaft is provided with a rocker arm support seat and is driven to rotate, the connecting arm is fixed to the rotating shaft, and the rocker arm gripper is fixed to the connecting arm.

[0053] The foreign fiber removal process of the foreign fiber removal device 40 of this embodiment is as follows:

[0054] The cotton inlet is connected to an external device (such as a cotton plucker, etc.), and natural fibers such as cotton or wool are sucked into the cotton box. After the cotton box is opened, the natural fibers fall onto the conveyor belt 211. The conveyor belt 211 conveys natural fibers such as cotton / wool at a certain speed. The shooting mechanism 31 performs machine vision detection on the cotton / wool natural fibers on the conveyor belt and transmits them to the computing mechanism. The AI ​​model carried by the computing mechanism calculates the position of the foreign fibers and sends it to the removal control unit. The removal control unit controls the movement of the linear module 41 in the foreign fiber removal device 40, and controls the multi-link gripping mechanism in the gripper module 42 to grab the foreign fibers at the designated position at the designated time and throw them into the foreign fiber collection port. The cotton / wool natural fibers continue to pass through the flipping mechanism 22 to flip and scatter the cotton / wool natural fibers again, and the turned cotton / wool natural fibers are transported through the conveyor belt 211. The foreign fibers are identified again by the shooting mechanism, and the foreign fibers are removed by the second set of gripper modules 42. The cotton / wool natural fibers that have been detected and removed twice for foreign fibers fall into the suction system and enter the subsequent process. If necessary, they can also enter the circulation system for the second round of foreign fiber removal.

[0055] In one embodiment, the foreign fiber removal device is a signal-controlled jet mechanism for spraying foreign fibers in the fiber at a specific time and sending them into a foreign fiber collection port. The jet mechanism includes a solenoid valve controlled by a removal control unit and a jet tank electrically connected to the solenoid valve.

[0056] The cleaning control unit controls the connection of the electromagnetic valve and can also control the direction of the nozzle of the jet tank to ensure that the foreign fibers are removed. The foreign fiber removal process of the foreign fiber removal device 40 of this embodiment is as follows:

[0057] The cotton inlet is connected to an external device (such as a cotton plucker, etc.), and natural fibers such as cotton or wool are sucked into the cotton box. After the cotton box is opened, the natural fibers fall onto the conveyor belt 211. The conveyor belt 211 conveys natural fibers such as cotton / wool at a certain speed. The shooting mechanism 31 performs machine vision detection on the cotton / wool natural fibers on the conveyor belt and transmits it to the computing mechanism. The AI ​​model carried by the computing mechanism calculates the position of foreign fibers and sends it to the removal control unit. The removal control unit controls one or more solenoid valves at the corresponding position in the foreign fiber removal device 40 to spray at a specified time, spraying foreign fibers into the foreign fiber collection port, and the cotton / wool natural fibers continue to pass through the flipping mechanism 22 to flip and break up the cotton / wool natural fibers again, and the cotton / wool natural fibers after flipping are transported through the conveyor belt 211, and foreign fibers are identified again by the shooting mechanism, and the foreign fibers are removed by the second set of foreign fiber removal devices 40. The cotton / wool natural fibers that have been detected and removed twice fall into the suction system and enter the subsequent process. If necessary, they can also enter the circulation system for the second round of foreign fiber removal.

[0058] The above are only specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A system for removing foreign matter from natural fibers, characterized in that: include: The opening device is used to moderately open the natural fibers and make the natural fibers fall evenly. Moderate opening means that the height of the fiber layer is 6-10cm when the natural fibers are naturally spread out; A fiber conveying device for conveying natural fibers that have been dropped and evenly spread, the fiber conveying device being provided with a plurality of positioning marks, having at least two conveying mechanisms connected by a flipping mechanism, the vertical height of the first conveying mechanism being higher than the vertical height of the second conveying mechanism, and the flipping mechanism flipping at least part of the natural fibers conveyed by the first conveying mechanism; A foreign fiber detection device, used to detect the natural fibers transported by the fiber transport device, identify the foreign fibers in the natural fibers and generate the current foreign fiber coordinates corresponding to the foreign fiber removal; a foreign fiber removing device for removing the foreign fibers according to the current foreign fiber coordinates corresponding to the foreign fiber removal identified by the foreign fiber detecting device; Wherein, the foreign fiber detection device comprises: A photographing mechanism, disposed above the fiber conveying device, for photographing the natural fibers conveyed by the fiber conveying device; A computing mechanism, for receiving the photographed photos, performing computing analysis, and generating a cleaning time and a current foreign fiber coordinate corresponding to the foreign fiber cleaning, wherein the computing mechanism uses at least two deep learning algorithms selected from the group consisting of a YOLO neural network algorithm, an R-CNN neural network algorithm, a Fast R-CNN neural network algorithm, a Faster R-CNN neural network algorithm, a RetinaNet neural network algorithm, an EfficientDet neural network algorithm, an ATSS neural network algorithm, and an SSD neural network algorithm to train a foreign fiber recognition model; a cleaning control unit, used for controlling the foreign fiber cleaning device to clean the foreign fibers according to the current foreign fiber coordinates, The operation mechanism includes a model recognition unit for identifying foreign fibers and a coordinate calculation unit for determining the coordinates of foreign fibers. The model recognition unit uses at least two deep learning algorithms selected from the group consisting of a YOLO neural network algorithm, an R-CNN neural network algorithm, a Fast R-CNN neural network algorithm, a Faster R-CNN neural network algorithm, a RetinaNet neural network algorithm, an EfficientDet neural network algorithm, an ATSS neural network algorithm, and an SSD neural network algorithm to respectively train sample images of foreign fibers until overfitting occurs; then a prediction operation is performed on the training set, and a data enhancement method is used to obtain more unlabeled foreign fibers, and the detection frames of the predicted unlabeled foreign fibers are merged with the original detection frames to generate new training data for the next stage of detection network training; the training is iteratively performed by using the above method until it is found that the recall rate no longer increases, and the training is stopped; The model recognition unit uses the horizontal plane where the conveyor belt is located as a coordinate system, and the coordinate system uses the positioning mark as the origin; the model recognition unit uses different initial recognition models to recognize the test pictures of foreign fibers, and corrects the current initial recognition model according to the recognition result of another initial recognition model until the result deviation of the two models does not exceed the preset range value.

2. The foreign fiber removal system according to claim 1, characterized in that: The coordinate calculation unit receives the historical foreign fiber coordinates corresponding to the foreign fiber indicated by the positioning mark of the model recognition unit, and obtains the conveying speed of the fiber conveying device. The coordinate calculation unit generates a foreign fiber removal execution time corresponding to the foreign fiber and a current foreign fiber coordinate corresponding to the foreign fiber removal according to the historical foreign fiber coordinates and the conveying speed.

3. The foreign fiber removal system according to claim 1, characterized in that: The flipping mechanism is arranged between the tail end of the first conveying mechanism and the head end of the second conveying mechanism, and comprises a motor, a roller body connected to the output shaft of the motor, and a plurality of flipping paddles arranged on the roller body.

4. The foreign fiber removal system according to claim 1, characterized in that: The height difference between the vertical height of the first conveying mechanism and the vertical height of the second conveying mechanism ranges from 40 to 60 cm.

5. The foreign fiber removal system according to claim 1, characterized in that: The foreign fiber removal device comprises a linear module and a gripper module, wherein the linear module is used to control the movement of the gripper module, and the gripper module comprises a cylinder and a multi-link gripping mechanism that can release and close multiple claws through the cylinder.

6. The foreign fiber removal system according to claim 5, characterized in that: The multi-link gripping mechanism comprises an L-shaped gripper connecting seat, a gripper driving assembly and a gripper base. The vertical end surface of the gripper connection seat is connected to the linear module, and the gripper connection seat is provided with a triangular fixed shaft for fixing the gripper base at the L angle; The gripper drive assembly is arranged on the horizontal end surface of the gripper connection seat and passes through the horizontal end surface, including a gripper drive motor with an output shaft passing through the gripper connection seat downward and a rocker support seat connected to the output shaft; The gripper base is fixedly connected to the triangular fixed shaft by bolts, and the gripper base includes a rotating shaft, a connecting arm and a rocker arm gripper. The rotating shaft is provided with the rocker arm support seat and is driven to rotate, the connecting arm is fixed to the rotating shaft, and the rocker arm gripper is fixed to the connecting arm.

7. The foreign fiber removal system according to claim 1, characterized in that: The foreign fiber removal device is a signal-controlled jet mechanism for spraying foreign fibers in the fiber at a specific time and sending them into a foreign fiber collection port. The jet mechanism includes a solenoid valve controlled by the removal control unit and a jet tank electrically connected to the solenoid valve.

Images