Clamping method of multi-point flexible packaging bag clamping device

Abstract

This invention discloses a clamping method for a multi-point flexible packaging bag clamping device, belonging to the field of clamping devices. A multi-point flexible packaging bag clamping device includes: a support frame; the multi-point flexible packaging bag clamping device further includes: an opening mechanism located on one side of the support frame for opening the packaging bag; a bag-flipping mechanism located on one side of the opening mechanism for flipping the packaging bag; a rotating mechanism mounted on the support frame; a connecting rod mounted on the rotating mechanism; a bag-feeding mechanism mounted on the connecting rod; a clamping mechanism mounted on the bag-feeding mechanism and located below the bag-feeding mechanism; and suction cup frames mounted on the clamping mechanism and located below the clamping mechanism; wherein the clamping mechanism drives a pair of suction cup frames to clamp the packaging bag. The beneficial effect of this application is that it provides a multi-point flexible packaging bag clamping device and clamping method that increases the clamping stability by adjusting the clamping angle and position of the packaging bag through the rotation of a first slider and the lateral movement of a second slider.

Description

Technical Field

[0001] This invention relates to the field of clamping devices, and more specifically, to a clamping method for a multi-point flexible packaging bag clamping device. Background Technology

[0002] In existing packaging equipment, manual handling of packaging bags is often used, which is inefficient.

[0003] Chinese utility model patent CN205346294U discloses a packaging bag clamping device, including a frame, a bag mouth clamping mechanism, and a bag body fixing mechanism. The bag body fixing mechanism is installed near the middle of the frame, and the bag mouth clamping mechanism is installed at the top of the frame. A slider is provided on the frame, and a slide rail is provided at the bottom of the frame. The slider cooperates with the slide rail. The bag mouth clamping mechanism includes a clamping plate and a cylinder a. The clamping plate is installed on the cylinder a. The bag body fixing mechanism includes a support plate, a pad, and a cylinder b. The cylinder b is fixed on the support plate, and the pad is installed on the cylinder b.

[0004] The above structure cannot be adapted to packaging bags of different sizes, nor can the state of the packaging bag be adjusted to improve the stability of the clamp. Summary of the Invention

[0005] The purpose of this invention is to provide a clamping method for a multi-point flexible packaging bag clamping device. This invention can adjust the clamping angle and position of the packaging bag by rotating the first slider and laterally moving the second slider, thereby increasing the stability of the clamping. The real-time status of the packaging bag is detected by a CCD camera, and the information on the bag's position and tilt angle is input. After calculation, the first and second sliders are driven to adjust the tilt angle and position of the packaging bag respectively, further increasing the stability of the clamping.

[0006] The summary section of this application is intended to provide a brief overview of the concepts, which will be described in detail in the detailed description section below. This summary section is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0007] To address the technical problems mentioned in the background section, some embodiments of this application provide a multi-point flexible packaging bag clamping device, including:

[0008] support;

[0009] Its features are:

[0010] The multi-point flexible packaging bag clamping device also includes:

[0011] An opening mechanism, located on one side of the bracket, is used to open the packaging bag.

[0012] The bag-flipping mechanism, located on one side of the opening mechanism, is used to flip the packaging bag.

[0013] The rotary mechanism is mounted on the support frame;

[0014] The connecting rod is mounted on the rotary mechanism;

[0015] The bag feeding mechanism is located on the connecting rod;

[0016] The clamping mechanism is located on the bag feeding mechanism and below it.

[0017] The suction cup holder is mounted on the clamping mechanism and located below the clamping mechanism.

[0018] The clamping mechanism drives a pair of suction cup holders to clamp the packaging bag.

[0019] Furthermore, the suction cup holder includes:

[0020] Frame;

[0021] The first slider is rotatably mounted on the frame;

[0022] The second slider is slidably mounted on the frame and is located on both sides of the first slider;

[0023] A cylinder is mounted on the first and second sliders;

[0024] A fixed suction cup is mounted on the cylinder;

[0025] The cylinder drives the fixed suction cup to move back and forth; the frame is equipped with a camera for collecting information about the packaging bags.

[0026] A clamping method for a multi-point flexible packaging bag clamping device, characterized in that:

[0027] Includes the following steps:

[0028] S1. The packaging bag is positioned at multiple points using a fixed suction cup;

[0029] S2. Collect information about the packaging bag using a camera;

[0030] S3. The packaging bag is tilted and adjusted by rotating the first slider;

[0031] S4. Adjust the position of the packaging bag by sliding the second slider laterally.

[0032] Further, step S2 includes the following steps:

[0033] A. Preprocess the image;

[0034] B. Perform image training;

[0035] C. Region segmentation.

[0036] Furthermore, in step A, the original image is first filtered, then converted to grayscale, and finally enhanced.

[0037] Image filtering uses bilateral filtering, and the formula is:

[0038] ;

[0039] In the formula, This is the filtered image. For spatial domain weights, For grayscale weights, For normalization parameters, For noisy images, For pixels The neighborhood range of the location;

[0040] in,

[0041] ;

[0042] ;

[0043] .

[0044] Furthermore, image grayscale conversion employs a weighted average method, where the three components are weighted according to their relative importance. The calculation formula is as follows:

[0045] ;

[0046] in, The three primary colors, These are the weighting coefficients;

[0047] Image enhancement processing first uses the Sobel operator for edge detection, which is performed for any point in the image. Two types of convolutional kernels are used for edge detection in the horizontal and vertical directions, respectively:

[0048] ;

[0049] ;

[0050] In the formula, and For point The horizontal and vertical edge response intensities at a given location represent the likelihood that the edge is an image edge.

[0051] Next, feature extraction is performed. Starting from the top right corner of the image, all pixels in the image are traversed from left to right and from top to bottom. The following operations are performed on the pixels: the gradient value of the edge detection image is obtained. If the gradient of a certain point is greater than the threshold T and the gradient value of the pixel at a horizontal distance L is also greater than the threshold T, then this point is the inner ring edge point. All pixel values ​​between these two points are set to 255, and the pixel values ​​of pixels that do not meet the conditions are set to 0. The image enhancement is thus completed.

[0052] Furthermore, the region segmentation involves inputting the packaged bag image with labeled boxes obtained after network training into the segmentation network, performing image segmentation algorithms on the tilt and width of the woven bag to segment the woven bag region, and improving the segmentation for cases with missing segmentation edges and unsatisfactory segmentation under lighting conditions.

[0053] The improved algorithm takes an input tensor, where N represents the number of images in the input batch, H, W, and C represent the height, width, and number of feature channels of the tensor, respectively, and BN represents standardizing each feature channel using the mean and standard deviation of a mini-batch of images.

[0054] ;

[0055] In the formula, and These are affine parameters learned from the training data. and These are the mean and standard deviation calculated for each feature channel of the current batch of images;

[0056] ;

[0057] ;

[0058] In the formula, N represents the number of image samples in a batch of input; H represents the height of the input tensor; and W represents the width of the input tensor. It is a very small constant; Let IN represent the pixel value at the height h and width w of the c-th channel of the input tensor in the n-th sample. IN is the normalized feature channel of each image.

[0059] ;

[0060] in and Let represent the mean and standard deviation of each feature channel of the image, respectively, and their definitions are as follows:

[0061] ;

[0062] .

[0063] Furthermore, in step S3, the step of adjusting the tilt of the packaging bag involves first detecting the tilt angle of the packaging bag, then determining whether the state of the packaging bag is stable based on the tilt angle, and then rotating the unstable packaging bag to a stable state by using a fixed suction cup and the first slider.

[0064] Further, in step B, the image training step first collects a number of woven bag images and labels them. Then, the labeled dataset is divided into a training set and a validation set in a 1:1 ratio. The dataset is then fed into the network for pre-training to obtain pre-training weights. Using these weights, the dataset is fed into the improved network again for training. Finally, a script is written to calculate the accuracy of the validation set.

[0065] Furthermore, the network is improved by using the Wiou loss function, assuming the anchor box is... The target box is The formula for calculating wiou is:

[0066] ;

[0067] ;

[0068] ;

[0069] ;

[0070] ;

[0071] ;

[0072] in, , This refers to the size of the overlapping portion of the two frames. , The size of the minimum bounding box. These are the weighting coefficients. Calculate the similarity of the aspect ratio of the two boxes. These represent the x and y coordinates of the center point of the anchor frame, as well as the width and height of the anchor frame. These represent the x and y coordinates of the center point of the prediction box, and the width and height of the anchor box, respectively. , The minimum bounding box size; the superscript * indicates that... , Separate it from the computation graph and transform it from a variable into a constant;

[0073] Introducing GAM attention, given a feature map After passing through two independent attention sub-modules, the original feature map is first corrected through the channel attention mechanism module to obtain an intermediate state. Then, the feature map is corrected again by the spatial attention mechanism module to obtain the final feature map. ;

[0074] The definition of an intermediate state is:

[0075] ;

[0076] in, This is a channel attention diagram. This represents element-based multiplication.

[0077] Furthermore, in step S4, the position of the packaging bag is adjusted, and the width of the packaging bag is set to the default value. When the width of the packaging bag is detected to exceed the threshold, the fixed suction cup on the first slider retracts, and the second slider moves to clamp the packaging bag.

[0078] Compared with the prior art, the present invention has the following beneficial effects:

[0079] 1. By rotating the first slider, the tilt angle of the packaging bag is adjusted, so that the packaging bag remains horizontal during the insertion process, which improves the efficiency of insertion and the stability of the packaging bag during the production process.

[0080] 2. The second slider's lateral sliding mechanism adapts to packaging bags of different sizes and shapes, increasing clamping stability.

[0081] 3. The camera collects the status information of the packaging bag, enabling the first and second sliders to automatically adjust the packaging bag according to its status, making operation convenient. Attached Figure Description

[0082] The accompanying drawings, which form part of this application, are used to provide a further understanding of the application and to make other features, objects, and advantages of the application more apparent. The illustrative embodiments and descriptions of this application are used to explain the application and do not constitute an undue limitation of the application.

[0083] Furthermore, throughout the accompanying drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and the elements are not necessarily drawn to scale.

[0084] In the attached diagram:

[0085] Figure 1 This is an overall schematic diagram according to Embodiment 1 of this application;

[0086] Figure 2 This is a structural schematic diagram of a part of Embodiment 1, mainly showing the support structure;

[0087] Figure 3 This is a structural schematic diagram of a part of Embodiment 1, mainly showing the suction cup holder structure;

[0088] Figure 4 This is a flowchart of the detection and adjustment process;

[0089] Figure 5 This is a diagram showing the packaging bag tilted.

[0090] Figure 6 This is a flowchart of the image preprocessing process;

[0091] Figure 7 It is a convolution kernel template image;

[0092] Figure 8 This is a flowchart of the feature extraction process;

[0093] Figure 9 This is an image training flowchart;

[0094] Figure 10 This is a diagram of the GAM attention module structure;

[0095] Figure 11 This is a diagram of the channel attention submodule;

[0096] Figure 12 This is an improved DeepLab V3 network structure diagram;

[0097] Figure 13 This is a flowchart of the process for adjusting the tilt of the packaging bag;

[0098] Figure 14 This is a flowchart of the packaging bag position adjustment process;

[0099] Figure 15 This is an overall schematic diagram based on Embodiment 3 of this application.

[0100] Figure label:

[0101] 100. Multi-point flexible packaging bag clamping device;

[0102] 101. Bracket;

[0103] 102. Slewing mechanism;

[0104] 103. Connecting rod;

[0105] 104. Bag delivery mechanism;

[0106] 105. Clamping mechanism;

[0107] 106, suction cup holder; 106a, frame; 106b, first slider; 106c, second slider; 106d, cylinder; 106e, fixed suction cup;

[0108] 107. Opening mechanism;

[0109] 108. Bag-flipping mechanism;

[0110] 109. Workbench. Detailed Implementation

[0111] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0112] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other.

[0113] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0114] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0115] This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0116] Example 1

[0117] Reference Figure 1-3 A multi-point flexible packaging bag clamping device 100 includes: an opening mechanism 107, a bag flipping mechanism 108, a support 101, a rotating mechanism 102, a connecting rod 103, a bag feeding mechanism 104, a clamping mechanism 105, and a suction cup frame 106. The opening mechanism 107 is fixed on a workbench 109 and is used to open the opening of the packaging bag. The bag flipping mechanism 108 is fixed on the workbench and is used to flip the packaging bag onto the workbench. The support 101 is located on one side of the workbench. The rotating mechanism 102 is mounted on the support 101. The connecting rod 103 is fixed on the rotating mechanism 102. The bag feeding mechanism 104 is mounted on the connecting rod 103 at the end away from the rotating mechanism 102. The clamping mechanism 105 is mounted on the bag feeding mechanism 104 and located below the bag feeding mechanism 104. There is a pair of suction cup frames 106, which are mounted on the clamping mechanism 105 and located below the clamping mechanism 105.

[0118] Reference Figure 3 The suction cup holder 106 includes: a frame 106a, a first slider 106b, a second slider 106c, a cylinder 106d, and a fixed suction cup 106e; the first slider 106b is rotatably mounted on the frame 106a and located at the middle of the frame 106a; there are two second sliders 106c, which are laterally slidably mounted on the frame 106a and located on both sides of the first slider 106b; there are two cylinders 106d, which are arranged around the first slider 106b and the second slider 106c; the fixed suction cup 106e is fixed on the cylinder 106d.

[0119] A CCD camera for collecting information about packaging bags is installed on the frame 106a.

[0120] First, the opening mechanism 107 picks up the placed packaging bag using a suction cup. Then, the gripper in the opening mechanism 107 opens the bag opening. Next, the flipping mechanism 108 flips the packaging bag over to the worktable 109 using a rotating gripper for the bag inserting machine to clamp. Then, a pair of hinged rods in the clamping mechanism 105 close, clamping the packaging bag using a fixed suction cup 106e on the suction cup frame 106. Subsequently, the motor in the rotary mechanism 102 starts, driving the connecting rod to rotate and sending the packaging bag out of the worktable 109. Under the action of the electric guide rail in the bag feeding mechanism 104, the packaging bag is fed into the feeding mechanism.

[0121] The lateral sliding of the second slider 106c can flexibly adjust the position of the fixed suction cup 106e according to the size of the packaging bag, achieving the effect of eliminating the need for replacement.

[0122] Example 2

[0123] Reference Figure 4-14 A clamping method for a multi-point flexible packaging bag clamping device includes the following steps:

[0124] S1. The packaging bag is positioned at multiple points using a fixed suction cup;

[0125] S2. Collect information about the packaging bag using a camera;

[0126] S3. The packaging bag is tilted and adjusted by rotating the first slider;

[0127] S4. Adjust the position of the packaging bag by sliding the second slider laterally.

[0128] Reference Figure 5 The CCD camera determines the coordinates of two vertices A and B of the packaging bag by taking pictures, calculates the tilt angle α of the packaging bag and the distance AB, and the width of the packaging bag, thereby determining whether the tilt angle of the packaging bag exceeds the threshold of 5° and whether the width of the packaging bag exceeds the normal processing size of 600mm.

[0129] Inclination:

[0130] ;

[0131] Packaging bag width:

[0132] ;

[0133] In the formula, Let A be the coordinates of point A on the X-axis; Let B be the coordinate of the point on the X-axis; Let A be the coordinate of the point on the Y-axis; Let B be the coordinates of point B on the Y-axis.

[0134] Step S2 includes the following steps:

[0135] A. Preprocess the image;

[0136] B. Perform image training;

[0137] C. Region segmentation.

[0138] In step A, the image preprocessing involves first filtering the original image, then converting it to grayscale, and finally enhancing the image.

[0139] Image filtering uses bilateral filtering, and its formula is as follows:

[0140] ;

[0141] In the formula, This is the filtered image. For spatial domain weights, For grayscale weights, For normalization parameters, For noisy images, For pixels The neighborhood range of the location;

[0142] in,

[0143] ;

[0144] ;

[0145] .

[0146] Image grayscale conversion uses a weighted average method to calculate a weighted average of the three components according to their importance. The calculation formula is as follows:

[0147] ;

[0148] in, The three primary colors, These are the weighting coefficients;

[0149] Reference Figure 7 Image enhancement processing first uses the Sobel operator for edge detection, and then uses a 3x3 convolution kernel to perform a convolution operation on the grayscale image to further detect edges. The convolution kernel template is as follows: Figure 7 Suppose the image that needs edge detection is like... Figure 7 As shown, for any point P0 in this image, edge detection in the horizontal and vertical directions is performed using the two convolution kernels shown respectively:

[0150] ;

[0151] ;

[0152] In the formula, and For point The horizontal and vertical edge response intensities at a given location represent the likelihood that the edge is an image edge.

[0153] Reference Figure 6-7 Then, feature extraction is performed, starting from the top right corner of the image, traversing all pixels from left to right and top to bottom, and performing the following operations on each pixel:

[0154] The first step is to obtain the gradient value g(x, y);

[0155] The second step is to use the following formula to determine whether the pixel may be an edge point of the inner support ring: g(x,y)>T&&g(x,y)>T;

[0156] Third, if the formula is true, this pixel may be the edge point of the inner support ring, and set all pixel values ​​between these two points to 255; if the formula is false, set the pixel value of the pixel to 0.

[0157] Where g(x,y) is the gradient value of the edge detection image, T is the gradient threshold, and L is the lane width.

[0158] If the gradient g(x,y) at a certain point is greater than the threshold T and the gradient value of the pixel g(x+L,y) at a horizontal distance L is also greater than the threshold T, then this point is a possible inner support ring edge point. Set all pixel values ​​between these two points to 255, and set the pixel values ​​of pixels that do not meet the conditions to 0. This completes the image enhancement.

[0159] Reference Figure 9For image training, firstly, a number of woven bag images were collected and labeled using the labelimg software. Next, the labeled dataset was divided into a training set and a validation set in a 1:1 ratio. Then, the dataset was fed into an ultralytics network for pre-training to obtain pre-trained weights. These weights were then used to train the dataset again into the improved ultralytics network. Finally, a script was written to calculate the accuracy on the validation set.

[0160] Improved Ultralytics Network:

[0161] (1) Using the Wiou loss function: Assuming the anchor box is The target box is The formula for calculating wiou is as follows:

[0162] ;

[0163] ;

[0164] ;

[0165] ;

[0166] ;

[0167] ;

[0168] in, , This refers to the size of the overlapping portion of the two frames. , The size of the minimum bounding box. These are the weighting coefficients. Calculate the similarity of the aspect ratio of the two boxes. These represent the x and y coordinates of the center point of the anchor frame, as well as the width and height of the anchor frame. These represent the x and y coordinates of the center point of the prediction box, and the width and height of the anchor box, respectively. , The minimum bounding box size; the superscript * indicates that... , Separating it from the computation graph and turning it from a variable into a constant can prevent... This generates a gradient that hinders convergence.

[0169] (2) Introducing GAM attention

[0170] This algorithm model introduces the GAM (global attention mechanism). GAM is a novel attention module following CBAM, and its structure is as follows: Figure 10 As shown.

[0171] GAM is a module that combines channel attention and spatial attention mechanisms, and it's an attention mechanism that can amplify the interaction of global dimensional features while reducing diffuse information. Given a feature map... After passing through two independent attention sub-modules, the original feature map is first corrected through the channel attention mechanism module to obtain an intermediate state. Then, the feature map is corrected again by the spatial attention mechanism module to obtain the final feature map. The channel attention submodule is as follows: Figure 11 As shown.

[0172] The channel attention submodule uses a three-dimensional arrangement to preserve three-dimensional information, and then uses a two-layer multilayer perceptron (MLP) to amplify cross-dimensional channel spatial dependencies.

[0173] The definition of an intermediate state is:

[0174] ;

[0175] in, This is a channel attention diagram. This represents element-based multiplication.

[0176] Input feature map The size is CxHxW, and two 7x7 convolutions are used to model the non-linear relationships between pixels in the 7x7 square, so the parameters can capture more of the relationships between pixels in the pixel space.

[0177] The definition of an intermediate state is:

[0178] ;

[0179] in It is a spatial attention diagram.

[0180] Reference Figure 11 The DeepLabV3-based region segmentation algorithm obtains woven bag images with bounding boxes after training the Ultralytics network. These images are then used as input to the DeepLabV3 segmentation network to calculate the tilt and width of the woven bags. A deep learning-based image segmentation algorithm is then used to segment the woven bag region, and improvements are made to address issues such as missing segmentation edges and suboptimal segmentation under lighting conditions.

[0181] Improvement: An improved algorithm incorporating IBN-ResNet is used. Given an input tensor, N represents the number of images in the input batch, and H, W, and C represent the height, width, and number of feature channels of the tensor, respectively. BN means standardizing each feature channel using the mean and standard deviation of a mini-batch of images, i.e.:

[0182] ;

[0183] in, and These are affine parameters learned from the training data, with the aim of recovering the feature distribution prior to BN. and These are the mean and standard deviation calculated for each feature channel of the current batch of images.

[0184] ;

[0185] Where N represents the number of image samples in a batch of input; H represents the height of the input tensor; and W represents the width of the input tensor. It is a very small constant; This represents the pixel value at the height h and width w of the c-th channel of the input tensor in the n-th sample. IN is the normalized feature channel of each image;

[0186] ;

[0187] in and Let represent the mean and standard deviation of each feature channel of the image, respectively, and their definitions are as follows:

[0188] ;

[0189] .

[0190] IBN-Net is proposed, a standardized structure that combines IN and BN. It uses both IN and BN in shallow layers and mainly uses BN in deep layers, which can better learn features and has better generalization performance.

[0191] Inclination calculation:

[0192] The segmented image is then labeled with angles, such as... Figure 5 .

[0193] Reference Figure 13The adjustment of the packaging bag tilt is as follows: if the CCD detects that the packaging bag is tilted by more than 5°, it is considered that the state of the packaging bag is unstable, which will affect the accuracy of bag insertion. Therefore, a rotating slider is required for correction. The suction cup on the rotating slider 1 first performs a clamping operation to prevent it from sliding during the correction process. Under the action of the rotating cylinder, it rotates α° to make the packaging bag in a horizontal position, thus completing the tilt correction. If the tilt of the packaging bag exceeds 40°, it is considered that the bag flipping mechanism has failed to pick up the packaging bag and has exceeded the correction limit. The bag insertion machine puts the packaging bag into the waste area, uploads the waste information, and proceeds to the next bag insertion operation to ensure the normal operation of the production process.

[0194] Reference Figure 13 The packaging bag position is adjusted. The default width of the packaging bag is 520mm. If the CCD detects that the width of the packaging bag exceeds 600mm, the suction cup on the rotating slider will retract so that the horizontal slider can bear the clamping operation. The horizontal slider 2.3 moves to the position where the clamping force is most suitable to clamp the packaging bag (the packaging bag is subjected to the most uniform torque), preventing it from deforming under the influence of gravity and suction, and improving the accuracy of inserting the bag.

[0195] Force analysis shows that the deformation between the support points is minimal when the two fulcrums of the clamp are 6 / 25L from the ends. Therefore, the distance between the transverse sliders 2.5 should be 13 / 25AB.

[0196] By making the above adjustments, we can ensure that the packaging bags remain stable throughout the production process and improve the accuracy of bag insertion. Furthermore, this adjustment method is highly flexible and can be adapted to different production scenarios and needs.

[0197] In summary, the support unit is a key component of the multi-point flexible clamping system, and its Z-axis adjustment is achieved via a cylinder. This design allows the support unit to be flexibly adjusted to accommodate packaging bags of various sizes and shapes. Furthermore, the system is equipped with multiple support units, which enables the clamps to better fit the packaging bags, improving packaging efficiency.

[0198] The lateral sliding of the second slider can adapt to packaging bags of different sizes and shapes, making the clamping more stable. This design makes the system highly adaptable, enabling efficient clamping of both large and small packaging bags.

[0199] Furthermore, the rotation of the first slider can correct the tilt of the packaging bag, ensuring that the bag remains horizontal during insertion. This design not only improves the efficiency of insertion but also guarantees the stability of the packaging bag during production.

[0200] The multi-point flexible packaging bag clamp device also has the following advantages:

[0201] 1. Suitable for packaging bags of different sizes: By sliding the second slider on the left and right sides of the suction cup frame laterally to adjust the fixed suction cup, it can adapt to packaging bags of various sizes and meet different production needs.

[0202] 2. Multi-point positioning: Multiple fixed suction cups enable multi-point positioning, which improves the stability of gripping the packaging bag and avoids problems such as the packaging bag falling off due to insufficient gripping points.

[0203] 3. Correction and adjustment: The packaging bag is corrected by rotating the first slider, so that the packaging bag is kept horizontal when inserted, which improves the insertion efficiency.

[0204] 4. Improved production efficiency: The design of the multi-point flexible packaging bag clamping device makes the packaging bag more stable during the production process, reducing production interruptions caused by packaging bag instability, thereby improving overall production efficiency.

[0205] Example 3

[0206] Reference Figure 15 The coordinated control of the multi-point flexible packaging bag clamping device and the cement packaging machine: In order to achieve coordinated operation between the multi-point flexible packaging bag clamping device and the cement packaging machine, given that the rotary material tower has a large inertia and it is difficult to adjust its rotation speed, the bag insertion efficiency can be improved and stability can be ensured by adjusting the position of the packaging bag on the multi-point flexible packaging bag clamping device.

[0207] Based on the standard 520mm packaging bag insertion, the bag feeding mechanism 4 adjusts the initial extension of the bag feeding mechanism according to the difference between the width of different sized packaging bags and the standard width, simulating the insertion action of the standard packaging bag opening, thereby simplifying the operation structure and improving the stability of bag insertion.

[0208] The above description is merely a selection of preferred embodiments of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of this disclosure.

Claims

1. A clamping method for a multi-point flexible packaging bag clamping device, comprising: Stent (101); Its features are: The multi-point flexible packaging bag clamping device also includes: An opening mechanism (107), located on one side of the bracket (101), is used to open the opening of the packaging bag; A bag-flipping mechanism (108) is located on one side of the opening mechanism (107) and is used to flip the packaging bag; A rotary mechanism (102) is mounted on the support (101); A connecting rod (103) is provided on the rotary mechanism (102); A bag feeding mechanism (104) is provided on the connecting rod (103); A clamping mechanism (105) is provided on the bag feeding mechanism (104) and located below the bag feeding mechanism (104); A suction cup holder (106) is disposed on the clamping mechanism (105) and located below the clamping mechanism (105); The clamping mechanism (105) drives a pair of suction cup holders (106) to clamp the packaging bag; The suction cup holder (106) includes: Frame (106a); The first slider (106b) is rotatably mounted on the frame (106a); The second slider (106c) is laterally slidably disposed on the frame (106a) and located on both sides of the first slider (106b); A cylinder (106d) is disposed on the first slider (106b) and the second slider (106c); A fixed suction cup (106e) is mounted on the cylinder (106d); The cylinder (106d) drives the fixed suction cup (106e) to move back and forth; the frame (106a) is equipped with a camera for collecting information about the packaging bag; The clamping method includes the following steps: S1. The packaging bag is positioned at multiple points using a fixed suction cup; S2. Collect status information of the packaging bag using a camera; S3. Based on the status information, the packaging bag is tilted and adjusted by rotating the first slider so that the packaging bag remains horizontal during the insertion process; S4. Based on the status information, adjust the position of the packaging bag by sliding the second slider laterally to adapt to packaging bags of different sizes and outlines.

2. The clamping method of the multi-point flexible packaging bag clamping device according to claim 1, characterized in that: Step S2 includes the following steps: A. Preprocess the image; B. Perform image training; C. Region segmentation; In step A, the original image is first filtered, then converted to grayscale, and finally enhanced. The image filtering uses bilateral filtering, and the formula is as follows: ； In the formula, This is the filtered image. For spatial domain weights, For grayscale weights, For normalization parameters, For noisy images, For pixels The neighborhood range of the location; in, ； ； 。 3. The clamping method of the multi-point flexible packaging bag clamping device according to claim 2, characterized in that: The image grayscale conversion is performed using a weighted average method, which calculates the average of the three components according to their importance. The calculation formula is as follows: ； in, The three primary colors, These are the weighting coefficients; The image enhancement process first uses the Sobel operator for edge detection, which is performed for any point in the image. Two types of convolutional kernels are used for edge detection in the horizontal and vertical directions, respectively: ； ； In the formula, and For point The horizontal and vertical edge response intensities at a given location represent the likelihood that the edge is an image edge. Next, feature extraction is performed. Starting from the top right corner of the image, all pixels in the image are traversed from left to right and from top to bottom. The following operations are performed on the pixels: the gradient value of the edge detection image is obtained. If the gradient of a certain point is greater than the threshold T and the gradient value of the pixel at a horizontal distance L is also greater than the threshold T, then this point is the inner ring edge point. All pixel values ​​between these two points are set to 255, and the pixel values ​​of pixels that do not meet the conditions are set to 0. The image enhancement is thus completed.

4. The clamping method of the multi-point flexible packaging bag clamping device according to claim 2, characterized in that: The region segmentation involves inputting the packaged bag image with labeled boxes obtained after network training into the segmentation network, performing image segmentation algorithms on the tilt and width of the woven bag to segment the woven bag region, and improving the segmentation for cases with missing segmentation edges and unsatisfactory segmentation under lighting conditions. The improved algorithm takes an input tensor, where N represents the number of images in the input batch, H, W, and C represent the height, width, and number of feature channels of the tensor, respectively, and BN represents standardizing each feature channel using the mean and standard deviation of a mini-batch of images. ； In the formula, and These are affine parameters learned from the training data. and These are the mean and standard deviation calculated for each feature channel of the current batch of images; ； ； In the formula, N represents the number of image samples in a batch of input; H represents the height of the input tensor; and W represents the width of the input tensor. It is a very small constant; Let IN represent the pixel value at the height h and width w of the c-th channel of the input tensor in the n-th sample. IN is the normalized feature channel of each image. ； in and Let represent the mean and standard deviation of each feature channel of the image, respectively, and their definitions are as follows: ； 。 5. The clamping method of the multi-point flexible packaging bag clamping device according to claim 1, characterized in that: The step of adjusting the tilt of the packaging bag in step S3 is to first detect the tilt angle of the packaging bag, then determine whether the state of the packaging bag is stable based on the tilt angle, and then rotate the unstable packaging bag to a stable state by using a fixed suction cup and the first slider.

6. The clamping method of the multi-point flexible packaging bag clamping device according to claim 2, characterized in that: In step B, the image training steps first involve collecting a number of woven bag images and labeling them. Then, the labeled dataset is divided into a training set and a validation set in a 1:1 ratio. The dataset is then fed into the network for pre-training to obtain pre-training weights. These weights are then used to feed the dataset back into the improved network for training. Finally, a script is written to calculate the accuracy of the validation set.

7. The clamping method of the multi-point flexible packaging bag clamping device according to claim 6, characterized in that: The improvements to the network include: using the Wiou loss function, assuming the anchor box is... The target box is The formula for calculating wiou is: ； ； ； ； ； ； in, , This refers to the size of the overlapping portion of the two frames. , The size of the minimum bounding box. These are the weighting coefficients. Calculate the similarity of the aspect ratio of the two boxes. These represent the x and y coordinates of the center point of the anchor frame, as well as the width and height of the anchor frame. These represent the x and y coordinates of the center point of the prediction box, and the width and height of the anchor box, respectively. , The minimum bounding box size; the superscript * indicates that... , Separate it from the computation graph and transform it from a variable into a constant; Introducing GAM attention, given a feature map After passing through two independent attention sub-modules, the original feature map is first corrected through the channel attention mechanism module to obtain an intermediate state. Then, the feature map is corrected again by the spatial attention mechanism module to obtain the final feature map. ; The definition of an intermediate state is: ； in, This is a channel attention diagram. This represents element-based multiplication.

8. The clamping method of a multi-point flexible packaging bag clamping device according to claim 6, characterized in that: In step S4, the position of the packaging bag is adjusted, and the width of the packaging bag is set to the default value. When the width of the packaging bag is detected to exceed the threshold, the fixed suction cup on the first slider retracts, and the second slider moves to clamp the packaging bag.

Images