Woven bag cutting device based on laser technology and control method thereof

Abstract

The invention provides a woven bag cutting device based on a laser technology and a control method of the woven bag cutting device. The control method comprises the steps that an active traction motor is controlled to pull woven base cloth placed on a base cloth feeding frame; the position of the woven base cloth is detected through a photoelectric sensor, and accurate positioning of the woven base cloth is ensured in cooperation with an EPC automatic deviation rectifying device; a spiral line type transition roller is controlled to preliminarily flatten the woven base cloth to obtain preliminarily flattened woven base cloth; controlling a flattening negative pressure box to adsorb the preliminarily flattened woven base cloth to obtain flattened optimized woven base cloth; controlling a laser cutting group to perform laser cutting on the flattened and optimized woven base cloth to obtain laser-cut woven base cloth; and controlling a buffer transition roller to convey the woven base cloth subjected to laser cutting.

Description

Technical Field

[0001] The present application relates to the field of industrial control, and more particularly, in an embodiment of the present application, to a woven bag cutting device based on laser technology and a control method thereof. Background Art

[0002] Woven bags are widely used in the packaging industry, especially in the transportation and storage of bulk materials. However, in the process of processing woven bags, traditional woven bag cutting devices can usually only cut straight bag openings, lack flexibility, and cannot adapt to application scenarios that require special-shaped cutting. This limits the design diversity and functionality of woven bags, especially for applications that require special opening shapes, traditional methods seem to be unable to cope with it. In addition, traditional woven bag cutting methods usually use mechanical knives or thermal cutting technology, and are controlled by fixed programs or manually. These methods may lead to problems such as uneven cutting edges, low efficiency, and material waste.

[0003] Therefore, an optimized control scheme for the woven bag cutting device is desired. Summary of the invention

[0004] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a woven bag cutting device based on laser technology and a control method thereof, which monitors and collects the flattened state image of the base fabric, and introduces an image processing and analysis algorithm based on artificial intelligence and deep learning at the back end to analyze the flattened state image of the base fabric, so as to capture and characterize the global flattened state characteristics of the base fabric, so as to identify and judge whether there are wrinkles on the flattened surface of the base fabric, thereby finding the unevenness or defects of the flattened base fabric, so as to immediately adjust the working parameters of the flattening negative pressure box, thereby improving the flatness of the fabric and ensuring that the woven bag finally cut out has good quality and consistent size. In this way, the dependence on manual intervention can be reduced, the degree of automation can be improved, and the production process can be made more stable and reliable, reducing the scrap rate caused by uneven materials.

[0005] According to one aspect of the present application, a control method for a woven bag cutting device based on laser technology is provided, which comprises: Controlling the active traction motor to pull the woven base fabric placed on the base fabric delivery rack; The position of the woven base fabric is detected by a photoelectric sensor, and the EPC automatic deviation correction device is used to ensure the accurate positioning of the woven base fabric; Controlling the spiral transition roller to preliminarily flatten the woven base fabric to obtain a preliminarily flattened woven base fabric; Controlling the flattening negative pressure box to adsorb the preliminarily flattened woven base fabric to obtain a flattened optimized woven base fabric; Controlling the laser cutting group to perform laser cutting on the flattened optimized woven base fabric to obtain the laser-cut woven base fabric; The buffer transition roller is controlled to transmit the woven base fabric after the laser cutting.

[0006] According to another aspect of the present application, a control system of a woven bag cutting device based on laser technology is provided, which comprises: A base fabric flattened state image acquisition module is used to acquire a base fabric flattened state image captured by an industrial camera; A base fabric flattened state image grid segmentation module, used for performing grid segmentation on the base fabric flattened state image to obtain a set of base fabric flattened grid granularity state images; A base fabric flattened grid granularity state image feature extraction module is used to perform texture feature extraction and spatial feature enhancement processing on each base fabric flattened grid granularity state image in the set of base fabric flattened grid granularity state images to obtain a set of texture grid granularity spatial enhancement feature maps after the base fabric is flattened; The flattened state consistency measurement coefficient calculation module is used to calculate the flattened state consistency measurement coefficient between any two base fabric texture grid granularity spatial enhancement feature maps after flattening in the set of the base fabric texture grid granularity spatial enhancement feature maps after flattening, and perform flattened state identification based on multiple flattened state consistency measurement coefficients to determine whether there are wrinkles and whether to generate parameter adjustment prompts for the flattened negative pressure box.

[0007] Compared with the prior art, the present application provides a woven bag cutting device based on laser technology and a control method thereof, which monitors and collects the flattened state image of the base fabric, and introduces an image processing and analysis algorithm based on artificial intelligence and deep learning at the back end to analyze the flattened state image of the base fabric, so as to capture and depict the global flattened state characteristics of the base fabric, thereby identifying and judging whether there are wrinkles on the flattened surface of the base fabric, thereby finding the unevenness or defects of the flattened base fabric, and adjusting the working parameters of the flattening negative pressure box in real time, thereby improving the flatness of the fabric and ensuring that the woven bag finally cut out has good quality and consistent size. In this way, the dependence on manual intervention can be reduced, the degree of automation can be improved, and the production process can be made more stable and reliable, reducing the scrap rate caused by uneven materials. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0009] Figure 1 This is a flow chart of a control method for a woven bag cutting device based on laser technology according to an embodiment of the present application.

[0010] Figure 2 This is a flow chart of controlling the flattening negative pressure box to adsorb the preliminarily flattened woven base fabric to obtain a flattened optimized woven base fabric in the control method of the woven bag cutting device based on laser technology according to an embodiment of the present application.

[0011] Figure 3 This is a data flow diagram for controlling the flattening negative pressure box to adsorb the preliminarily flattened woven base fabric to obtain a flattened optimized woven base fabric in the control method of the woven bag cutting device based on laser technology according to an embodiment of the present application. DETAILED DESCRIPTION

[0012] Various exemplary embodiments, features and aspects of the present application will be described in detail below with reference to the accompanying drawings. The same reference numerals in the accompanying drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless otherwise specified.

[0013] Woven bag cutting is the process of making woven materials into the desired size and shape, usually using professional cutting tools to ensure precise and neat edges. However, traditional woven bag cutting devices can usually only cut straight bag mouths, and because they use mechanical knives or thermal cutting technology and rely on fixed procedures or manual control, they will lead to problems such as uneven cutting edges, low efficiency and material waste.

[0014] With the development of laser technology, it is possible to use laser for precise cutting, which can provide higher precision, faster speed and better cutting quality. Among them, laser cutting technology is an advanced process that uses high-energy laser beams to accurately cut materials. By focusing the high energy density of the laser beam, it can melt, vaporize or reach the combustion point in a very short time, thereby realizing the cutting of various materials such as metals and non-metals.

[0015] In response to the above technical problems, the present application proposes a control method for a woven bag cutting device based on laser technology. Figure 1 FIG. 1 is a flow chart of a control method of a woven bag cutting device based on laser technology according to an embodiment of the present application. Figure 1As shown, according to the control method of the woven bag cutting device based on laser technology in the embodiment of the present application, it includes: S110, controlling the active traction motor to pull the woven base cloth placed on the base cloth delivery rack; S120, detecting the position of the woven base cloth by a photoelectric sensor, and cooperating with the EPC automatic correction device to ensure the accurate positioning of the woven base cloth; S130, controlling the spiral transition roller to preliminarily flatten the woven base cloth to obtain a preliminarily flattened woven base cloth; S140, controlling the flattening negative pressure box to adsorb the preliminarily flattened woven base cloth to obtain a flattened optimized woven base cloth; S150, controlling the laser cutting group to laser cut the flattened optimized woven base cloth to obtain a laser-cut woven base cloth; S160, controlling the buffer transition roller to transmit the laser-cut woven base cloth.

[0016] In the technical solution of the present application, laser technology is used to cut the woven bag into a special-shaped pattern to ensure the connection state of the tear line shape. After the folding and center seam are bonded into a tube, the cam drives the upper rubber roller to press down and the lower rubber roller to squeeze the bag using the speed ratio difference between the front and rear rubber rollers, and finally forms a trapezoidal bag mouth product, which solves the problem that the traditional center seam bag making machine can only cut straight bag mouths. The woven bag cutting device based on laser technology consists of a base cloth feeding unit, a flattening laser cutting unit, a folding and tube forming unit, a pressing and breaking unit, a counting and conveying unit, a main control electrical box, a center seam material / adhesive bonding unit, a laser control electrical box and a water tank. Among them, the base cloth feeding rack unit is used to automatically feed the base cloth material to ensure a continuous and stable supply of raw materials to improve production efficiency. The flattening laser cutting unit uses laser technology to accurately cut the base cloth and flatten the material at the same time to ensure that the edges after cutting are neat and reduce the difficulty of subsequent processing. The folding and tube-forming unit folds the cut materials and forms a tubular structure, which is suitable for making bags, pipes and other products, improving the appearance and functionality of the finished products. The pressing and breaking unit firmly presses the materials by applying pressure and performs a breaking test to ensure that the strength of the bonding parts meets the standards. The counting and conveying unit is used to track the production quantity in real time and convey the finished products to the next process to optimize the production process. The main control electrical box is a centralized control of the power system of the entire production line to ensure the normal operation of the equipment and stable production. The center seam / glue bonding unit uses center seam material or glue to bond the various parts during the production process to enhance the overall strength and durability of the product. The laser control electrical box is specifically used to control the power supply and cooling system of the laser cutting equipment to ensure that the laser operates at the optimal temperature and extend its service life.

[0017] In the control method of the woven bag cutting device based on laser technology, the step S110 controls the active traction motor to pull the woven base fabric placed on the base fabric delivery rack. It should be understood that the active traction motor is a motor dedicated to controlling the movement of materials, and its main function is to ensure the smooth supply of materials (such as woven base fabrics) during the production process through precise speed and torque control. In the technical solution of the present application, the active traction motor is responsible for pulling the pp / pe woven base fabric from the base fabric delivery rack to ensure that the braiding machine can continuously obtain the required base fabric during operation. The control of this process can be achieved through a closed-loop feedback system to ensure that the tension and speed of the base fabric are always kept in the optimal range, thereby avoiding the woven base fabric from being loose or too tight, so as to maintain the tension consistent during the weaving process, which helps to prevent fabric defects such as wrinkles or breaks, and realizes a higher degree of automation, reduces dependence on manual operation, and improves production efficiency. It should be understood that the traditional base fabric delivery rack is a passive traction type, which is pulled by the main traction motor in front, and the woven bag cutting device based on laser technology of the present application is changed to active traction.

[0018] In the control method of the woven bag cutting device based on laser technology, the step S120 detects the position of the woven base fabric by a photoelectric sensor, and cooperates with the EPC automatic deviation correction device to ensure the accurate positioning of the woven base fabric. It should be understood that the photoelectric sensor, as a non-contact detection device, can monitor the position of the woven base fabric in real time. Its working principle is to use the emission and reception of light. When the base fabric passes through the sensor, the sensor will detect the reflection or shielding of light, thereby judging the actual position of the base fabric. The EPC (electronic climate control) automatic deviation correction device monitors the material position in real time, and combines the sensor feedback to quickly calculate the required deviation correction amount and automatically adjust the operating parameters of related equipment. In the technical solution of the present application, the EPC automatic deviation correction device is closely coordinated with the photoelectric sensor to form a closed-loop control system. When the photoelectric sensor detects that the base fabric deviates from the ideal position, it will transmit the position information to the EPC device. The EPC device calculates the required deviation correction amount by analyzing the data fed back by the sensor, and automatically adjusts the operating parameters of the traction motor to quickly correct the position of the base fabric. Through this collaborative working mechanism, the photoelectric sensor and the EPC automatic deviation correction device can effectively reduce the risk of base fabric deviation and ensure the continuity and stability of the weaving process.

[0019] In the control method of the woven bag cutting device based on laser technology, the step S130 controls the spiral transition roller to preliminarily flatten the woven base fabric to obtain a preliminarily flattened woven base fabric. It should be understood that the spiral transition roller is a key equipment used in weaving and processing, and has a special spiral texture design. The structure of this roller can provide a uniform contact area, thereby applying a stable pressure when the base fabric passes through, effectively reducing wrinkles and unevenness. In the technical solution of the present application, the design of the spiral transition roller allows uniform pressure to be generated during operation to obtain a uniform flattening effect. Among them, by adjusting the rotation speed of the roller, the pressure applied and the arrangement of the spiral transition roller, different degrees of flattening of the base fabric can be achieved, effectively reducing the wrinkles and unevenness of the base fabric, thereby achieving a preliminary flattening effect, providing a good foundation for subsequent processes.

[0020] In the control method of the woven bag cutting device based on laser technology, the step S140 controls the flattening negative pressure box to adsorb the preliminary flattened woven base fabric to obtain a flattened optimized woven base fabric. It should be understood that the flattening negative pressure box firmly fixes the base fabric on the working surface through the negative pressure adsorption force, thereby effectively eliminating the residual wrinkles and unevenness on the surface of the fabric, so that it can achieve the effect of optimized flattening to ensure the stability and consistency of the material in subsequent processing. It is worth mentioning that, considering that laser cutting has extremely high requirements for the flatness of the material, a flattening negative pressure box device is designed in the technical solution of the present application. After the fabric is initially flattened by the spiral transition roller, the fabric is forced to be adsorbed on the surface of the flattening negative pressure box by the flattening negative pressure box, so that the fabric is flattened by the flattening negative pressure box before laser cutting. The flattening negative pressure box is a rectangular sealed body, which has an air volume adjustment port, a fan interface, and an array of multiple close-position air inlet flat plates.

[0021] However, to achieve efficient woven bag production, it is not enough to rely solely on laser cutting. Before cutting, it is necessary to ensure that the woven base fabric remains flat and wrinkle-free during the conveying process, so as to ensure that the cut product size is accurate and the appearance is good. If the woven base fabric is not fully flattened before entering the laser cutting area, it may cause cutting errors or product quality degradation. In the technical solution of the present application, although a spiral transition roller and a flattening negative pressure box are used to reduce wrinkles to ensure the flatness of the woven base fabric and the quality of subsequent laser cutting, it is still possible that the surface of the fabric may be slightly uneven due to various reasons. For example, factors such as the material properties of the fabric itself, changes in ambient temperature and humidity, and uneven distribution of mechanical tension may cause the fabric to be slightly deformed or wrinkled during the conveying process. Therefore, in the step of controlling the flattening negative pressure box to adsorb the preliminary flattened woven base fabric to obtain a flattened optimized woven base fabric, it is necessary to monitor the flattening state of the base fabric in real time. Once unevenness or defects are found, measures should be taken immediately to correct them, such as adjusting the parameters of the flattening negative pressure box.

[0022] Specifically, the technical concept of the present application is to monitor and collect the flattened state image of the base fabric through an industrial camera, and introduce an image processing and analysis algorithm based on artificial intelligence and deep learning at the back end to analyze the flattened state image of the base fabric, so as to capture and characterize the global flattened state characteristics of the base fabric, so as to identify and judge whether there are wrinkles on the flattened surface of the base fabric. In this way, it is possible to timely find the unevenness or defects of the base fabric during the woven bag cutting process. When wrinkles or other defects are detected in the fabric, the system can immediately adjust the working parameters of the flattening negative pressure box, such as changing the pressure distribution of the airflow or adjusting the size of the adsorption force, so as to improve the flatness of the fabric and ensure that the woven bag finally cut has good quality and consistent size. This visual detection and correction mechanism can realize the control of the woven bag cutting device in a more intelligent way, which can not only improve the automation degree of woven bag cutting and reduce the dependence on manual intervention, but also make the production process more stable and reliable, and reduce the scrap rate caused by uneven materials.

[0023] Figure 2 This is a flow chart of controlling the flattening negative pressure box to adsorb the preliminarily flattened woven base fabric to obtain a flattened optimized woven base fabric in the control method of the woven bag cutting device based on laser technology according to an embodiment of the present application. Figure 3 This is a data flow diagram for controlling the flattening negative pressure box to absorb the preliminary flattened woven base fabric to obtain a flattened optimized woven base fabric in the control method of the woven bag cutting device based on laser technology according to an embodiment of the present application. Figure 2 and Figure 3As shown, in an embodiment of the present application, the step S140, controlling the flattening negative pressure box to adsorb the preliminary flattened woven base fabric to obtain a flattened optimized woven base fabric, includes: S141, acquiring a base fabric flattened state image captured by an industrial camera; S142, gridding the base fabric flattened state image to obtain a set of base fabric flattened grid granularity state images; S143, performing texture feature extraction and spatial feature enhancement processing on each base fabric flattened grid granularity state image in the set of base fabric flattened grid granularity state images to obtain a set of base fabric flattened texture grid granularity spatial enhancement feature maps; S144, calculating the flattening state consistency measurement coefficient between any two base fabric flattened texture grid granularity spatial enhancement feature maps in the set of base fabric flattened texture grid granularity spatial enhancement feature maps, and performing flattening state identification based on multiple flattening state consistency measurement coefficients to determine whether there are wrinkles, and determine whether to generate parameter adjustment prompts for the flattening negative pressure box.

[0024] Specifically, in the steps S141 and S142, first, an image of the flattened state of the base fabric captured by an industrial camera is acquired. Then, the image of the flattened state of the base fabric is grid-cut to obtain a set of images of the flattened grid granularity state of the base fabric. By dividing the image into multiple grids, the image information in each grid can be processed and analyzed separately. This method allows the system to detect irregularities or wrinkles within a small range on the surface of the base fabric, and thus more accurately identify the overall flattened state of the base fabric.

[0025] In an embodiment of the present application, the step S143, respectively performing texture feature extraction and spatial feature enhancement processing on each base fabric flattened mesh granularity state image in the set of base fabric flattened mesh granularity state images to obtain a set of spatially enhanced feature maps of the texture mesh granularity after the base fabric is flattened, includes: S1431, respectively inputting each base fabric flattened mesh granularity state image in the set of base fabric flattened mesh granularity state images into a texture feature extractor based on a multi-layer dilated convolution structure to obtain a set of texture mesh granularity feature maps after the base fabric is flattened; S1432, respectively performing spatially constrained feature enhancement processing on each base fabric flattened texture mesh granularity feature map in the set of base fabric flattened texture mesh granularity feature maps to obtain a set of spatially enhanced feature maps of the texture mesh granularity after the base fabric is flattened.

[0026] Specifically, in step S1431, each base fabric flattened grid granularity state image in the set of base fabric flattened grid granularity state images is input into a texture feature extractor based on a multi-layer hole convolution structure to obtain a set of texture grid granularity feature maps after the base fabric is flattened. It should be understood that, by using the texture feature extractor based on a multi-layer hole convolution structure, texture features can be extracted from the base fabric flattened grid granularity state image in each grid, so as to perform subsequent recognition and judgment tasks of the wrinkle condition of the base fabric after flattening.

[0027] Specifically, in step S1432, each of the base fabric flattened texture grid granularity feature maps in the set of base fabric flattened texture grid granularity feature maps is subjected to spatial constraint feature enhancement processing to obtain a set of base fabric flattened texture grid granularity spatial enhancement feature maps. It should be understood that since the base fabric flattened texture grid granularity feature map contains different aspects of features and semantic information about the base fabric flattened texture grid area, these feature information have a mutual correlation relationship, which plays an important role in the subsequent wrinkle recognition and parameter adjustment tasks. For example, the wrinkles of the base fabric after flattening may be presented in different local areas within the base fabric flattened texture grid area. Based on this, in order to improve the quality and expressiveness of the texture grid granularity feature map after the base fabric is flattened, thereby further strengthening the relevant features generated by these wrinkles, so as to better identify whether there are wrinkles, in the technical solution of the present application, each of the texture grid granularity feature maps after the base fabric is flattened in the set of texture grid granularity feature maps after the base fabric is flattened is further subjected to spatially constrained feature enhancement processing to obtain a set of spatially enhanced feature maps of the texture grid granularity after the base fabric is flattened. It is worth mentioning that the spatially constrained feature enhancement processing can better capture the detailed features of the surface of the base fabric by combining the local features of the texture grid after the base fabric is flattened with the global context information, while retaining the overall structural information, which is helpful for subsequent wrinkle recognition tasks.

[0028] In the step S1432, each of the base fabric flattened texture mesh granularity feature maps in the set of the base fabric flattened texture mesh granularity feature maps is subjected to spatial constraint feature enhancement processing to obtain a set of base fabric flattened texture mesh granularity spatial enhanced feature maps, including: S1432-1, after feature dispersion is performed on the base fabric flattened texture mesh granularity feature map to obtain a set of base fabric flattened texture mesh pixel granularity feature vectors, semantic association constraint score calculation is performed on the set of base fabric flattened texture mesh pixel granularity feature vectors to obtain a base fabric flattened texture mesh pixel granularity semantic association spatial constraint score topological representation; S1432-2, based on the base fabric flattened texture mesh pixel granularity semantic association spatial constraint score topological representation, the base fabric flattened texture mesh granularity feature map is enhanced to obtain a base fabric flattened texture mesh granularity spatial enhanced feature map.

[0029] In an embodiment of the present application, the step S1432-1, after feature dispersion of the base fabric flattened texture grid granularity feature map to obtain a set of base fabric flattened texture grid pixel granularity feature vectors, performs semantic association constraint score calculation on the set of base fabric flattened texture grid pixel granularity feature vectors to obtain a topological representation of the semantic association spatial constraint score of the base fabric flattened texture grid pixel granularity, includes: S1432-11, feature dispersion of the base fabric flattened texture grid granularity feature map along the channel dimension to obtain a set of base fabric flattened texture grid pixel granularity feature vectors; S1432-12, calculates the semantic association score between any two base fabric flattened texture grid pixel granularity feature vectors in the set of base fabric flattened texture grid pixel granularity feature vectors. Obtain a topological matrix of semantic association scores of pixel granularity of the texture grid after the base fabric is flattened; S1432-13, based on the spatial distance between any two pixel granularity feature vectors of the texture grid after the base fabric is flattened in the set of pixel granularity feature vectors of the texture grid after the base fabric is flattened, perform spatial attention attenuation modulation on the topological matrix of semantic association scores of pixel granularity of the texture grid after the base fabric is flattened to obtain a topological matrix of spatial soft constraint scores of the semantic association of pixel granularity of the texture grid after the base fabric is flattened; S1432-14, perform atrous convolution coding on the topological matrix of spatial soft constraint scores of the semantic association of pixel granularity of the texture grid after the base fabric is flattened to obtain a topological feature matrix of spatial soft constraint scores of the semantic association of pixel granularity of the texture grid after the base fabric is flattened as a topological representation of the spatial constraint scores of the semantic association of pixel granularity of the texture grid after the base fabric is flattened.

[0030] Specifically, in step S1432-11, the texture grid granularity feature map after the base fabric is flattened is feature scattered along the channel dimension to obtain a set of texture grid pixel granularity feature vectors after the base fabric is flattened. It should be understood that the texture grid granularity feature map is obtained by analyzing the surface features of the base fabric and contains rich texture information. In the technical solution of the present application, the texture features of the fabric surface can be effectively extracted and quantified by feature scattered.

[0031] In an embodiment of the present application, the step S1432-11, which performs feature dispersion on the texture grid granularity feature map of the base fabric after flattening along the channel dimension to obtain a set of pixel granularity feature vectors of the texture grid of the base fabric after flattening, includes: performing feature reshaping processing on the texture grid granularity feature map of the base fabric after flattening along the channel dimension to obtain a set of base fabric flattened texture grid feature vectors; mapping each base fabric flattened texture grid feature vector in the set of base fabric flattened texture grid feature vectors to the Poincare space to obtain a set of pixel granularity feature vectors of the texture grid of the base fabric after flattening.

[0032] Specifically, in the step S1432-12, the semantic association score between any two base fabric flattened texture grid pixel granularity feature vectors in the set of base fabric flattened texture grid pixel granularity feature vectors is calculated to obtain a base fabric flattened texture grid pixel granularity semantic association score topological matrix. It should be understood that this step first maps the base fabric flattened texture grid pixel granularity feature vectors to the Poincare space, and calculates the Poincare distance between each mapped base fabric flattened texture grid pixel granularity feature vector as the semantic association score between any two base fabric flattened texture grid pixel granularity feature vectors. Among them, each pixel granularity feature vector is mapped from the Euclidean space to the Poincare sphere, with the purpose of ensuring that each base fabric flattened texture grid pixel granularity feature vector can adapt to the geometric characteristics of the hyperbolic space while maintaining the original base fabric flattened texture grid pixel granularity semantic information.

[0033] In hyperbolic space, as the distance between two points increases, the actual distance grows exponentially, which means that many nodes can be densely placed near the origin (e.g., similar base fabric flattened texture mesh features), while they can be placed more sparsely far away from the origin (e.g., different base fabric flattened texture mesh features). Due to the distance characteristics of the Poincare space, even slightly different features can be mapped to relatively far positions, which helps to improve the model's ability to distinguish between different categories of base fabric flattened texture mesh features, that is, to distinguish the surface states of different base fabric flattened mesh regions. Compared with Euclidean space, the Poincare space can maintain large differences between features at a smaller space cost. In addition, the Poincare distance not only reflects the degree of physical proximity, but also implies the semantic similarity of different base fabric flattened texture mesh local regions, which can help the model better understand the semantic association between pixels, so as to better perform subsequent wrinkle recognition and detection tasks.

[0034] In an embodiment of the present application, the step S1432-12, calculating the semantic association score between any two base fabric flattened texture grid pixel granularity feature vectors in the set of base fabric flattened texture grid pixel granularity feature vectors to obtain the base fabric flattened texture grid pixel granularity semantic association score topological matrix, includes: calculating the first in the set of base fabric flattened texture grid pixel granularity feature vectors The texture grid pixel granularity feature vector and the first The square of the position difference between the feature vectors of the texture grid pixel granularity after the base cloth is flattened is used to obtain the semantic difference of the texture grid pixel granularity after the base cloth is flattened. The constant 1 is calculated and the first The difference between the squares of the norms of the pixel granularity feature vectors of the texture grid after flattening the first base cloth is used to obtain the The semantic representation of the pixel granularity of the texture grid after the base cloth is flattened; the calculation constant 1 and the The difference between the squares of the norms of the pixel granularity feature vectors of the texture grid after flattening the first base cloth is used to obtain the The pixel granularity semantic representation of the texture grid after flattening the base cloth; calculating the first The semantic representation of the pixel granularity of the texture grid after flattening the base cloth and the After multiplying the products of the semantic representations of the pixel granularity of the flattened texture grid of the base fabric, the semantic difference-norm representation of the pixel granularity of the flattened texture grid of the base fabric is divided by the value after the product to obtain the semantic association information representation of the pixel granularity of the flattened texture grid of the base fabric; after calculating the sum of the double of the semantic association information representation of the pixel granularity of the flattened texture grid of the base fabric and the constant 1, the obtained value is input into the inverse hyperbolic cosine function to obtain the semantic association score value of the pixel granularity of the flattened texture grid of the base fabric; and the multiple semantic association score values ​​of the pixel granularity of the flattened texture grid of the base fabric are arranged in a matrix form to obtain the semantic association score topological matrix of the pixel granularity of the flattened texture grid of the base fabric.

[0035] Specifically, in the step S1432-13, the spatial attention attenuation modulation is performed on the semantic association score topology matrix of the texture grid pixel granularity after the base fabric is flattened based on the spatial distance between any two feature vectors of the texture grid pixel granularity after the base fabric is flattened in the set of the feature vectors of the texture grid pixel granularity after the base fabric is flattened to obtain the semantic association score topology matrix of the texture grid pixel granularity after the base fabric is flattened. It should be understood that considering that the texture grid features of the base fabric flattened in adjacent areas in reality often have stronger correlations than those in distant areas, the spatial attention attenuation modulation is performed on the semantic association score topology matrix of the texture grid pixel granularity after the base fabric is flattened, which can simulate this phenomenon, and such processing helps to construct a spatial soft constraint score topology matrix that is more in line with the actual situation. By introducing spatial soft constraints, the module can adjust the semantic relationship between pixels more finely, thereby improving the distinction between features. This is particularly important for distinguishing between a normal flattened state of the base fabric and a slightly wrinkled state, because they may be very similar visually, but the impact on subsequent laser cutting and the cutting quality assurance of woven bags are quite different. In particular, through the spatial constraint mechanism, the module can emphasize the correlation between adjacent pixels, which helps to enhance the spatial coherence of the feature map. In the process of sensing the flat state of the base fabric and identifying wrinkles, the wrinkled area is often not isolated, but has a certain spatial connection with other surface features. Therefore, maintaining this coherence is crucial for accurately identifying wrinkled areas and parameter control tasks.

[0036] In an embodiment of the present application, the step S1432-13 performs spatial attention attenuation modulation on the semantic association score topology matrix of the texture grid pixel granularity after the base fabric is flattened based on the spatial distance between any two of the set of the texture grid pixel granularity feature vectors after the base fabric is flattened to obtain the semantic association score topology matrix of the texture grid pixel granularity after the base fabric is flattened, including: calculating the first The texture grid pixel granularity feature vector after flattening the base cloth and the The spatial distance between the pixel granularity feature vectors of the texture grid after flattening the base cloth is obtained to obtain the pixel granularity semantic space distance representation; the natural exponential function value with the pixel granularity semantic space distance representation as an exponential power and the natural constant e as the base is calculated to obtain the pixel granularity semantic space distance class support representation; after multiplying the pixel granularity semantic space distance class support representation by the spatial attention attenuation modulation hyperparameter to obtain the spatial distance attention attenuation representation, the pixel granularity semantic association score topological matrix of the texture grid after flattening the base cloth is converted into the pixel granularity semantic space distance class support representation. The eigenvalue of the position is divided by the spatial distance attention attenuation representation to obtain the semantic association space soft constraint score of the texture grid pixel granularity after the base cloth is flattened; and multiple semantic association space soft constraint scores of the texture grid pixel granularity after the base cloth is flattened are arranged in a matrix form to obtain a topological matrix of the semantic association space soft constraint scores of the texture grid pixel granularity after the base cloth is flattened.

[0037] Specifically, in step S1432-14, the topological matrix of the semantic association space soft constraint score of the texture grid pixel granularity after the base fabric is flattened is subjected to hole convolution coding to obtain the topological feature matrix of the semantic association space soft constraint score of the texture grid pixel granularity after the base fabric is flattened as the topological representation of the semantic association space constraint score of the texture grid pixel granularity after the base fabric is flattened. It should be understood that the topological matrix of the semantic association space soft constraint score of the texture grid pixel granularity after the base fabric is flattened is subjected to hole convolution coding, so as to capture the semantic association topological features between the feature vectors of the texture grid pixel granularity after each base fabric is flattened.

[0038] In an embodiment of the present application, the step S1432-2, based on the topological representation of the semantic association space constraint score of the pixel granularity of the texture grid after the base fabric is flattened, enhances the granularity feature map of the texture grid after the base fabric is flattened to obtain the spatial enhanced feature map of the texture grid after the base fabric is flattened, includes: S1432-21, inputting the set of feature vectors of the texture grid after the base fabric is flattened and the spatial soft constraint score topological matrix of the semantic association space of the pixel granularity of the texture grid after the base fabric is flattened into the graph convolutional coding module to obtain the set of context pixel granularity feature vectors of the texture grid after the base fabric is flattened; S1432-22, performing feature shape reshaping on the set of context pixel granularity feature vectors of the texture grid after the base fabric is flattened to obtain the spatial enhanced feature map of the texture grid after the base fabric is flattened.

[0039] Specifically, in the step S1432-21, the set of feature vectors of the texture grid after the base fabric is flattened and the topological matrix of the soft constraint score of the semantic association space of the pixel granularity of the texture grid after the base fabric is flattened are input into the graph convolutional coding module to obtain the set of feature vectors of the context pixel granularity of the texture grid after the base fabric is flattened. It should be understood that the input of the graph convolutional coding module is in the form of a graph, in which the nodes represent the feature vectors of the texture grid after the base fabric is flattened, and the edges represent the relationship between the nodes, that is, the topological features of the soft constraint score of the pixel granularity semantic association space of the texture grid after the base fabric is flattened. By aggregating the feature information of the texture grid after the base fabric is flattened of the neighboring nodes to update the current node state, the learning of the entire graph structure is realized, and in this way, the global semantic context association feature expression between the pixel granularity features of the texture grid after the base fabric is flattened in the area of ​​the texture grid after the base fabric is flattened is enhanced.

[0040] Specifically, in the step S1432-22, the set of the feature vectors of the texture grid context pixel granularity after the base fabric is flattened is reshaped to obtain the spatial enhancement feature map of the texture grid granularity after the base fabric is flattened. It should be understood that, through the feature shape reshaping, the set of the feature vectors of the texture grid context pixel granularity after the base fabric is flattened can be rearranged so as to form an image form corresponding to the actual fabric structure in space, thereby effectively capturing the local characteristics and overall distribution of the texture. Specifically, the values ​​of the feature vectors of the texture grid context pixel granularity after the base fabric is flattened in the set of the feature vectors of the texture grid context pixel granularity after the base fabric is flattened can be mapped to specific pixel positions by using an interpolation method, thereby forming a more dense feature map, i.e., the spatial enhancement feature map of the texture grid granularity after the base fabric is flattened. In addition, the upsampling layer in the convolutional neural network can also be used to automatically enhance the details of the image by learning the context information of the features, and generate the spatial enhancement feature map of the texture grid granularity after the base fabric is flattened. The texture grid granularity spatial enhancement feature map obtained by flattening the base fabric not only maintains the granularity information of the base fabric, but also enhances the spatial correlation between different features. This helps to extract richer contextual information and make the texture analysis results more accurate.

[0041] Specifically, the following spatial constraint enhancement formula is used to perform spatial constraint feature enhancement processing on each of the set of the texture grid granularity feature maps after the base fabric is flattened to obtain a set of spatial enhancement feature maps of the texture grid granularity after the base fabric is flattened; Wherein, the spatial constraint enhancement formula is:

[0042] ; in, is the texture grid granularity feature map after the base fabric is flattened, are the height, width and number of channels of the texture grid granularity feature map after the base fabric is flattened, For feature reshaping, and is the set of texture mesh feature vectors after the base fabric is flattened, The value of is the number of feature vectors in the set of texture mesh feature vectors after the base fabric is flattened. and is the feature transformation weight matrix, and are the first and The pixel granularity feature vector of the texture grid after flattening the base cloth, is the one-norm of the vector, is the inverse hyperbolic cosine function, is the semantic association score calculation operation, is the topological matrix of the semantic association score of the texture grid pixel granularity after the base fabric is flattened. The eigenvalues ​​of the positions, It is the spatial distance calculation operation between vectors. is the spatial attention decay modulation hyperparameter, is the topological matrix of the soft constraint score in the semantic association space of the texture grid pixel granularity after the base fabric is flattened. The eigenvalues ​​of the positions, is the soft constraint score topological matrix of the texture grid pixel granularity semantic association space after the base fabric is flattened, It is the hole convolution coding process. is the topological feature matrix of the soft constraint score of the texture grid pixel granularity semantic association space after the base fabric is flattened. is the graph convolutional encoding operation, is the set of pixel granularity feature vectors of the texture grid context after the base fabric is flattened. It is each base fabric flattened texture grid granularity spatial enhancement feature map in the set of base fabric flattened texture grid granularity spatial enhancement feature maps.

[0043] In an embodiment of the present application, the step S144, calculates the flattened state consistency measurement coefficient between any two base fabric texture grid granularity spatial enhancement feature maps in the set of the base fabric texture grid granularity spatial enhancement feature maps after flattening, and performs flattened state identification based on multiple flattened state consistency measurement coefficients to determine whether there are wrinkles, and determines whether to generate a parameter adjustment prompt for a flattened negative pressure box, including: S1441, calculates the flattened state consistency measurement coefficient between any two base fabric texture grid granularity spatial enhancement feature maps in the set of the base fabric texture grid granularity spatial enhancement feature maps after flattening to obtain a base fabric global flattened state measurement representation matrix composed of multiple flattened state consistency measurement coefficients; S1442, inputs the base fabric global flattened state measurement representation matrix into a classifier-based flattened state identifier to obtain a recognition result, and the recognition result is used to indicate whether there are wrinkles; S1443, in response to the recognition result that wrinkles are present, generates a parameter adjustment prompt for a flattened negative pressure box.

[0044] Specifically, in the step S1441, the flattened state consistency measurement coefficient between any two base fabric flattened texture grid granularity spatial reinforcement feature maps in the set of the base fabric flattened texture grid granularity spatial reinforcement feature maps is calculated to obtain a base fabric global flattened state measurement representation matrix composed of multiple flattened state consistency measurement coefficients. It should be understood that by calculating the flattened state consistency measurement coefficient between the texture grid granularity spatial reinforcement feature maps of different base fabrics after flattening, the spatial reinforcement feature similarity or difference between these different base fabric flattened texture grid areas can be measured, so as to quantify and compare the flattened states in different grid areas to obtain a comprehensive evaluation of the flattened state of the entire base fabric. This flattened state consistency measurement coefficient can reflect the consistency or uniformity of the flattened state between different grid areas. If the consistency measurement coefficient of two grid areas is high, it indicates that the two areas are relatively close in the flattened state, otherwise there may be wrinkles or other inconsistencies. By comparing these coefficients between different grids, a base fabric global flattened state measurement representation matrix can be constructed, which will contain the consistency measurement coefficients between all grid areas. It is worth mentioning that in the technical solution of the present application, the flattened state consistency measurement coefficient between the texture grid granularity spatial enhancement feature maps after the base fabric is flattened can be calculated by cosine similarity. The formula of cosine similarity is: ; in, is the first texture grid granularity spatial enhancement feature vector after flattening the first texture grid granularity spatial enhancement feature map, is the second texture grid granularity spatial enhancement feature vector after flattening the second texture grid granularity spatial enhancement feature map, represents the dot product, represents the norm of a vector, Represents the cosine similarity between the first texture grid granularity spatial enhancement feature vector and the second texture grid granularity spatial enhancement feature vector.

[0045] In particular, considering that the value range of cosine similarity is [-1, 1], in this metric, the closer the value is to 1, the more consistent the flattened states of the two feature maps are.

[0046] Specifically, in step S1442, the base fabric global flattened state metric representation matrix is ​​input into a flattened state identifier based on a classifier to obtain a recognition result, and the recognition result is used to indicate whether wrinkles exist. In other words, the global flattened state metric representation feature of the base fabric is used to perform classification processing to identify and determine whether wrinkles exist on the flattened surface of the base fabric.

[0047] Specifically, in step S1443, in response to the recognition result that wrinkles exist, a parameter adjustment prompt for the flattening negative pressure box is generated. It should be understood that in this way, it is possible to promptly find uneven or defective areas of the base fabric during the woven bag cutting process. When wrinkles or other defects are detected in the fabric, the system can immediately adjust the working parameters of the flattening negative pressure box, such as changing the pressure distribution of the airflow or adjusting the size of the adsorption force, thereby improving the flatness of the fabric and ensuring that the woven bag finally cut has good quality and consistent size. This visual detection and correction mechanism can realize the control of the woven bag cutting device in a more intelligent way, which can not only improve the automation of woven bag cutting and reduce dependence on manual intervention, but also make the production process more stable and reliable, and reduce the scrap rate caused by uneven materials.

[0048] Specifically, in the technical solution of the present application, each base fabric flattened texture grid granularity feature map in the set of base fabric flattened texture grid granularity feature maps respectively represents the texture features of the local area determined by gridding segmentation of the base fabric flattened state image. When the feature enhancement based on spatial soft constraints is performed on each base fabric flattened texture grid granularity feature map in the set of base fabric flattened texture grid granularity feature maps, it can improve the spatial domain expression significance of the texture features of each base fabric flattened texture grid granularity feature map in the set of base fabric flattened texture grid granularity feature maps, but it will also increase the difference in feature manifold expression between each base fabric flattened texture grid granularity feature map in the set of base fabric flattened texture grid granularity feature maps, which will lead to the lack of long-distance difference semantic encoding representation of the feature manifold of the base fabric global flattened state metric representation matrix and the base fabric global flattened state metric representation vector, affecting the accuracy of the recognition result obtained by the classifier-based flattened state identifier.

[0049] Based on this, in a preferred embodiment, the base fabric global flattened state metric representation matrix is ​​input into a classifier-based flattened state identifier to obtain an identification result, including: Expanding the base fabric global flattened state metric representation matrix into a base fabric global flattened state metric representation vector; Arranging the eigenvalues ​​of the base fabric global flattened state metric representation vector in ascending order to form a base fabric global flattened state metric sequence encoding vector; In response to the base fabric global flattened state measurement sequence encoding vector The eigenvalue and The absolute value of the difference between the eigenvalues ​​is less than or equal to the distance difference hyperparameter , calculate the The eigenvalues ​​are similar to the The weighted sum between the eigenvalues ​​is the optimized Eigenvalue ,in, and are the first and second order encoding vectors of the base fabric global flattened state measurement order. Eigenvalues ​​and Eigenvalues, For the optimized eigenvalues, is the first weight hyperparameter, is the second weight hyperparameter; In response to the base fabric global flattened state measurement sequence encoding vector The eigenvalue and The absolute value of the difference between the eigenvalues ​​is greater than the distance difference hyperparameter ; Calculate the square root of the sum of the squares of all eigenvalues ​​of the base fabric global flattened state metric representation vector, multiply the square root by 2 and then divide it by the square of the length of the base fabric global flattened state metric representation vector to obtain the base fabric global flattened state metric space primitive value ; The base fabric is flattened to measure the primitive value of the state space Multiply by the After the eigenvalue, calculate the product with the first The weighted reduction between the eigenvalues ​​is the optimized Eigenvalue ,in, is the third weight hyperparameter, is the fourth weight hyperparameter; The optimization of the first step of combining the base fabric global flattened state metric order encoding vector The first eigenvalue of the base fabric global flattened state metric sequence encoding vector is obtained by using the eigenvalue to obtain an optimized base fabric global flattened state metric representation vector, wherein the first eigenvalue of the base fabric global flattened state metric sequence encoding vector remains unchanged; The optimized base fabric global flattened state metric representation vector is input into the classifier-based flattened state identifier to obtain the identification result.

[0050] In this way, in order to solve the problem that the feature set of the base fabric global flattened state metric representation vector has insufficient global cross-domain fine-grained aggregation representation capability due to the long distance exceeding the predetermined local distribution interval threshold under the predetermined eigenvalue sequence distribution, the high-dimensional feature space primitive representation of the base fabric global flattened state metric representation vector based on self-inner product fusion is used to capture the complex structure of the global network interaction of its eigenvalues, so as to reconstruct the cross-domain fine-grained aggregation relationship between the eigenvalues ​​of the base fabric global flattened state metric representation vector by simulating the scale-based high-dimensional feature space potential primitive, so as to realize the encoding reconstruction of the real sequence distribution behavior of the base fabric global flattened state metric representation vector under the long distance, improve the encoding expression effect of the base fabric global flattened state metric representation vector, and improve the accuracy of the recognition result obtained by inputting the flattened state identifier based on the classifier. In this way, it is possible to timely find the unevenness or wrinkles of the base fabric during the woven bag cutting process, and adjust the working parameters of the flattening negative pressure box accordingly to ensure that the woven bag finally cut has good quality and consistent size.

[0051] In the control method of the woven bag cutting device based on laser technology, the step S150 controls the laser cutting group to perform laser cutting on the flattened optimized woven base cloth to obtain the woven base cloth after laser cutting. It should be understood that in the technical solution of the present application, the subject architecture of the laser cutting group is an aluminum profile structure, and the laser emitter can be adjusted up and down, left and right, front and back through a screw adjustment mechanism to adjust products with different widths and different cutting positions, and can also be adjusted at different positions through a software system. Specifically, special-shaped graphic cutting is a method of focusing high-temperature penetration cutting on the surface of the fabric by using a radio frequency excited metal tube packaged CO2 laser. The graphic shape is cut into various graphic shapes on the fabric through a 3D galvanometer control system to achieve the purpose of special-shaped cutting.

[0052] In the control method of the woven bag cutting device based on laser technology, the step S160 controls the buffer transition roller to transmit the woven base fabric after laser cutting. It should be understood that after laser cutting, the fabric forms a state of easy-tear line connection (i.e., dotted line cutting), which means that the fabric is not completely cut off, but retains some tiny connection points. At this time, if the subsequent processing is directly carried out, these weak connection points may be easily broken due to the heavy back part of the fabric. To solve this problem, a buffer transition roller is introduced. Among them, the design of the buffer transition roller takes into account the bearing capacity problem. It only bears a small amount of weight, and the main load is borne by the active transmission rubber roller of the base fabric. In this way, even if the roll behind the fabric is heavy, it will not cause excessive pressure on the part that has just been cut, thereby helping the fabric to smoothly transition from one processing stage to the next stage, reducing sudden tension changes, and thus ensuring the integrity of the fabric structure. In this way, the fabric that has been cut with special-shaped graphics but has not been completely separated can be processed and protected to avoid the problem of breakage caused by excessive weight of the fabric roll. In particular, the laser cutting group is a buffer transition roller that delivers the cloth from the active conveying roller of the base cloth to the laser cutting group. It is worth mentioning that the buffer transition roller mainly serves to connect the cloth to the dotted line after laser cutting. When the cloth roll is too heavy, it is very easy to break. In the technical solution of this application, it is improved to a buffer transition roller, and the laser virtual cut only carries the weight of one transition roller. The main load is on the active conveying rubber roller of the base cloth, so the cloth roll will not break after cutting due to being too large.

[0053] In summary, the control method of the woven bag cutting device based on laser technology according to the embodiment of the present application is explained, which collects the image of the flattened state of the base fabric through monitoring, and introduces the image processing and analysis algorithm based on artificial intelligence and deep learning at the back end to analyze the image of the flattened state of the base fabric, so as to capture and depict the global flattened state characteristics of the base fabric, so as to identify and judge whether there are wrinkles on the flattened surface of the base fabric, thereby finding the unevenness or defects of the flattened base fabric, so as to adjust the working parameters of the flattening negative pressure box in real time, thereby improving the flatness of the fabric and ensuring that the woven bag finally cut out has good quality and consistent size. In this way, the dependence on manual intervention can be reduced, the degree of automation can be improved, and the production process can be made more stable and reliable, reducing the scrap rate caused by uneven materials.

[0054] In particular, in a specific example of the present application, a woven bag cutting device based on laser technology is also provided, and the woven bag cutting device based on laser technology is used to execute the control method of the woven bag cutting device based on laser technology.

[0055] Here, those skilled in the art can understand that the specific operations of each step in the control system of the above-mentioned woven bag cutting device based on laser technology have been referred to above. Figures 1 to 3 The control method of the woven bag cutting device based on laser technology has been introduced in detail, and therefore, its repeated description will be omitted.

[0056] As described above, the control system 100 of the woven bag cutting device based on laser technology according to the embodiment of the present application can be implemented in various terminal devices. In one example, the control system 100 of the woven bag cutting device based on laser technology can be integrated into the terminal device as a software module and / or a hardware module. For example, the control system 100 of the woven bag cutting device based on laser technology can be a software module in the operating system of the terminal device, or can be an application developed for the terminal device; of course, the control system 100 of the woven bag cutting device based on laser technology can also be one of the many hardware modules of the terminal device.

[0057] Alternatively, in another example, the control system 100 of the woven bag cutting device based on laser technology and the terminal device may also be separate devices, and the control system 100 of the woven bag cutting device based on laser technology may be connected to the terminal device via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0058] In summary, the control system of the woven bag cutting device based on laser technology according to the embodiment of the present application is explained, which monitors and collects the flattened state image of the base fabric, and introduces an image processing and analysis algorithm based on artificial intelligence and deep learning at the back end to analyze the flattened state image of the base fabric, so as to capture and depict the global flattened state characteristics of the base fabric, thereby identifying and judging whether there are wrinkles on the flattened surface of the base fabric, thereby finding the unevenness or defects of the flattened base fabric, and adjusting the working parameters of the flattening negative pressure box in real time, thereby improving the flatness of the fabric and ensuring that the woven bag finally cut out has good quality and consistent size. In this way, the reliance on manual intervention can be reduced, the degree of automation can be improved, and the production process can be made more stable and reliable, reducing the scrap rate caused by uneven materials.

Claims

1. A control method for a woven bag cutting device based on laser technology, characterized in that: include: Controlling the active traction motor to pull the woven base fabric placed on the base fabric delivery rack; The position of the woven base fabric is detected by a photoelectric sensor, and the EPC automatic deviation correction device is used to ensure the accurate positioning of the woven base fabric; Controlling the spiral transition roller to preliminarily flatten the woven base fabric to obtain a preliminarily flattened woven base fabric; Controlling the flattening negative pressure box to adsorb the preliminarily flattened woven base fabric to obtain a flattened optimized woven base fabric; Controlling the laser cutting group to perform laser cutting on the flattened optimized woven base fabric to obtain the laser-cut woven base fabric; The buffer transition roller is controlled to transmit the woven base fabric after the laser cutting.

2. The control method of the woven bag cutting device based on laser technology according to claim 1 is characterized in that: Controlling the flattening negative pressure box to adsorb the preliminarily flattened woven base fabric to obtain a flattened optimized woven base fabric, comprising: Acquire a flattened state image of the base fabric captured by an industrial camera; Grid segmentation is performed on the base fabric flattened state image to obtain a set of base fabric flattened grid granularity state images; Performing texture feature extraction and spatial feature enhancement processing on each base fabric flattened grid granularity state image in the set of base fabric flattened grid granularity state images to obtain a set of spatial enhancement feature maps of texture grid granularity after the base fabric is flattened; The flattened state consistency measurement coefficient between any two base fabric flattened texture grid granularity spatial enhancement feature maps in the set of the base fabric flattened texture grid granularity spatial enhancement feature maps is calculated, and the flattened state is identified based on multiple flattened state consistency measurement coefficients to determine whether there are wrinkles and whether to generate parameter adjustment prompts for the flattened negative pressure box.

3. The control method of the woven bag cutting device based on laser technology according to claim 2 is characterized in that: The texture feature extraction and spatial feature enhancement processing are respectively performed on each base fabric flattened mesh granularity state image in the set of base fabric flattened mesh granularity state images to obtain a set of spatial enhancement feature maps of texture mesh granularity after the base fabric is flattened, including: Inputting each base fabric flattened mesh granularity state image in the set of base fabric flattened mesh granularity state images into a texture feature extractor based on a multi-layer hole convolution structure to obtain a set of texture mesh granularity feature maps after the base fabric is flattened; The method comprises: performing spatial constraint feature enhancement processing on each base fabric flattened texture mesh granularity feature map in the set of base fabric flattened texture mesh granularity feature maps to obtain a set of base fabric flattened texture mesh granularity spatial enhanced feature maps, including: performing feature dispersion on the base fabric flattened texture mesh granularity feature map to obtain a set of base fabric flattened texture mesh pixel granularity feature vectors, performing semantic association constraint score calculation on the set of base fabric flattened texture mesh pixel granularity feature vectors to obtain a base fabric flattened texture mesh pixel granularity semantic association spatial constraint score topological representation; enhancing the base fabric flattened texture mesh granularity feature map based on the base fabric flattened texture mesh pixel granularity semantic association spatial constraint score topological representation to obtain a base fabric flattened texture mesh granularity spatial enhanced feature map.

4. The control method of the woven bag cutting device based on laser technology according to claim 3 is characterized in that: After the feature map of the texture grid granularity after the base fabric is flattened is feature scattered to obtain a set of feature vectors of the pixel granularity of the texture grid after the base fabric is flattened, the set of feature vectors of the pixel granularity of the texture grid after the base fabric is flattened is subjected to semantic association constraint score calculation to obtain a topological representation of the semantic association space constraint score of the pixel granularity of the texture grid after the base fabric is flattened, including: Performing feature dispersion on the texture grid granularity feature map after the base fabric is flattened along the channel dimension to obtain a set of pixel granularity feature vectors of the texture grid after the base fabric is flattened; Calculating the semantic association score between any two pixel granularity feature vectors of the texture grid after flattening the base fabric in the set of pixel granularity feature vectors of the texture grid after flattening the base fabric to obtain a topological matrix of semantic association scores of pixel granularity of the texture grid after flattening the base fabric; Based on the spatial distance between any two base fabric flattened texture grid pixel granularity feature vectors in the set of base fabric flattened texture grid pixel granularity feature vectors, the base fabric flattened texture grid pixel granularity semantic association score topology matrix is ​​subjected to spatial attention attenuation modulation to obtain the base fabric flattened texture grid pixel granularity semantic association spatial soft constraint score topology matrix; The soft constraint score topology matrix of the semantic association space of the pixel granularity of the texture grid after the base fabric is flattened is subjected to dilated convolution coding to obtain a soft constraint score topology feature matrix of the semantic association space of the pixel granularity of the texture grid after the base fabric is flattened as a topological representation of the semantic association space constraint score of the pixel granularity of the texture grid after the base fabric is flattened.

5. The control method of the woven bag cutting device based on laser technology according to claim 4 is characterized in that: The feature dispersion of the texture grid granularity feature map after the base fabric is flattened is performed along the channel dimension to obtain a set of pixel granularity feature vectors of the texture grid after the base fabric is flattened, including: Performing feature reshaping processing on the texture grid granularity feature map after the base fabric is flattened along the channel dimension to obtain a set of texture grid feature vectors after the base fabric is flattened; Each base fabric flattened texture grid feature vector in the set of base fabric flattened texture grid feature vectors is mapped to the Poincare space to obtain a set of base fabric flattened texture grid pixel granularity feature vectors.

6. The control method of the woven bag cutting device based on laser technology according to claim 5 is characterized in that: Calculating the semantic association score between any two base fabric flattened texture grid pixel granularity feature vectors in the set of base fabric flattened texture grid pixel granularity feature vectors to obtain a base fabric flattened texture grid pixel granularity semantic association score topological matrix, including: Calculate the first pixel granularity feature vector of the texture grid after the base fabric is flattened. The texture grid pixel granularity feature vector and the first The square of the position difference between the feature vectors of the pixel granularity of the texture grid after flattening the base cloth is obtained to obtain the semantic difference of the pixel granularity of the texture grid after flattening the base cloth. Calculate the constant 1 and the The difference between the squares of the norms of the pixel granularity feature vectors of the texture grid after the base cloth is flattened is used to obtain the semantic representation of the pixel granularity of the texture grid after the base cloth is flattened; Calculate the constant 1 and the The difference between the squares of the norms of the pixel granularity feature vectors of the texture grid after the base cloth is flattened is used to obtain the semantic representation of the pixel granularity of the texture grid after the base cloth is flattened; Calculate the The semantic representation of the pixel granularity of the texture grid after flattening the base cloth and the After multiplying the semantic representations of pixel granularity of the flattened texture grid of the base fabric, a semantic difference-norm representation of pixel granularity of the flattened texture grid of the base fabric is divided by the value after the multiplication to obtain the semantic association information representation of pixel granularity of the flattened texture grid of the base fabric; After calculating the sum of the double of the semantic association information of the pixel granularity of the texture grid after the base fabric is flattened and the constant 1, the obtained value is input into the inverse hyperbolic cosine function to obtain the semantic association score value of the pixel granularity of the texture grid after the base fabric is flattened; Arrange the plurality of pixel granularity semantic association score values ​​of the texture grid after the base fabric is flattened into a matrix form to obtain a topological matrix of the pixel granularity semantic association score of the texture grid after the base fabric is flattened.

7. The control method of the woven bag cutting device based on laser technology according to claim 6 is characterized in that: Based on the spatial distance between any two base fabric flattened texture grid pixel granularity feature vectors in the set of base fabric flattened texture grid pixel granularity feature vectors, the base fabric flattened texture grid pixel granularity semantic association score topology matrix is ​​subjected to spatial attention attenuation modulation to obtain the base fabric flattened texture grid pixel granularity semantic association spatial soft constraint score topology matrix, including: Calculate the The texture grid pixel granularity feature vector after flattening the base cloth and the The spatial distance between the pixel-granularity feature vectors of the texture grid after flattening the base cloth is used to obtain the pixel-granularity semantic space distance representation; Calculating a natural exponential function value with the pixel granularity semantic space distance represented as an exponential power and the natural constant e as the base to obtain a pixel granularity semantic space distance class support representation; After multiplying the pixel-granular semantic space distance class support representation by the spatial attention attenuation modulation hyperparameter to obtain the spatial distance attention attenuation representation, the pixel-granular semantic association score topology matrix of the texture grid after flattening the base cloth is The feature value of the position is divided by the spatial distance attention attenuation representation to obtain the texture grid pixel granularity semantic association space soft constraint score after the base cloth is flattened; Arrange the plurality of the soft constraint scores of the semantic association space of the pixel granularity of the texture grid after the base fabric is flattened into a matrix form to obtain a topological matrix of the soft constraint scores of the semantic association space of the pixel granularity of the texture grid after the base fabric is flattened.

8. The control method of the woven bag cutting device based on laser technology according to claim 7 is characterized in that: The method further comprises: enhancing the feature map of the texture grid granularity after the base fabric is flattened based on the semantic association space constraint score topological representation of the pixel granularity of the texture grid after the base fabric is flattened to obtain the spatial enhanced feature map of the texture grid granularity after the base fabric is flattened, including: Inputting the set of texture grid feature vectors after the base fabric is flattened and the soft constraint score topology matrix of the pixel granularity semantic association space of the texture grid after the base fabric is flattened into a graph convolutional coding module to obtain a set of texture grid context pixel granularity feature vectors after the base fabric is flattened; The set of the context pixel granularity feature vectors of the texture grid after the base fabric is flattened is reshaped to obtain a granularity spatial enhancement feature map of the texture grid after the base fabric is flattened.

9. The control method of the woven bag cutting device based on laser technology according to claim 8 is characterized in that: Calculating the flattened state consistency measurement coefficient between any two base fabric flattened texture grid granularity spatial enhancement feature maps in the set of the base fabric flattened texture grid granularity spatial enhancement feature maps, and performing flattened state recognition based on a plurality of the flattened state consistency measurement coefficients to determine whether wrinkles exist and whether to generate a parameter adjustment prompt for a flattened negative pressure box, including: Calculating the flattened state consistency measurement coefficient between any two base fabric flattened texture grid granularity spatial enhancement feature maps in the set of base fabric flattened texture grid granularity spatial enhancement feature maps to obtain a base fabric global flattened state measurement representation matrix composed of a plurality of flattened state consistency measurement coefficients; Inputting the base fabric global flattening state measurement representation matrix into a classifier-based flattening state identifier to obtain a recognition result, wherein the recognition result is used to indicate whether there are wrinkles; In response to the identification result that wrinkles exist, a parameter adjustment prompt for flattening the negative pressure box is generated.

10. A woven bag cutting device based on laser technology, characterized in that: The woven bag cutting device based on laser technology is used to execute the control method of the woven bag cutting device based on laser technology as described in any one of claims 1-9.

Images