Anti-deviation control method and system for groove placement of pin product conveyor belt

Abstract

The invention discloses an anti-deviation control method and system for groove placement of a pin product conveying belt, and relates to the technical field of conveying belt control. The method comprises the steps that a product clamping path is determined, and the clamping force is determined; motor current identification is carried out, and the pin product is clamped to the position above a groove of a product conveying belt; performing image acquisition by using a visual acquisition assembly to obtain a clamped image set, and performing lateral deviation identification to obtain lateral deviation control parameters; retrieving to obtain a downward movement force-speed mapping group set at the placement moment; and optimizing longitudinal control parameters of the product longitudinal moving device to obtain a target longitudinal control scheme, and performing anti-offset control. The technical problems that in the prior art, when the pin products are placed in the groove of the conveying belt, control is not accurate, the products are prone to being damaged, and precision and efficiency are low are solved, and the technical effects that the pin products are accurately controlled in the process of being placed in the groove of the conveying belt, product damage is avoided, and the placing precision and efficiency are improved are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of conveyor belt control, and particularly to an anti-offset control method and system for placing pin products in the grooves of a conveyor belt. Background Art

[0002] In modern industrial production, the transportation and placement of pin products, such as processors with pins, is a key link. In traditional technologies, when controlling the placement of pin products in the corresponding grooves of a conveyor belt, many technical problems are faced. On the one hand, due to the inability to accurately obtain the storage position of the pin products and the contour position of the conveyor belt grooves, it is difficult for the product longitudinal movement device to determine a precise product clamping path, which may cause situations such as collisions and drops during the product handling process, affecting production efficiency and product quality. On the other hand, during the process of placing the product into the conveyor belt groove, it is difficult to reasonably determine the clamping force of the product longitudinal movement device according to the actual situation of the product. If the clamping force is too large, it may cause damage to the product; if the clamping force is too small, the product may slip during the handling process. At the same time, the control of the force and speed during the vertical downward movement of the product is not precise enough, there are problems such as insufficient force control accuracy and force-displacement response delay, which easily lead to collisions between the product and the conveyor belt when approaching the groove, resulting in deformation of the pins of the product.

[0003] The prior art has technical problems such as inaccurate control when placing pin products in the conveyor belt groove, easy damage to the products, and low accuracy and efficiency. Summary of the Invention

[0004] The present application provides an anti-offset control method and system for placing pin products in the grooves of a conveyor belt, which are used to solve the technical problems in the prior art of inaccurate control when placing pin products in the conveyor belt groove, easy damage to the products, and low accuracy and efficiency.

[0005] In view of the above problems, the present application provides an anti-offset control method and system for placing pin products in the grooves of a conveyor belt.

[0006] In the first aspect of the present application, an anti-offset control method for placing pin products in the grooves of a conveyor belt is provided. The method includes:

[0007] Determine the product clamping path according to the storage position of the pin product and the contour position of the conveyor belt groove, and determine the clamping force based on the quality and clamping contact area of the pin product; identify the motor current based on the clamping force and the product clamping path, and control the motor to drive the product longitudinal movement device to clamp the pin product above the groove of the product conveyor belt according to the product clamping path; use the visual acquisition component to perform image acquisition to obtain a set of clamping images, and perform horizontal offset identification to obtain horizontal offset control parameters; retrieve the set of downward force-speed mapping groups at the placement moment; traverse the set of downward force-speed mapping groups at the placement moment for centralized analysis to determine the centralized interval of the downward force-speed mapping groups at the placement moment; use the centralized interval of the downward force-speed mapping groups at the placement moment as a constraint to optimize the longitudinal control parameters of the product longitudinal movement device to obtain a target longitudinal control scheme, and combine the horizontal offset control parameters to perform anti-offset control for placing the product in the conveyor belt groove.

[0008] In the second aspect of the present application, an anti-offset control system for placing a pin product in a conveyor belt groove is provided. The system includes:

[0009] A clamping force determination module, which is used to determine the product clamping path according to the storage position of the pin product and the contour position of the conveyor belt groove, and determine the clamping force based on the quality and clamping contact area of the pin product; a motor output current sequence acquisition module, which identifies the motor current based on the clamping force and the product clamping path, and controls the motor to drive the product longitudinal movement device to clamp the pin product above the groove of the product conveyor belt according to the product clamping path; a horizontal offset control parameter acquisition module, which is used to use the visual acquisition component to perform image acquisition to obtain a set of clamping images, and perform horizontal offset identification to obtain horizontal offset control parameters; a mapping group set acquisition module, which is used to retrieve the set of downward force-speed mapping groups at the placement moment; a centralized analysis module, which is used to traverse the set of downward force-speed mapping groups at the placement moment for centralized analysis to determine the centralized interval of the downward force-speed mapping groups at the placement moment; a target longitudinal control scheme acquisition module, which is used to use the centralized interval of the downward force-speed mapping groups at the placement moment as a constraint to optimize the longitudinal control parameters of the product longitudinal movement device to obtain a target longitudinal control scheme, and combine the horizontal offset control parameters to perform anti-offset control for placing the product in the conveyor belt groove.

[0010] One or more technical solutions provided in the present application have at least the following technical effects or advantages:

[0011] Determine the product clamping path based on the storage position of the pin product and the contour position of the conveyor belt groove, and determine the clamping force based on the quality and clamping contact area of the pin product; perform motor current identification, and control the motor to drive the product longitudinal movement device to clamp the pin product above the groove of the product conveyor belt according to the product clamping path; use the visual acquisition component to perform image acquisition to obtain a set of clamping images, and perform lateral offset identification to obtain lateral offset control parameters; traverse the set of downward force-speed mapping groups at the placement moment for centralized analysis to determine the centralized interval of the downward force-speed mapping groups at the placement moment; optimize the longitudinal control parameters of the product longitudinal movement device with the centralized interval of the downward force-speed mapping groups at the placement moment as the constraint to obtain the target longitudinal control scheme, and combine the lateral offset control parameters to perform anti-offset control for placing the product in the conveyor belt groove. It achieves the technical effects of realizing precise control during the placement of the pin product in the conveyor belt groove, avoiding product damage, and improving the placement accuracy and efficiency. Description of the Drawings

[0012] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0013] Figure 1 Schematic flowchart of the anti-offset control method for placing a pin product in a conveyor belt groove provided by an embodiment of the present application;

[0014] Figure 2 Schematic structural diagram of the anti-offset control system for placing a pin product in a conveyor belt groove provided by an embodiment of the present application.

[0015] Explanation of reference numerals: Clamping force determination module 10, motor output current sequence acquisition module 20, lateral offset control parameter acquisition module 30, mapping group set acquisition module 40, centralized analysis module 50, target longitudinal control scheme acquisition module 60. Detailed Embodiments

[0016] The present application provides an anti-offset control method and system for placing a pin product in a conveyor belt groove, which is used to solve the technical problems of inaccurate control, easy product damage, low accuracy and efficiency when placing a pin product in a conveyor belt groove in the prior art.

[0017] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application.

[0018] Embodiment 1, as Figure 1 shown, the present application provides an anti-offset control method for placing pin products in the conveyor belt grooves, and the method includes:

[0019] Step S100: Determine the product clamping path according to the storage position of the pin product and the contour position of the conveyor belt groove, and determine the clamping force based on the quality and clamping contact area of the pin product.

[0020] Specifically, first in the data acquisition link, the multi-modal sensing fusion technology is used, and an industrial-grade vision camera with high frame rate and high resolution is deployed, combined with a high-precision lidar and an ultrasonic sensor array to synchronously capture the storage position of the pin product and the contour position of the conveyor belt groove. The vision camera is based on the deep learning object detection and positioning algorithm to accurately analyze the image pixel information, obtain the coordinates (x, y, z) of the pin product in the three-dimensional space and the attitude angle information, with an accuracy of sub-millimeter level; the lidar emits laser beams and receives the reflected light, and uses the triangulation method to construct a point cloud model of the conveyor belt groove contour, accurately measuring the three-dimensional geometric parameters of the groove. The ultrasonic sensor assists in calibrating the close-range details, filling the data blind area, and ensuring the integrity and accuracy of the position data. Then it enters the stage of constructing and training the path planning model based on the convolutional neural network (CNN). For the large-scale sample set collected in the early stage, each sample contains the storage position coordinate group of the pin product, the contour position parameter set of the conveyor belt groove, and the corresponding accurate product clamping path coordinate sequence. The constructed CNN model has the first convolutional layer configured with N1 convolutional kernels with a size of k1×k1 and a stride of S1 to perform primary feature extraction on the position data and generate a feature map; after being reduced in dimension by the maximum pooling layer with a size of p1×p1, it is passed to the subsequent multi-layer convolutional and pooling structures, and the size of the deep convolutional kernels decreases gradually, gradually abstracting the features; the end fully connected layer integrates the global features. According to the supervised learning mechanism, the mean square error loss function is used to measure the deviation between the predicted clamping path and the real path, and the random gradient descent method and the adaptive learning rate strategy are used to iteratively update the model weight parameters. After several rounds of training, until the accuracy of the model on the validation set exceeds 95%, the path recognizer training is completed. In the real-time operation stage, the newly collected pin product and groove position data are input into the trained path recognizer after being preprocessed by standardization, and a precise product clamping path presented in the form of a coordinate point sequence is quickly generated.

[0021] Determine the clamping force of the product longitudinal movement device according to the formula It is determined as follows. First, calculate the gravity F_gravity of the pin product. Given that the mass of the pin product is m, according to the gravity calculation formula F_gravity = m×g, where g is the acceleration due to gravity. The safety factor k usually takes values between 1.2 and 1.5. However, considering the influence of the clamping contact area, its value is more precise. When the clamping contact area is large, the friction force distribution between the product and the fixture is more uniform, and the product is less likely to slip during the clamping process. For example, if the clamping contact area is 1.5 times or more of the standard contact area, the safety factor k can be set to 1.2. On the contrary, when the clamping contact area is small, the friction force distribution is relatively uneven, and the risk of product slipping increases. For example, if the clamping contact area is 0.5 times or less of the standard contact area, the safety factor k can be set to 1.5. For the clamping contact area between the two, the value of the safety factor k can be determined by linear interpolation or other appropriate functional relationships. Determine the dynamic force F_dynamic, which depends on the motion states such as the acceleration of the product longitudinal movement device during the process of clamping and moving the product. If the product longitudinal movement device is moving at a constant speed, F_dynamic may be 0. However, if it is in the process of acceleration or deceleration, the dynamic force needs to be calculated according to Newton's second law F = ma, where a is the acceleration and m is the product mass. Finally, through determine the clamping force of the product longitudinal movement device.

[0022] Step S200: Identify the motor current based on the clamping force and the product clamping path, and control the motor to drive the product longitudinal movement device to clamp the pin product above the groove of the product conveyor belt according to the product clamping path according to the identification result.

[0023] Specifically, the known clamping force is determined by comprehensively considering factors such as product quality, safety factor, friction coefficient, and dynamic force. Regarding the product clamping path, it is regarded as a trajectory composed of a series of discrete points. Calculate the distance between adjacent points, and obtain the speed by the ratio of the distance to the time interval (assuming the time interval for movement between two points is known). Then calculate the acceleration according to the ratio of the change in speed to the time interval. According to the dynamics formula, the force required at each point on the path can be obtained from the product quality and acceleration. Combining with the lever arm length related to the structure of the product longitudinal movement device, calculate the torque by the product of the force and the lever arm. The motor has its own torque-current characteristic curve, which reflects the corresponding relationship between torque and current, usually obtained through experimental determination. Substitute the torque values calculated at each point previously into the torque-current characteristic curve equation to obtain the motor output current values corresponding to each path point, and these current values form the motor output current sequence. After obtaining the motor output current sequence, input each current value in the sequence into the motor controller in turn. The motor controller controls the input voltage of the motor by using pulse width modulation technology according to the input current value. Control the speed and rotation direction of the motor by adjusting the input voltage, so that the motor rotates in the desired manner. The rotating motor generates corresponding power, and transmits the power to the product longitudinal movement device through the connection structure with the product longitudinal movement device. Driven by the motor, the product longitudinal movement device moves along the pre-determined product clamping path, and finally clamps the pin product above the groove of the product conveyor belt.

[0024] Step S300: Use the vision acquisition component to perform image acquisition to obtain a set of clamping images, and perform horizontal offset recognition to obtain horizontal offset control parameters.

[0025] Specifically, an image acquisition is carried out by using a visual acquisition component, which is usually composed of devices such as a camera. Its installation position and angle are carefully designed to clearly obtain an image containing the pin product and the groove of the product conveyor belt. During the process that the pin product is clamped by the product longitudinal movement device to above the conveyor belt groove, the visual acquisition component continuously works and acquires images at a certain frequency, thereby obtaining a series of images, which together constitute a set of clamping images. Then, a horizontal offset recognition is performed on the set of clamping images, which is a process involving image processing and analysis. First, it is necessary to preprocess the image, such as grayscale conversion, filtering and other operations, to improve the quality and clarity of the image for subsequent analysis. Then, through an edge detection algorithm, such as the Canny edge detection algorithm, the edges of the product and the groove in the image are detected to obtain a clear edge contour. On this basis, by calculating the relative position relationship between the product edge and the groove edge in the horizontal direction, it is determined whether the product has a horizontal offset. After the above analysis and calculation, the offset amount and the offset direction are finally determined. The offset amount refers to the distance that the product deviates from the center of the groove in the horizontal direction, and is obtained by calculating the coordinate difference between the product edge and the center of the groove in the horizontal direction. The offset direction refers to whether the product deviates to the left or right relative to the center of the groove. If the product edge is on the left side of the center of the groove, the offset direction is to the left; if the product edge is on the right side of the center of the groove, the offset direction is to the right. Finally, the obtained offset amount and offset direction are used as horizontal offset control parameters. These parameters will be used for subsequent control operations, such as by adjusting the horizontal movement of the product longitudinal movement device to enable the product to be accurately placed at the center position of the conveyor belt groove, thereby realizing the anti-offset control of the product placement in the conveyor belt groove.

[0026] Step S400: Retrieve and obtain the set of downward force-speed mapping groups at the placement moment.

[0027] Specifically, it is necessary to obtain the historical qualified placement data set of pin products, and the formation of this set is based on a large number of past production practice records. During the production process, each time a pin product is successfully placed in the conveyor belt groove, relevant data will be recorded, including various parameters of the product itself, relevant characteristics of the conveyor belt, operating conditions during the placement process, and information such as the final placement result. These data are systematically sorted and stored to form the historical qualified placement data set of pin products. Next, determine the downward force - speed mapping group at the placement moment as the index. During the product placement process, the downward force and speed are two key operating parameters. The downward force - speed mapping group at the placement moment records the corresponding relationship between the force value and the corresponding speed value used for the product to move downward at each specific moment. This corresponding relationship is of great significance for understanding and optimizing the product placement process because different combinations of force and speed will result in different placement effects. Then, use the determined index to retrieve the historical qualified placement data set of pin products. During the retrieval process, traverse the entire historical data set to find those data records that meet the conditions of the downward force - speed mapping group at the placement moment. For each data record, it will be checked whether the included downward force and speed mapping group match the index. If the match is successful, the data record will be extracted as the qualified data. After the retrieval process, finally obtain the set of downward force - speed mapping groups at the placement moment. Each element in this set is selected from the historical data set and they all meet the index conditions of the downward force - speed mapping group at the placement moment. This set will serve as an important data basis for subsequent analysis and optimization steps, providing a basis for further determining reasonable product placement parameters and strategies.

[0028] Step S500: Traverse the set of downward force - speed mapping groups at the placement moment for centralized analysis to determine the centralized interval of the downward force - speed mapping groups at the placement moment.

[0029] Specifically, a two-dimensional tolerance threshold analysis space is constructed based on the set of downward force-velocity mapping groups at the placement moment, so as to obtain a set of analysis points, where each analysis point corresponds to a downward force-velocity mapping group at the placement moment. Then, multiple analysis points are randomly extracted from the set of analysis points, and neighborhoods are constructed for these points according to a preset iteration bandwidth, obtaining multiple analysis point neighborhoods. Then, these neighborhoods are traversed to calculate the neighborhood density, and the analysis point corresponding to the maximum neighborhood density is determined as the direction analysis point, and the remaining analysis points are used as following analysis points. Taking the direction of the direction analysis point as the iteration direction, the following analysis points are iterated according to the preset iteration bandwidth to obtain multiple following iterative analysis points, and then neighborhoods are constructed for them and it is judged whether the maximum neighborhood density of their neighborhoods is greater than or equal to the neighborhood density of the direction analysis point. If so, the direction analysis point is updated. This iterative process is repeated until the preset maximum number of iterations is satisfied. Finally, the neighborhood corresponding to the direction analysis point obtained in the last iteration is used as the concentrated interval in the set of downward force-velocity mapping groups at the placement moment.

[0030] Step S600: With the concentrated interval in the set of downward force-velocity mapping groups at the placement moment as a constraint, perform optimization of the longitudinal control parameters of the product longitudinal movement device to obtain a target longitudinal control scheme, and combine the lateral offset control parameters to perform anti-offset control for the placement of the product conveyor belt groove.

[0031] Specifically, perform longitudinal distance feature recognition on the set of clamping images to determine the vertical distance between the pin product and the plane where the groove of the product conveyor belt is located as the longitudinal distance. Then, obtain the preset downward movement time, and combine the longitudinal distance to determine the initial downward force-velocity mapping group. Then, with the concentrated interval in the set of downward force-velocity mapping groups at the placement moment as a constraint, use the scheme identifier to identify the initial downward force-velocity mapping group and the longitudinal distance, obtaining multiple longitudinal control schemes. Then, these schemes are screened, and with the concentrated interval as the standard, multiple preliminarily screened longitudinal control schemes are obtained. Then, obtain the preliminarily screened downward force-velocity mapping groups at the placement moment of these preliminarily screened schemes, and perform control scheme fitness analysis based on them to obtain multiple preliminarily screened fitness values. Finally, the preliminarily screened longitudinal control scheme corresponding to the maximum value among the multiple preliminarily screened fitness values is determined as the target longitudinal control scheme.

[0032] Based on the previously obtained lateral offset control parameters and the target longitudinal control scheme, anti-offset control for placing pin products in the grooves of the product conveyor belt is achieved. First, using the offset amount and offset direction information in the lateral offset control parameters, the offset amount clarifies the distance by which the product deviates from the center of the groove in the lateral direction, and the offset direction indicates whether it is offset to the left or right. According to this information, the position of the product longitudinal movement device in the lateral direction is adjusted so that the product can move towards the center of the groove, thereby correcting the lateral offset. At the same time, the placement of the product in the longitudinal direction is controlled according to the target longitudinal control scheme. The target longitudinal control scheme has considered factors such as the vertical distance between the product and the plane where the conveyor belt groove is located, the preset downward movement time, and reasonable downward movement force and speed. According to this scheme, the downward movement process of the product in the longitudinal direction is precisely controlled to ensure that the product can be stably placed in the appropriate position within the groove, avoiding offset or improper placement in the longitudinal direction. Through the coordinated control of the lateral and longitudinal directions, anti-offset control for placing pin products in the grooves of the product conveyor belt is ultimately achieved, improving the accuracy and stability of product placement.

[0033] In a possible implementation manner, step S500 further includes:

[0034] Step S510: Based on the set of downward force-speed mapping groups at the placement moment, construct a two-dimensional tolerance threshold analysis space to obtain a set of analysis points, where each analysis point corresponds to a downward force-speed mapping group at the placement moment.

[0035] Step S520: Randomly extract multiple analysis points from the set of analysis points, and construct neighborhoods for the multiple analysis points according to a preset iteration bandwidth to obtain multiple analysis point neighborhoods.

[0036] Step S530: Traverse the multiple analysis point neighborhoods to calculate the neighborhood densities, obtain multiple analysis point neighborhood densities, take the analysis point corresponding to the maximum value of the multiple analysis point neighborhood densities as the direction analysis point, and take the remaining multiple analysis points as multiple following analysis points.

[0037] Step S540: Use the direction of the direction analysis point as the iteration direction, and perform iteration on the multiple following analysis points according to the preset iteration bandwidth to obtain multiple following iteration analysis points.

[0038] Step S550: Construct multiple neighborhoods of the multiple following iterative analysis points, and determine whether the maximum neighborhood density of the multiple neighborhoods of the following iterative analysis points is greater than or equal to the analysis point neighborhood density of the direction analysis point. If so, update the following iterative analysis point corresponding to the maximum neighborhood density of the multiple neighborhoods of the following iterative analysis points to the direction analysis point, and continue the iteration until the preset maximum number of iterations is satisfied. Take the neighborhood corresponding to the direction analysis point obtained in the last iteration as the interval of the placement time downward force-velocity mapping group set.

[0039] Specifically, construct a two-dimensional tolerance threshold analysis space based on the placement time downward force-velocity mapping group set. This space is a virtual environment for analyzing and processing data, and its two dimensions correspond to the downward force and velocity respectively. In this space, each placement time downward force-velocity mapping group is regarded as a data point. By processing all the mapping groups in the set and corresponding them one by one to this two-dimensional space, an analysis point set is obtained, where each analysis point exactly corresponds to a placement time downward force-velocity mapping group.

[0040] Randomly select multiple analysis points from the constructed analysis point set, and these analysis points correspond to the placement time downward force-velocity mapping groups. Then, for each selected analysis point, construct its neighborhood according to the preset iteration bandwidth. The preset iteration bandwidth determines the range of the neighborhood. Centered on the analysis point, within the two-dimensional tolerance threshold analysis space, an area containing several other potential analysis points is determined according to this bandwidth. Do such an operation for each selected analysis point, and finally obtain multiple analysis point neighborhoods.

[0041] Calculate the neighborhood density for the obtained multiple analysis point neighborhoods. Neighborhood density is an index to measure the distribution density of analysis points within a neighborhood. The specific calculation method is to count the number of analysis points in each analysis point neighborhood respectively, and divide the statistical result by the area of the corresponding analysis point neighborhood, so as to obtain the density value of each analysis point neighborhood, that is, the multiple analysis point neighborhood densities. By comparing these neighborhood density values, take the analysis point corresponding to the maximum neighborhood density as the direction analysis point, because the neighborhood where this analysis point is located has the highest analysis point density, representing a trend of the data, and take the remaining multiple analysis points as multiple following analysis points.

[0042] Analyze the direction of the point in a determined direction as the iterative direction, which is based on the data trend direction indicated by the density of analysis points in the neighborhood where the analysis point is located. Then, for multiple following analysis points, perform iterative operations according to a preset iterative bandwidth. That is, in a two-dimensional tolerance threshold analysis space, starting from each following analysis point, move a distance of the preset iterative bandwidth along the determined iterative direction to obtain new position points, and these new position points are the multiple following iterative analysis points.

[0043] Construct respective neighborhoods for the multiple following iterative analysis points, that is, the neighborhoods of the multiple following iterative analysis points. The way to construct the neighborhood is similar to before. In the two-dimensional tolerance threshold analysis space, with each following iterative analysis point as the center, determine its neighborhood range according to a preset rule. Then, calculate the neighborhood density of these multiple neighborhoods of the following iterative analysis points, and find the maximum neighborhood density. Then compare this maximum value with the analysis point neighborhood density of the direction analysis point. If the maximum neighborhood density of the neighborhoods of the multiple following iterative analysis points is greater than or equal to the analysis point neighborhood density of the direction analysis point, then update the following iterative analysis point corresponding to this maximum neighborhood density as the direction analysis point. This means that the trend of the data may have changed and the direction needs to be re-determined. After that, continue the iterative operation. Each iteration repeats the above processes of constructing the neighborhood, calculating the density, comparing and judging, and updating the direction analysis point until the preset maximum number of iterations is reached. When the preset maximum number of iterations is reached, use the neighborhood corresponding to the direction analysis point obtained in the last iteration as the placement moment downward force-velocity mapping group concentration interval. This concentration interval is determined after multiple iterations and analyses. It represents a relatively concentrated area of the placement moment downward force-velocity mapping group in the data space and is of great significance for subsequent control parameter optimization.

[0044] In a possible implementation manner, step S530 further includes:

[0045] Step S531: Count the number of analysis points in each of the multiple analysis point neighborhoods respectively, and divide the statistical result by the area of the corresponding multiple analysis point neighborhoods to obtain the densities of the multiple analysis point neighborhoods.

[0046] Specifically, for each neighborhood of the analysis point, detailed statistical operations need to be carried out. First, it is necessary to carefully determine the number of analysis points in each neighborhood of the analysis point. This requires a one-by-one check of the area covered by each neighborhood in the analysis space to accurately identify and count the analysis points therein. Then, calculate the areas of the corresponding multiple neighborhoods of the analysis points, which depends on the shape of the neighborhood in the analysis space and related preset parameters. For example, if the neighborhood is set to a regular geometric shape (such as a circle or a square), the area can be calculated according to the corresponding geometric formula. Finally, divide the number of analysis points obtained statistically by the corresponding neighborhood area to obtain the density value of each neighborhood of the analysis point. By such a calculation method, all neighborhoods of the analysis points are operated on, and finally, the densities of multiple neighborhoods of the analysis points are obtained. These density values will provide an important basis for subsequent determination of the direction analysis points and following analysis points.

[0047] In a possible implementation manner, step S600 further includes:

[0048] Step S610: Identify the longitudinal distance feature of the set of clamping images to obtain the longitudinal distance, where the longitudinal distance is the vertical distance between the pin product and the plane where the groove of the product conveyor belt is located.

[0049] Step S620: Obtain the preset downward movement time, and combine the longitudinal distance to determine the initial downward force-velocity mapping group.

[0050] Step S630: With the concentrated interval of the downward force-velocity mapping group at the placement moment as the constraint, optimize the longitudinal control scheme of the product longitudinal movement device according to the longitudinal distance and the initial downward force-velocity mapping group to obtain the target longitudinal control scheme.

[0051] Specifically, for the identification of the longitudinal distance feature of the set of clamping images, first, image processing techniques and algorithms need to be used. These techniques and algorithms can analyze and extract features of each element in the image. For each image in the set of clamping images, the relevant information of the pin product and the plane where the groove of the product conveyor belt is located needs to be carefully distinguished. By processing this information, the distance between the pin product and the plane where the conveyor belt groove is located in the vertical direction can be determined, and this distance is the required longitudinal distance. The accurate acquisition of this longitudinal distance is crucial for subsequent determination of control parameters and accurate placement of the product in the conveyor belt groove.

[0052] Obtain the preset downward movement time, which is a time parameter preset based on the technological requirements of product placement and past experience. It reflects the expected time for the product to fall from the initial position to the appropriate position in the conveyor belt groove under ideal conditions. Then, combine the obtained preset downward movement time with the obtained longitudinal distance. When determining the initial downward movement force-velocity mapping group, multiple physical factors need to be considered. According to the principles of dynamics, the falling process of the product is affected by various factors such as gravity, mass, and air resistance. In this case, the longitudinal distance and the preset downward movement time can be used as key constraints. Through Newton's second law (F = ma) and kinematic formulas, comprehensively consider factors such as the mass of the product, the longitudinal distance, and the preset downward movement time, and calculate the force required for the product to move downward in the initial stage and the corresponding velocity. There is a certain correspondence between these forces and velocities, and they jointly constitute the initial downward movement force-velocity mapping group, which provides important basic data for optimizing the longitudinal control scheme of the subsequent product longitudinal movement device.

[0053] The concentrated interval of the downward force - speed mapping group at the placement moment is obtained by deeply analyzing and processing the historical qualified placement data set of pin products. This concentrated interval determines the reasonable value ranges of the downward force and speed when successfully placing products in the past, providing key constraint conditions for optimizing the longitudinal control scheme of the current product's longitudinal movement device. Use the scheme identifier to identify the longitudinal distance and the initial downward force - speed mapping group, generating multiple possible longitudinal control schemes. These schemes will consider different combinations of force and speed to meet the placement requirements of the product in the longitudinal direction. With the concentrated interval of the downward force - speed mapping group at the placement moment as a constraint, screen the multiple generated longitudinal control schemes. Only the schemes that meet the requirements of the concentrated interval will be retained, that is, the downward force and speed values in the scheme must be within the reasonable range defined by the concentrated interval, which can exclude some obviously unreasonable schemes and narrow the scope of optimization. Obtain the initially screened downward force - speed mapping groups at the placement moment corresponding to the multiple longitudinal control schemes after preliminary screening, and conduct fitness analysis of the control schemes based on these mapping groups. For example, conduct frequent item mining on the vibration data of the pin product in the groove placement to obtain multiple initially screened placement amplitude sets and multiple initially screened placement vibration time sets, and perform weighted calculations on these sets to determine the fitness. Consider factors such as the accuracy and stability of product placement. For example, if the product vibrates greatly or the placement time is too long when placing the product, its fitness will be lower. Determine the target longitudinal control scheme as the initially screened longitudinal control scheme corresponding to the maximum value among the multiple initially screened fitness values. This scheme can ensure the accuracy and stability of product placement to the greatest extent while meeting the constraints of the concentrated interval, meeting the requirements for accurate placement of the product in the conveyor belt groove. Through the above process, comprehensively considering the constraints of the concentrated interval and the longitudinal distance and the initial downward force - speed mapping group, the target longitudinal control scheme is finally obtained, providing an effective control strategy for accurate placement of the product.

[0054] In a possible implementation manner, step S630 further includes:

[0055] Step S631: Use the scheme identifier to identify the initial downward force - speed mapping group and the longitudinal distance, obtaining multiple longitudinal control schemes.

[0056] Step S632: With the concentrated interval of the downward force - speed mapping group at the placement moment as a constraint, screen the multiple longitudinal control schemes to obtain multiple initially screened longitudinal control schemes.

[0057] Step S633: Obtain the multiple initially screened downward force - speed mapping groups at the placement moment of the multiple initially screened longitudinal control schemes, and conduct fitness analysis of the control schemes based on the multiple initially screened downward force - speed mapping groups at the placement moment, obtaining multiple initially screened fitness values.

[0058] Step S634: Use the preliminary screening longitudinal control scheme corresponding to the maximum value among the multiple preliminary screening fitness values as the target longitudinal control scheme.

[0059] Specifically, the scheme recognizer is constructed using a neural network algorithm. First, a large amount of historical data containing different longitudinal distances, downward movement forces and speeds, and corresponding product placement results is collected, and data cleaning and normalization preprocessing are performed. Then, a multi-layer perceptron structure including an input layer (corresponding to 3 nodes for longitudinal distance, force, and speed), two hidden layers with 5 nodes each layer, and an output layer is designed. The preprocessed data is divided into a training set and a test set. During training, the input layer receives the data, which is calculated and transmitted through the hidden layer according to the ReLU function, and a prediction result is obtained at the output layer. The mean square error loss function is calculated by comparing with the actual result, and the weights and biases are adjusted using the backpropagation algorithm until the stop condition is met. Finally, the model is evaluated using the test set. If it meets the requirements, when a new longitudinal distance and downward movement force-speed mapping group are input, multiple longitudinal control schemes can be generated at the output layer. The scheme recognizer receives two key input information: the initial downward movement force-speed mapping group and the longitudinal distance. The initial downward movement force-speed mapping group describes the corresponding relationship between the force and speed during the initial downward movement of the product, and the longitudinal distance specifies the vertical distance between the pin product and the plane of the groove of the product conveyor belt. Through comprehensive analysis and calculation of various possible situations, the scheme recognizer finally outputs multiple longitudinal control schemes, each of which represents a possible control strategy and provides a basis for subsequent screening and optimization.

[0060] Using the concentrated interval of the downward movement force-speed mapping group at the placement moment as the key constraint condition, a screening operation is carried out on multiple longitudinal control schemes. The concentrated interval is determined through in-depth analysis of historical data, and it defines the reasonable value range of the downward movement force and speed during successful product placement in the past. For each longitudinal control scheme, carefully check whether the involved downward movement force and speed values are within this concentrated interval. Only those schemes whose force and speed values fully meet the requirements of the concentrated interval will be screened out, and finally multiple preliminary screening longitudinal control schemes are obtained. These preliminary screening schemes are consistent with historical experience in terms of the values of force and speed and are more likely to achieve accurate placement of the product in the groove of the conveyor belt.

[0061] Obtain multiple initial screening vertical control scheme - corresponding multiple initial screening placement time downward force - speed mapping groups, which record the specific values of the downward force and speed involved at the placement time for each initial screening vertical control scheme. Then, based on these mapping groups, conduct a fitness analysis of the control schemes. This analysis process comprehensively considers multiple factors, such as the accuracy and stability of product placement, etc. By performing frequent item mining on the vibration data of the pin product groove placement, obtain multiple initial screening placement amplitude sets and multiple initial screening placement vibration time sets. Then, perform weighted calculations on these sets to determine the fitness of each initial screening vertical control scheme. For example, if a scheme causes a large vibration amplitude or a long placement time when placing the product, then its score in the weighted calculation may be lower, that is, its fitness is lower. In this way, analyze all the initial screening vertical control schemes and finally obtain multiple initial screening fitness values.

[0062] Compare all the initial screening fitness values and find the maximum value among them. Then, search for the initial screening vertical control scheme corresponding to this maximum value. This initial screening vertical control scheme, on the premise of meeting the interval constraints of the placement time downward force - speed mapping group set, through the fitness analysis of multiple initial screening vertical control schemes, performs optimally in terms of the accuracy and stability of product placement, etc. Therefore, determine it as the target vertical control scheme, which can better achieve the accurate placement of the product in the conveyor belt groove.

[0063] In a possible implementation manner, step S633 further includes:

[0064] Step S6331: Based on the multiple initial screening placement time downward force - speed mapping groups, conduct frequent item mining on the vibration data of the pin product groove placement to obtain multiple initial screening placement amplitude sets and multiple initial screening placement vibration time sets.

[0065] Step S6332: Perform weighted calculations on the multiple initial screening placement amplitude sets and the multiple initial screening placement vibration time sets to obtain the multiple initial screening fitness values.

[0066] Specifically, these mapping group data are connected to a professional big data analysis platform, and the high-performance data processing engine built into the platform is used to parse the pin product groove placement vibration data corresponding to each mapping group with a time accuracy of milliseconds. An improved frequent item mining algorithm based on the Apriori algorithm is adopted. These algorithms are customized and tuned to accurately adapt to the characteristics of vibration data in industrial production scenarios. Regarding the time series characteristics of vibration data, a sliding window technique is introduced to count the frequency of vibration events triggered by different downward force-velocity combinations within the dynamic window range, focusing on the two key indicators of amplitude and vibration time. When mining frequent amplitude items, the algorithm discretizes continuous amplitude values at intervals of, for example, 0.1 mm according to the preset amplitude resolution, and then conducts frequency statistics. The amplitude intervals that reach the set frequency threshold are clustered to construct multiple preliminary screening placement amplitude sets. For the vibration time, a time series analysis algorithm is also used, combined with the time benchmark required for normal product placement, with a precision unit of 0.01 s, to identify the frequently occurring vibration duration ranges and generate multiple preliminary screening placement vibration time sets. The entire process makes full use of parallel computing technology and invokes cluster computing resources to ensure the efficient and accurate completion of the frequent item mining task under massive vibration data and complex computing requirements, providing solid data support for subsequent fitness evaluation.

[0067] According to the process characteristics of product placement and the key points of quality control, the weight coefficients corresponding to amplitude and vibration time are scientifically and reasonably set. For example, if the current production link has extremely high requirements for the accuracy of product placement and even a tiny amplitude deviation may cause subsequent assembly failures, the weight coefficient of amplitude will be significantly increased; conversely, if the production rhythm is tight and more emphasis is placed on the rapid placement of products to improve overall efficiency, the weight of vibration time will dominate. For each preliminary screening placement amplitude set, the amplitude data elements are extracted one by one, each amplitude value is multiplied by the preset amplitude weight coefficient, and the results of these products are accumulated to obtain a value representing the performance of this preliminary screening scheme in the amplitude dimension. At the same time, the same operation is carried out on multiple preliminary screening placement vibration time sets. The vibration time data is accurately captured from each set, multiplied by the corresponding vibration time weight coefficient, and then accumulated to obtain a value reflecting the characteristics of vibration time. Finally, the weighted values obtained by each preliminary screening scheme in the amplitude and vibration time dimensions are added together to determine the preliminary screening fitness value for each preliminary screening longitudinal control scheme. These preliminary screening fitness values accurately quantify the comprehensive effectiveness of each scheme in balancing placement stability and operation timeliness.

[0068] In a possible implementation manner, step S100 further includes:

[0069] Step S110: Obtain the storage positions of multiple sample pin products, the contour positions of multiple sample conveyor belt grooves, and the clamping paths of multiple sample products as training data.

[0070] Step S120: Use the training data to perform supervised training on the framework constructed based on the convolutional neural network, learn the one-to-one mapping relationship between the storage positions of the pin products and the contour positions of the conveyor belt grooves and the product clamping paths, until the training converges, and obtain a trained path recognizer.

[0071] Step S130: Based on the path recognizer, identify the storage positions of the pin products and the contour positions of the conveyor belt grooves, and obtain the product clamping path, where the product clamping path is the movement path for the product longitudinal movement device to clamp the pin product from the storage position of the pin product to above the contour position of the conveyor belt groove.

[0072] Specifically, for the acquisition of the storage positions of the sample pin products, a high-precision positioning sensor network is deployed in the warehousing area. These sensors use multi-source fusion technologies such as laser ranging and visual recognition. They can not only accurately locate the three-dimensional coordinates of the products on the shelves, but also precisely capture their placement angles, locking them from all directions of the x, y, and z axes to ensure that the storage positions of each pin product are recorded without deviation, accumulating a large amount of sample position information under different layouts. At the same time, for the determination of the contour positions of the sample conveyor belt grooves, professional contour scanning equipment is enabled to scan section by section along the running track of the conveyor belt. Using the principle of structured light 3D scanning or high-precision line laser scanning technology, a contour model with millimeter-level accuracy is generated, covering the length, width, and height dimensions of the grooves, the edge radian, and the internal fine structural features. The conveyor belt grooves under different working conditions are scanned repeatedly to collect a rich variety of contour samples. In terms of the acquisition of the product clamping paths, the robotic arm motion control system plays a key role. When performing the operation of clamping the sample product, the built-in encoder and motion tracking module of the system record the parameter changes such as the rotation angle, extension length, and movement speed of the robotic arm joints in real time, integrating them into a complete motion trajectory data, accurately depicting the whole process path from the initial position to clamping the product and then placing it at the target point. Through repeated clamping experiments, a large number of sample product clamping paths under different starting conditions, product distributions, and task requirements are accumulated. Finally, these three types of key data are organically integrated to form the training data basis for subsequent training.

[0073] Based on the powerful feature extraction and pattern recognition capabilities of the convolutional neural network, a framework is constructed. A large amount of collected training data is input into it for supervised training. During the training process, the convolutional neural network uses multiple convolutional kernels to progressively extract the image features of the storage positions of pin products and the contour position features of the conveyor belt grooves, continuously adjusting the weights of internal neurons and fully learning the unique one-to-two mapping rules between these two and the product clamping path. Each round of training iteration is evaluated and optimized according to the preset loss function, and the network parameters are efficiently corrected using the backpropagation algorithm. This process continues until the model performance indicators are stable and the loss function converges to a minimum value, announcing the successful acquisition of a path recognizer that has completed training and has accurate recognition capabilities.

[0074] The entire process enters the actual application stage. Relying on the trained path recognizer to carry out key recognition work, first, using high-precision vision acquisition devices and position sensors, quickly and accurately capture the storage position information of pin products in the current actual scenario. Whether it is the three-dimensional coordinates of the product on the shelf or the attitude details such as its placement angle, etc., are all completely and accurately recorded; at the same time, advanced contour scanning technology is used to synchronously obtain the contour position of the conveyor belt groove, not missing any subtle contour changes, including key geometric information such as the length, width, and height dimensions, arc, and internal structure characteristics of the groove. Immediately afterwards, the two types of real-time position data obtained are accurately input into the trained path recognizer according to the input format and specifications preset by the path recognizer. The path recognizer then activates its internal complex neural network structure. The input data first enters each convolutional layer through the input layer. The convolutional kernels quickly extract and match the data features according to the patterns learned during the previous training, accurately identifying the combination of product storage and groove contour features similar to those in the training data; subsequently, after a series of in-depth processes in the pooling layer and the fully connected layer, combining the rich experience and mapping rules learned in the past, efficiently calculate and generate the corresponding unique product clamping path. Finally, the output clamping path is presented in the form of accurate coordinate sequences, robotic arm action instruction sets, etc., specifying in detail how to move from the initial position of the robotic arm to the pin product for clamping along the optimal trajectory and then along what route to smoothly place the product at the specified position of the conveyor belt groove, thereby ensuring the efficient, accurate, and safe completion of the entire product handling process.

[0075] Embodiment 2, based on the same inventive concept as the anti-offset control method for placing pin products in the conveyor belt groove in the foregoing embodiment, as Figure 2 shown, the present application provides an anti-offset control system for placing pin products in the conveyor belt groove. The system in the embodiment of the present application and the method embodiment are based on the same inventive concept. Among them, the system includes:

[0076] The clamping force determination module 10 is configured to determine the product clamping path according to the storage position of the pin product and the profile position of the conveyor belt groove, and determine the clamping force based on the quality and clamping contact area of the pin product.

[0077] The motor output current sequence acquisition module 20 is configured to identify the motor current based on the clamping force and the product clamping path, and control the motor to drive the product longitudinal movement device to clamp the pin product above the groove of the product conveyor belt according to the recognition result.

[0078] The lateral offset control parameter acquisition module 30 is configured to use the vision acquisition component to perform image acquisition, obtain a set of clamping images, and perform lateral offset recognition to obtain lateral offset control parameters.

[0079] The mapping group set acquisition module 40 is configured to retrieve and obtain the set of downward force - speed mapping groups at the placement moment.

[0080] The centralized analysis module 50 is configured to traverse the set of downward force - speed mapping groups at the placement moment for centralized analysis to determine the centralized interval of the downward force - speed mapping groups at the placement moment.

[0081] The target longitudinal control scheme acquisition module 60 is configured to optimize the longitudinal control parameters of the product longitudinal movement device with the centralized interval of the downward force - speed mapping groups at the placement moment as a constraint, obtain the target longitudinal control scheme, and perform anti - offset control for placing the product in the groove of the product conveyor belt in combination with the lateral offset control parameters.

[0082] Furthermore, the centralized analysis module 50 further includes:

[0083] The analysis point set acquisition unit is configured to construct a two - dimensional tolerance threshold analysis space based on the set of downward force - speed mapping groups at the placement moment, and obtain an analysis point set, where each analysis point corresponds to a downward force - speed mapping group at the placement moment.

[0084] The analysis point neighborhood acquisition unit is configured to randomly extract multiple analysis points from the analysis point set, and construct neighborhoods for the multiple analysis points according to a preset iteration bandwidth to obtain multiple analysis point neighborhoods.

[0085] Analysis point neighborhood density acquisition unit, which is used to traverse the multiple analysis point neighborhoods to calculate the neighborhood density, obtain multiple analysis point neighborhood densities, take the analysis point corresponding to the maximum value of the multiple analysis point neighborhood densities as the direction analysis point, and take the remaining multiple analysis points as multiple following analysis points.

[0086] Following iterative analysis point acquisition unit, which is used to take the direction of the direction analysis point as the iterative direction, and iterate the multiple following analysis points according to a preset iterative bandwidth to obtain multiple following iterative analysis points.

[0087] Neighborhood density maximum value judgment unit, which is used to construct multiple following iterative analysis point neighborhoods of the multiple following iterative analysis points, and judge whether the maximum value of the neighborhood density of the multiple following iterative analysis point neighborhoods is greater than or equal to the analysis point neighborhood density of the direction analysis point. If so, update the following iterative analysis point corresponding to the maximum value of the neighborhood density of the multiple following iterative analysis point neighborhoods as the direction analysis point, and continue the iteration until the preset maximum number of iterations is satisfied. Take the neighborhood corresponding to the direction analysis point obtained in the last iteration as the concentrated interval of the placement time downward force - speed mapping group.

[0088] Further, the analysis point neighborhood density acquisition unit further includes:

[0089] Analysis point quantity statistics unit, which is used to respectively count the number of analysis points in the multiple analysis point neighborhoods, and divide the statistical result by the area of the corresponding multiple analysis point neighborhoods to obtain the multiple analysis point neighborhood densities.

[0090] Further, the target longitudinal control scheme acquisition module 60 further includes:

[0091] Longitudinal distance acquisition unit, which is used to perform longitudinal distance feature recognition on the clamping image set to obtain the longitudinal distance, where the longitudinal distance is the vertical distance between the pin product and the plane where the groove of the product conveyor belt is located.

[0092] Downward movement time acquisition unit, which is used to obtain a preset downward movement time, and combine the longitudinal distance to determine the initial downward force - speed mapping group.

[0093] Longitudinal control scheme optimization unit, which is used to take the concentrated interval of the placement time downward force - speed mapping group as a constraint, and optimize the longitudinal control scheme of the product longitudinal movement device according to the longitudinal distance and the initial downward force - speed mapping group to obtain the target longitudinal control scheme.

[0094] Further, the vertical control scheme optimization unit further includes:

[0095] A vertical control scheme acquisition unit, which is used to identify the initial downward force - speed mapping group and the vertical distance by using a scheme identifier to obtain multiple vertical control schemes.

[0096] A preliminary screening vertical control scheme acquisition unit, which is used to screen the multiple vertical control schemes by taking the concentrated interval of the downward force - speed mapping group at the placement moment as a constraint to obtain multiple preliminary screening vertical control schemes.

[0097] A preliminary screening fitness acquisition unit, which is used to obtain multiple preliminary screening downward force - speed mapping groups at the placement moment of the multiple preliminary screening vertical control schemes, and perform control scheme fitness analysis based on the multiple preliminary screening downward force - speed mapping groups at the placement moment to obtain multiple preliminary screening fitness values.

[0098] A target vertical control scheme determination unit, which is used to take the preliminary screening vertical control scheme corresponding to the maximum value among the multiple preliminary screening fitness values as the target vertical control scheme.

[0099] Further, the preliminary screening fitness acquisition unit further includes:

[0100] A preliminary screening placement amplitude set acquisition unit, which is used to perform frequent item mining on the vibration data of the pin product groove placement based on the multiple preliminary screening downward force - speed mapping groups at the placement moment to obtain multiple preliminary screening placement amplitude sets and multiple preliminary screening placement vibration time sets.

[0101] A weighted calculation unit, which is used to perform weighted calculation on the multiple preliminary screening placement amplitude sets and the multiple preliminary screening placement vibration time sets to obtain the multiple preliminary screening fitness values.

[0102] Further, the clamping force determination module 10 further includes:

[0103] A training data acquisition unit, which is used to obtain multiple sample pin product storage positions, multiple sample conveyor belt groove contour positions, and multiple sample product clamping paths as training data.

[0104] A path identifier acquisition unit, which is used to perform supervised training on a framework constructed based on a convolutional neural network by using the training data to learn the one - to - one mapping relationship between the pin product storage position, the conveyor belt groove contour position, and the product clamping path until the training converges to obtain a trained path identifier.

[0105] A product clamping path acquisition unit, which acquires the product clamping path based on the recognition of the storage position of the pin product and the contour position of the conveyor belt groove by the path recognizer, where the product clamping path is the movement path for the product longitudinal movement device to clamp the pin product from the storage position of the pin product to above the contour position of the conveyor belt groove.

[0106] It should be noted that the above sequence of embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. And the above describes specific embodiments of this specification. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0107] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

[0108] This specification and the drawings are only exemplary descriptions of the present application and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and modifications.

Claims

1. Anti-deviation control method for placing the groove of the conveyor belt of the pin product, characterized in that: The method comprises: Determine the product clamping path based on the storage position of the pin product and the position of the conveyor belt groove profile, and determine the clamping force based on the mass of the pin product and the clamping contact area; Motor current identification is performed based on the clamping force and the product clamping path, and the motor is controlled to drive the product longitudinal moving device to clamp the pin product to the top of the groove of the product conveyor belt according to the product clamping path according to the identification result; The visual acquisition component is used to acquire images, obtain a set of clamped images, and perform lateral offset recognition to obtain lateral offset control parameters; Retrieve and obtain a set of downward force-speed mapping groups at the time of placement; Traversing the set of downward force-speed mapping groups at the time of placement to construct the neighborhood of the analysis point, and iteratively analyzing the constructed neighborhood to determine the concentrated interval of the downward force-speed mapping group at the time of placement; Taking the concentrated interval of the downward force-speed mapping group at the placement time as a constraint, the longitudinal control parameters of the product longitudinal moving device are optimized to obtain the target longitudinal control scheme, and the anti-deviation control of the product conveyor belt groove placement is carried out in combination with the lateral offset control parameters.

2. The anti-deviating control method for placing the groove of the pin product conveyor belt according to claim 1, characterized in that: Traverse the set of downward force-speed mapping groups at the time of placement to construct the neighborhood of the analysis point, and iteratively analyze the constructed neighborhood to determine the concentrated interval of the downward force-speed mapping group at the time of placement, including: Based on the placement moment downward movement force-speed mapping group set, a two-dimensional tolerance threshold analysis space is constructed to obtain a set of analysis points, wherein each analysis point corresponds to a placement moment downward movement force-speed mapping group; Randomly extracting multiple analysis points from the analysis point set, and constructing neighborhoods for the multiple analysis points according to a preset iteration bandwidth to obtain multiple analysis point neighborhoods; Traversing the neighborhoods of the plurality of analysis points to perform neighborhood density calculation to obtain the neighborhood densities of the plurality of analysis points, taking the analysis point corresponding to the maximum value of the neighborhood densities of the plurality of analysis points as the direction analysis point, and taking the remaining plurality of analysis points as the plurality of follow-up analysis points; Taking the direction of the directional analysis point as the iteration direction, iterating the multiple follow-up analysis points according to a preset iteration bandwidth to obtain multiple follow-up iteration analysis points; Construct multiple follow-up iterative analysis point neighborhoods of the multiple follow-up iterative analysis points, determine whether the maximum value of the neighborhood density of the multiple follow-up iterative analysis point neighborhoods is greater than or equal to the analysis point neighborhood density of the directional analysis point, and if so, update the follow-up iterative analysis point corresponding to the maximum value of the neighborhood density of the multiple follow-up iterative analysis point neighborhoods to the directional analysis point, continue to iterate until the preset maximum number of iterations is met, and use the neighborhood corresponding to the directional analysis point obtained in the last iteration as the concentrated interval of the downward force-speed mapping group at the placement moment.

3. The anti-deviating control method for placing the groove of the pin product conveyor belt according to claim 2, characterized in that: The number of analysis points in the neighborhoods of the multiple analysis points is counted respectively, and the statistical result is compared with the area of ​​the corresponding neighborhoods of the multiple analysis points to obtain the neighborhood density of the multiple analysis points.

4. The anti-deviating control method for placing the groove of the pin product conveyor belt according to claim 1, characterized in that: Taking the concentrated interval of the downward force-speed mapping group at the placement time as a constraint, the longitudinal control parameters of the product longitudinal moving device are optimized to obtain the target longitudinal control solution, including: Performing longitudinal distance feature recognition on the clamped image set to obtain a longitudinal distance, wherein the longitudinal distance is a vertical distance between the pin product and the plane where the groove of the product conveyor belt is located; Obtaining a preset downward movement time, and determining an initial downward movement force-speed mapping group in combination with the longitudinal distance; Taking the concentrated interval of the downward force-speed mapping group at the placement moment as a constraint, the longitudinal control scheme of the product longitudinal moving device is optimized according to the longitudinal distance and the initial downward force-speed mapping group to obtain the target longitudinal control scheme.

5. The anti-deviating control method for placing the groove of the pin product conveyor belt according to claim 4, characterized in that: Taking the concentrated interval of the downward force-speed mapping group at the placement time as a constraint, optimizing the longitudinal control scheme of the product longitudinal moving device according to the longitudinal distance and the initial downward force-speed mapping group to obtain the target longitudinal control scheme, including: Using a scheme identifier to identify the initial downward force-speed mapping group and the longitudinal distance to obtain multiple longitudinal control schemes; Taking the concentrated interval of the downward force-speed mapping group at the placement time as a constraint, screening the multiple longitudinal control schemes to obtain multiple preliminary screened longitudinal control schemes; Acquire multiple initial screening placement moment downward force-speed mapping groups of the multiple initial screening longitudinal control schemes, perform control scheme fitness analysis based on the multiple initial screening placement moment downward force-speed mapping groups, and obtain multiple initial screening fitnesses; The initial screening longitudinal control scheme corresponding to the maximum value among the multiple initial screening fitnesses is used as the target longitudinal control scheme.

6. The anti-deviating control method for placing the groove of the pin product conveyor belt according to claim 5, characterized in that: include: Based on the multiple initial screening placement moment downward force-speed mapping groups, frequent item mining of pin product groove placement vibration data is performed to obtain multiple initial screening placement amplitude sets and multiple initial screening placement vibration time sets; A weighted calculation is performed on the multiple initial screening placement amplitude sets and the multiple initial screening placement vibration time sets to obtain the multiple initial screening fitnesses.

7. The anti-deviating control method for placing the groove of the pin product conveyor belt according to claim 1, characterized in that: Determine the product picking path according to the pin product storage position and the conveyor belt groove contour position, including: Acquire multiple sample pin product storage locations, multiple sample conveyor belt groove contour locations, and multiple sample product gripping paths as training data; The framework based on the convolutional neural network is supervised and trained using the training data to learn the two-to-one mapping relationship between the storage position of the pin product and the contour position of the conveyor belt groove and the product gripping path until the training converges and a trained path identifier is obtained. The pin product storage position and the conveyor belt groove contour position are identified based on the path identifier to obtain the product clamping path, wherein the product clamping path is the movement path of the product longitudinal moving device to clamp the pin product from the pin product storage position to above the conveyor belt groove contour position. 8.Anti-deviation control system for the placement of grooves on the conveyor belt of tube pin products, characterized in that: The system is used to implement the anti-deviation control method for placing a groove of a conveyor belt of a tube pin product according to any one of claims 1 to 7, and the system comprises: A clamping force determination module, the clamping force determination module is used to determine the product clamping path according to the storage position of the pin product and the position of the conveyor belt groove contour, and determine the clamping force based on the mass of the pin product and the clamping contact area; A motor output current sequence acquisition module, which performs motor current identification based on the clamping force and the product clamping path, and controls the motor to drive the product longitudinal moving device to clamp the pin product to the top of the groove of the product conveyor belt according to the product clamping path according to the identification result; A lateral offset control parameter acquisition module, which is used to use a visual acquisition component to perform image acquisition, obtain a clamped image set, perform lateral offset recognition, and obtain lateral offset control parameters; A mapping group set acquisition module, the mapping group set acquisition module is used to retrieve and obtain a downward movement force-speed mapping group set at the placement time; A centralized analysis module, the centralized analysis module is used to traverse the set of downward force-speed mapping groups at the placement time for centralized analysis, and determine a centralized interval of the downward force-speed mapping group at the placement time; A target longitudinal control scheme acquisition module is used to optimize the longitudinal control parameters of the product longitudinal moving device based on the concentrated interval of the downward force-speed mapping group at the placement time as a constraint, obtain the target longitudinal control scheme, and perform anti-deviation control of the product conveyor belt groove placement in combination with the lateral offset control parameters.

Images