Automatic large ball crushing device for disc pelletizer

Abstract

The invention discloses an automatic large ball crushing device for a disc pelletizer, and belongs to the technical field of pelletizing equipment. In the disc pelletizing process, the problem that the pelletizing quality and the subsequent production process are affected by large pellets often occurs. The device comprises an image acquisition module for acquiring image data of a pelletizing disc area, a control unit with a built-in image analysis and three-dimensional path planning algorithm, a six-degree-of-freedom mechanical arm close to the pelletizing disc area and a crushing hand mounted at the tail end of the mechanical arm. The image analysis algorithm detects the large ball and generates coordinate information, the path planning algorithm calculates an obstacle avoidance track, the mechanical arm positions according to the track, and the crushing hand crushes the large ball. The device is mainly used for automatically detecting and crushing large balls generated in the disc pelletizer, and the pelletizing quality and the production efficiency are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of pelletizing equipment, and more specifically, to an automatic large ball crushing device for a disc pelletizing machine. Background Art

[0002] In industrial production, disc pelletizing machine is a widely used equipment in pelletizing process, for example, in mining, metallurgy and other industries, it is used to make powdered materials into spherical particles. However, in the process of disc pelletizing, some large balls with diameters exceeding the normal range often appear. The existence of these large balls will bring many adverse effects to the pelletizing production.

[0003] First of all, from the perspective of pelletizing quality, the appearance of large balls will destroy the uniformity of the entire pelletizing system. Qualified pellets have certain standard requirements in terms of size and density. The mixing of large balls makes the size distribution of the pellets uneven, which in turn affects the roasting and sintering of the pellets in subsequent processes. For example, during the sintering process, due to the large size of the large balls, the internal heat and mass transfer processes are relatively slow, which may lead to insufficient sintering and reduce the strength and quality of the sintered ore.

[0004] Secondly, the presence of large balls will also hinder the normal operation of the disc pelletizing machine. When the large balls roll in the pelletizing disc, they will increase the load of the disc, causing the motor and other driving equipment to consume more energy to maintain the operation of the disc. Long-term high-load conditions will accelerate the wear of the equipment, shorten the service life of the equipment, and increase the maintenance cost and downtime of the equipment.

[0005] Furthermore, large balls can also cause trouble in the subsequent transportation and processing stages. Large balls may block transportation pipelines, chutes and other equipment, affecting the normal transportation of materials and causing interruptions in the production process. At the same time, large balls are difficult to pass through some screening equipment and require additional manual intervention for processing, increasing labor costs and labor intensity.

[0006] The reasons for the appearance of large balls are relatively complex. On the one hand, the properties of the material may fluctuate, such as uneven particle size distribution and unstable water content, which may cause some materials to agglomerate and form large balls during the ball making process. On the other hand, unreasonable settings of the ball making machine's operating parameters, such as the speed, inclination, water spraying position and water spraying amount of the ball making disk, will also affect the formation process of the pellets and increase the probability of large balls.

[0007] In the past, when solving the problem of large balls, manual screening and crushing methods were usually adopted. This method is not only inefficient, but also labor-intensive, and it is difficult to process large balls in real time and accurately. At the same time, manual operation has certain subjectivity and uncertainty, which may miss some large balls or cause unnecessary damage to qualified pellets. With the continuous improvement of industrial automation level, some automated equipment has been tried to solve the problem of large balls, but the existing equipment on the market still has deficiencies in detection accuracy, crushing effect and adaptability, etc., and cannot meet the actual production needs. Therefore, it is of great practical significance to develop an automatic large ball crushing device for a disc pelletizer that is efficient and accurate. Summary of the Invention

[0008] One object of the present invention is to solve at least the above problems and provide at least the advantages described hereinafter.

[0009] To achieve these and other advantages in accordance with the present invention, there is provided an automatic large ball crushing device for a disc pelletizer, comprising: An image acquisition module for acquiring image data of the pelletizing disc area; A control unit communicatively connected to the image acquisition module, which is built-in with an image analysis algorithm and a three-dimensional path planning algorithm based on machine vision. The image analysis algorithm is configured to perform multi-target detection on the received image data, identify the sphere contour through an edge detection algorithm, calculate the equivalent diameter according to the sphere projection area, and when the detected equivalent diameter exceeds a preset threshold and the number of consecutive frames reaches a set value, it is determined that there are large balls, and the number of large balls is counted. When the number of large balls reaches a preset threshold, three-dimensional coordinate information of the large ball area is generated. The three-dimensional path planning algorithm is configured to calculate an obstacle avoidance motion trajectory based on the three-dimensional coordinate information of the large ball area and the preset starting point position and crushing area position; A six-degree-of-freedom robotic arm disposed at a position close to the pelletizing disc area. The joint drive assembly of the six-degree-of-freedom robotic arm is signal-connected to the control unit and performs spatial positioning based on the obstacle avoidance motion trajectory in response to a trigger command generated by the control unit; A crushing hand, which includes a mounting seat installed at the end of the six-degree-of-freedom robotic arm, a plurality of static fingers installed on the mounting seat, a drive assembly installed on the mounting seat, and a plurality of moving fingers driven to rotate by the drive assembly. When the drive assembly drives the plurality of moving fingers to rotate, the plurality of moving fingers are inserted into the gaps between the plurality of static fingers one by one.

[0010] Preferably, the main body of the static finger is rod-shaped, with the end bent, and stop bars are provided on both sides of the outermost part.

[0011] Preferably, comb-like structures are provided on both side walls of the static finger, and triangular reinforcing ribs are provided at the tooth roots of the comb-like structures. The tooth gap G of the comb-like structure is set as follows: G = d max + 16 mm, and G < D min , where d max is the diameter of the qualified ball, and D min is the diameter of the large ball; Sawtooth structures are correspondingly provided on both side walls of the moving finger for the comb-like structures. The tips of the sawtooth structures face the comb-like structures, and the heads of the sawtooth structures are double-bevel wedges with an included angle of 30°.

[0012] Preferably, the driving assembly includes a cylinder mounted on the mounting base, a connecting rod hinged to the piston rod of the cylinder, and a mounting rod hinged to the other end of the connecting rod. Among them, the moving finger is fixed on the mounting rod. When the piston rod is in an extended state, it is the initial position. In the initial position, the moving finger and the static finger are in an open state.

[0013] Preferably, a double eccentric wheel vibrator is further included, which is mounted on the mounting base through a rubber spring composite shock absorption layer. The static finger is mounted on the double eccentric wheel vibrator. Among them, the vibration direction of the double eccentric wheel vibrator is set to vibrate obliquely at 45° relative to the length direction of the static finger. When the moisture content of the qualified ball is less than 10%, the vibration parameters of the double eccentric wheel vibrator are set as follows: the frequency is 20 Hz and the amplitude is 2 mm; When the moisture content of the qualified ball is not less than 10%, the vibration parameters of the double eccentric wheel vibrator are set as follows: the frequency is 12 Hz and the amplitude is 5 mm.

[0014] Preferably, the rubber spring composite shock absorption layer includes a bottom plate, a telescopic sleeve provided on the bottom plate, a spring sleeved in the telescopic sleeve, and a top plate provided at the top of the telescopic sleeve. Among them, both ends of the spring are fixed to the bottom plate and the top plate respectively. A rubber layer is laid on the outer surface of the spring, buffer layers are laid on the bottom surface and the top surface of the top plate, and the double eccentric wheel vibrator is mounted on the top plate.

[0015] Preferably, the image analysis algorithm specifically includes the following steps: S1. Convert the received image data into a grayscale image and the HSV color space. Based on the distribution differences of the hue H, saturation S, and brightness V of the sphere and the material layer in the HSV color space, set the three-channel joint threshold range for dynamic segmentation to generate a preliminary binary mask; S2. Extract the brightness channel of the HSV color space, calculate the brightness mean μ and variance σ² of the preliminary segmentation region, and set the variance threshold σ th , for those satisfying σ² > σth The segmented area is determined as the dynamic material layer, and σ²≤σ is retained th The smoothed area is used as the candidate sphere set; S3. Calculate the image contrast Contrast using the gray-level co-occurrence matrix. The texture feature is the contrast. According to the texture difference between the sphere and the material layer, a preset contrast threshold C th , segment the image to generate a texture feature binary mask. Among them, the contrast higher than the contrast threshold C th is the material layer area, and the one not higher than the contrast threshold C th is used as the candidate sphere set; Among them, P ( i , j ) is the joint probability of the pixel pair ( i , j ) in the gray-level co-occurrence matrix, N is the number of gray levels; S4. Fuse the color and texture feature masks, and select the pixels with gray values in the range of [μ - 2σ, μ + 2σ] and texture classification as spheres as the initial seed points. Set the dynamic similarity criterion, and iteratively merge adjacent similar pixels until convergence to output the refined sphere region segmentation result. Among them, the pixel difference threshold allowed by the dynamic similarity criterion is: gray difference ΔGray≤15, HSV color distance ΔE≤8, and Euclidean distance of texture feature vector ΔF≤0.2; S5. Use Canny edge detection to extract the closed contour for the segmented sphere region, calculate the convex hull of the contour and geometric parameters. The geometric parameters include area A, perimeter L, and shape factor SF = 4πA / L². Among them, the area A is the number of pixels, and set the geometric constraint conditions as the area threshold and shape factor SF≧0.80; the contour that meets the geometric constraint conditions is determined as a valid sphere; S6. According to the projection area of the sphere and the spatial geometric relationship, establish an equivalent diameter model and calculate the equivalent diameter D eq , perform trajectory association on the detection results of consecutive multiple frames, and use Kalman filtering to track the sphere movement trajectory. The number of stable trajectories is the total number of real-time spheres; Among them, A proj is the projection area of the sphere region, which is converted by pixel area × single-pixel physical size, k is the ball disk inclination compensation coefficient, which is a pre-calibrated parameter, k = cosθ, and θ is the angle between the disk axis and the camera optical axis; S7. When the detected equivalent diameter D eq is greater than the preset diameter threshold D th , and the number of large balls is greater than the preset quantity threshold, trigger the reporting of the large ball coordinates and the crushing instruction.

[0016] The present invention has at least the following beneficial effects: First, through the collaborative work of the image acquisition module, the control unit, the six-degree-of-freedom robotic arm, and the crushing hand, the present invention realizes the automatic detection and crushing of large balls. The image analysis algorithm can accurately identify large balls, and the three-dimensional path planning algorithm ensures that the robotic arm avoids obstacles and reaches the position of the large balls. The crushing hand effectively crushes the large balls, improving the automation level and efficiency of the pelletizing production, reducing manual intervention, lowering the labor intensity, and at the same time improving the pelletizing quality.

[0017] Second, the design of the comb-like structure and the serrated structure enables the moving fingers and the static fingers to crush large balls more precisely when they cooperate. The reasonable setting of the comb gap ensures that qualified balls will not be accidentally crushed, while allowing large balls to be effectively crushed. The triangular reinforcing ribs enhance the structural strength of the static fingers and improve their durability. The head of the double-inclined-plane wedge-shaped serrated structure can more effectively cut into large balls and improve the crushing efficiency.

[0018] Third, the driving component has a simple and reliable structure and can accurately control the movement of the moving fingers. Through the telescopic movement of the piston rod of the cylinder, the connecting rod and the mounting rod are driven to move, thereby realizing the opening and closing actions of the moving fingers and the static fingers. The setting of the initial position makes the moving fingers and the static fingers in the open state during the movement of the robotic arm, avoiding interference with surrounding objects, and then performing the closing and crushing operation after reaching the position of the large balls, improving the accuracy and safety of the operation. In addition, the closing action of the finger-like fingers also plays a role in cleaning the adhesive material in each other's finger gaps.

[0019] Fourth, aiming at the problem that qualified balls are easily accidentally crushed, the double-eccentric vibrator works first before crushing large balls, and uses different vibration parameters to screen out qualified balls from the gaps between the static fingers, only retaining large balls for crushing, effectively avoiding unnecessary damage to qualified balls, significantly improving the product qualification rate, and reducing resource waste.

[0020] Fifthly, through multi-step processing and by comprehensively utilizing color, texture, and geometric features, the image analysis algorithm can accurately identify large balls from the image data in the pelletizing disk area. The fusion of color and texture features and the application of the dynamic similarity criterion improve the accuracy and refinement of sphere segmentation. The Kalman filter tracks the movement trajectory of the sphere, making the detection of large balls more stable and reliable. When the conditions are met, the large ball coordinate reporting and crushing instructions are triggered, providing an accurate information basis for subsequent crushing operations, thereby improving the working efficiency and accuracy of the entire large ball automatic crushing device.

[0021] Sixthly, through establishing a three-dimensional map model, the path planning algorithm accurately determines the working space of the robotic arm and the position information of various objects. Calculating the optimal target point coordinates ensures that the robotic arm can efficiently grasp large balls. The path search algorithm and optimization processing can find the optimal path from the starting point to each target point, and remove redundant nodes and smooth the path, making the movement of the robotic arm smoother and more efficient. At the same time, it avoids collisions with obstacles, improving the safety and reliability of the robotic arm operation.

[0022] Other advantages, objectives, and features of the present invention will be partially reflected by the following description and partially understood by those skilled in the art through the research and practice of the present invention. Brief Description of the Drawings

[0023] Figure 1 It is a schematic diagram of the overall structure of the automatic crushing device according to one of the technical solutions of the present invention; Figure 2 It is a detailed view of the crushing hand according to one of the technical solutions of the present invention; Figure 3 It is a view of two states of the crushing hand according to one of the technical solutions of the present invention; Figure 4 It is a detailed view of the comb-like structure according to one of the technical solutions of the present invention; Figure 5 It is a detailed view of the triangular reinforcing rib according to one of the technical solutions of the present invention; Figure 6 It is a detailed view of the serrated structure according to one of the technical solutions of the present invention; Figure 7 It is a detailed view of the rubber spring composite shock-absorbing layer according to one of the technical solutions of the present invention.

[0024] Reference numerals in the drawings: image acquisition module 1, six-degree-of-freedom robotic arm 2, pelletizing disc 3, crushing hand 4, mounting base 41, static finger 42, moving finger 43, stop bar 44, comb-like structure 5, triangular stiffener 6, serrated structure 7, air cylinder 45, connecting rod 46, mounting rod 47, double eccentric vibrator 8, rubber spring composite shock-absorbing layer 9, bottom plate 91, telescopic sleeve 92, spring 93, top plate 94, buffer layer 95, state 10 when the serrated structure is inserted between the static fingers. Detailed implementation manners

[0025] The present invention will be further described in detail below with reference to the accompanying drawings, so that those skilled in the art can implement it according to the text of the specification.

[0026] It should be noted that the experimental methods described in the following implementation schemes are all conventional methods unless otherwise specified, and the reagents and materials can be obtained from commercial channels unless otherwise specified; in the description of the present invention, the orientation or positional relationship indicated by the terms is based on the orientation or positional relationship shown in the accompanying drawings, which is only for the convenience of describing the present invention and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention.

[0027] As Figures 1 to 7 shown, the present invention provides a large ball automatic crushing device for a disc pelletizing machine, including: An image acquisition module 1, which is used to collect image data of the area of the pelletizing disc 3; A control unit, which is communicatively connected to the image acquisition module 1 and has an image analysis algorithm and a three-dimensional path planning algorithm based on machine vision built therein. The image analysis algorithm is configured to perform multi-target detection on the received image data, identify the sphere contour through an edge detection algorithm, calculate the equivalent diameter according to the projected area of the sphere, and when it is detected that the equivalent diameter exceeds a preset threshold and the number of consecutive frames reaches a set value, it is determined that there is a large ball, and the number of large balls is counted. When the number of large balls reaches a preset threshold, three-dimensional coordinate information of the large ball area is generated. The three-dimensional path planning algorithm is configured to calculate an obstacle avoidance motion trajectory based on the three-dimensional coordinate information of the large ball area and the preset starting point position and crushing area position; A six-degree-of-freedom robotic arm 2, which is arranged at a position close to the area of the pelletizing disc 3. The joint drive assembly of the six-degree-of-freedom robotic arm 2 is signal-connected to the control unit and performs spatial positioning based on the obstacle avoidance motion trajectory in response to the trigger instruction generated by the control unit; The crushing hand 4 includes a mounting base 41 installed at the end of the six-degree-of-freedom robotic arm 2, a plurality of static fingers 42 installed on the mounting base 41, a driving assembly installed on the mounting base 41, and a plurality of moving fingers 43 driven by the driving assembly to rotate. When the driving assembly drives the plurality of moving fingers 43 to rotate, the plurality of moving fingers 43 are inserted into the gaps between the plurality of static fingers 42 one by one. After performing the crushing action, continuously driving the moving fingers 43 to penetrate the static fingers 42 back and forth can effectively clean the adhesive material in the corresponding finger gaps, playing a role of maintenance-free.

[0028] In the above technical solution, aiming at the problems that the existing disc pelletizer produces large balls, which affect the quality and operation, and the manual treatment is inefficient and inaccurate, the image acquisition module 1 collects the image data of the pelletizing disc 3 in real time, and the image analysis algorithm of the control unit can accurately identify the large balls and count the quantity. When the number of large balls reaches the threshold, three-dimensional coordinate information is generated, and then the robotic arm automatically performs the crushing operation, getting rid of the inefficiency and inaccuracy of manual work, realizing the automatic detection and crushing of large balls, and improving the automation degree and efficiency of pelletizing production.

[0029] Regarding the problems of realizing the automatic detection and crushing of large balls and improving the automation degree and efficiency, the image acquisition module 1 provides a data basis, the image analysis algorithm accurately identifies large balls, the three-dimensional path planning algorithm plans an obstacle-avoiding motion trajectory, and the six-degree-of-freedom robotic arm 2 can accurately reach the position of the large ball according to the trajectory, and the crushing hand 4 completes the crushing. Each part works together to automate the whole process, greatly improving the production efficiency and reducing manual intervention and labor intensity.

[0030] For the problem that the robotic arm accurately reaches the position of the large ball in a complex environment and returns safely, the three-dimensional path planning algorithm of the control unit is based on the coordinates of the large ball and the preset position information, considering various objects and obstacles in the working space, and calculates the optimal obstacle-avoiding motion trajectory. The six-degree-of-freedom robotic arm 2 has high flexibility and positioning accuracy, can accurately move to the target position according to the trajectory, and safely return to the initial position after completing the crushing.

[0031] Specifically, the implementation of the image acquisition module 1: The image acquisition module 1 plays a key role in data collection in the whole automatic large ball crushing device. In practical applications, it is advisable to select a high-definition industrial camera as the image acquisition device. The selection of the installation position is crucial. It needs to be installed at a suitable height and angle above the pelletizing disc 3 to ensure that the image data of the pelletizing disc 3 area can be clearly and comprehensively collected. Generally speaking, the installation height should be determined according to the size of the pelletizing disc 3 and the viewing angle of the camera. For example, for a pelletizing disc 3 with a diameter of 7.5 meters, the camera can be installed directly above the pelletizing disc 3 at a position 2 - 3 meters away.

[0032] The installation angle of the camera should be vertically downward to avoid shooting blind spots. At the same time, to ensure the image quality, the camera should be cleaned and maintained regularly to prevent dust, water vapor, etc. from affecting the image clarity. In terms of data transmission, a stable and reliable communication method, such as Ethernet or fiber optic communication, is adopted to transmit the collected image data to the control unit in real time and accurately.

[0033] Implementation of the control unit: The control unit is the core of the entire device, responsible for processing and analyzing the image data and generating corresponding control instructions. A high-performance industrial computer is selected for the control unit, which has powerful computing power and data processing capabilities and can quickly process a large amount of image data.

[0034] The built-in image analysis algorithm based on machine vision is an important part of the control unit. During actual operation, first, preprocessing is performed on the received image data, including operations such as image enhancement and filtering to improve the image quality. Then, the sphere contour is identified through edge detection algorithms, and common edge detection algorithms include the Canny algorithm, etc. The equivalent diameter is calculated based on the projected area of the sphere, and an accurate mathematical model of the projected area and the equivalent diameter needs to be established. When the detected equivalent diameter exceeds the preset threshold and the number of consecutive frames reaches the set value, it is determined that a large ball exists, and the number of large balls is counted. When the number of large balls reaches the preset threshold, the three-dimensional coordinate information of the large ball area is generated.

[0035] The three-dimensional path planning algorithm then calculates the obstacle avoidance motion trajectory according to the three-dimensional coordinate information of the large ball area and the preset starting point position and the position of the crushing area, using a path search algorithm such as the A* algorithm. During the calculation process, factors such as the working space range of the robotic arm, the position and shape of the obstacles, etc. should be fully considered to ensure that the planned path is safe and efficient.

[0036] Implementation of the six-degree-of-freedom robotic arm 2: The six-degree-of-freedom robotic arm 2 is set at a position close to the pelletizing disc 3 area, and its base needs to be firmly installed on the ground or other stable support structures to ensure the stability of the robotic arm during movement.

[0037] The joint drive components of the robotic arm are signal-connected to the control unit. When the control unit generates a trigger instruction, the joint drive components control the movement of each joint of the robotic arm according to the obstacle avoidance motion trajectory to achieve spatial positioning. During the movement of the robotic arm, its motion state should be monitored in real time. The angle, speed, etc. of the joints are obtained through sensors installed at the joints, and this information is fed back to the control unit. The control unit adjusts the movement of the robotic arm according to the feedback information to ensure that the robotic arm can accurately reach the position of the large ball.

[0038] Meanwhile, to ensure the safety of the robotic arm, corresponding safety protection measures should be set up, such as limit switches, overload protection devices, etc. When the robotic arm moves beyond the preset range or encounters excessive resistance, the safety protection device can be triggered in a timely manner to stop the movement of the robotic arm and avoid equipment damage and casualties.

[0039] Implementation of the crushing hand 4: The crushing hand 4 is installed at the end of the six-degree-of-freedom robotic arm 2, and its mounting base 41 should be firmly fixed at the end of the robotic arm to ensure that there is no loosening or shaking during the crushing process.

[0040] The installation of the static fingers 42 and the moving fingers 43 should be precise to ensure that the moving fingers 43 can accurately insert into the gap between the static fingers 42 when rotating. The drive assembly is the key component for controlling the rotation of the moving fingers 43, and usually a motor or a cylinder 45 is used as the power source. During actual operation, when the robotic arm reaches the position of the large ball, the drive assembly drives the moving fingers 43 to rotate according to the instructions of the control unit, inserts into the gap between the static fingers 42, and crushes the large ball. As Figure 4 shown in the state 10 when the serrated structure inserts between the static fingers.

[0041] To improve the crushing effect, the materials of the static fingers 42 and the moving fingers 43 should have high strength and wear resistance, and materials such as alloy steel can be selected. At the same time, the shapes and sizes of the static fingers 42 and the moving fingers 43 should be optimized to better meet the crushing requirements of the large ball.

[0042] In another technical solution, the main body of the static finger 42 is rod-shaped, with a bent end, and blocking rods 44 are provided on both sides of the outermost part.

[0043] In the above technical solution, aiming at the problem that the large ball is likely to slip from both sides of the static finger 42 during crushing, resulting in crushing failure and reduced efficiency, the design of the bent end of the static finger 42 and the blocking rods 44 provided on both sides plays a good blocking role. When the large ball enters the crushing area, the blocking rods 44 can limit the lateral movement of the large ball and prevent it from escaping from both sides, so that the large ball can be stably located between the static finger 42 and the moving finger 43, thereby improving the success rate of the crushing operation and ensuring the crushing efficiency.

[0044] Regarding the problem that the static finger 42 needs to better limit and fix the large ball to improve the crushing effect, this structural design enhances the limiting ability of the static finger 42 to the large ball. The bent end part can initially block the rolling trend of the large ball, and the blocking rods 44 on both sides further strengthen the restraint on the large ball, making the large ball stable during the crushing process, enabling the moving finger 43 to more accurately insert into the gap between the static fingers 42 to crush the large ball, and effectively improving the crushing effect.

[0045] Specifically, the implementation of the main rod-shaped structure of the static finger 42: The main body of the static finger 42 adopts a rod-shaped structure, which has good mechanical properties and stability. In terms of material selection, considering that it needs to withstand the large pressure when crushing large balls, high-strength metal materials such as alloy steel are usually selected. Alloy steel has high strength and hardness, which can ensure that the static finger 42 does not deform or damage during long-term use.

[0046] The size design of the rod-shaped structure should be determined according to the actual crushing requirements and the size of the large balls. For large balls with a diameter of 45 - 300 mm to be crushed, the diameter of the rod is generally between 12 - 16 mm, and the length is adjusted according to the overall design of the crushing hand 4 and the size of the pelletizing disc 3, usually between 200 - 350 mm.

[0047] The implementation of the bending at the end of the static finger 42: The bending at the end of the static finger 42 is to better limit the large balls. The design of the bending angle is crucial and needs to be optimized according to the size of the large balls and the working mode of the crushing hand 4. Generally speaking, the bending angle between 30° - 60° is more appropriate.

[0048] The implementation of the installation of the retaining rod 44: The retaining rod 44 is installed on both sides of the outermost part of the static finger 42, playing a role in blocking the large balls from sliding to both sides. The material of the retaining rod 44 can also be selected as the same high-strength metal material as the main body of the static finger 42 to ensure its strength and wear resistance.

[0049] Overall assembly and debugging: After completing the processing and installation of each part of the static finger 42, it needs to be assembled onto the mounting seat 41 of the crushing hand 4. During the assembly process, it is necessary to ensure that the installation position of the static finger 42 is accurate and the clearance between it and the moving finger 43 is appropriate. After the installation is completed, debugging work needs to be carried out to check whether the movement of the static finger 42 and the moving finger 43 is smooth and whether the retaining rod 44 can effectively block the large balls. Through simulated crushing experiments, observe the movement of the large balls during the crushing process, and adjust and optimize the structure and installation of the static finger 42 to ensure that it can meet the actual crushing requirements.

[0050] In another technical solution, comb-shaped structures 5 are provided on both side walls of the static finger 42, triangular reinforcing ribs 6 are provided at the tooth roots of the comb-shaped structures 5, and the comb tooth gap G of the comb-shaped structure is set as: G = d max + 16 mm, and G < D min , where d max is the diameter of qualified balls, and D min is the diameter of large balls; On both side walls of the moving finger 43, sawtooth structures 7 are correspondingly provided for the comb-shaped structures 5. The tips of the sawtooth structures 7 face the comb-shaped structures 5, and the heads of the sawtooth structures 7 are double-bevel wedges with an included angle of 30°.

[0051] In the above technical solution, aiming at the problem of difficult to accurately distinguish qualified balls from large balls and avoid accidental crushing of qualified balls, the setting of the comb tooth gap G cleverly solves this problem. Since G = d max + 16mm, and G < D min , the qualified balls can smoothly pass through the comb tooth gap, while the large balls will be blocked, thus achieving accurate distinction between qualified balls and large balls, effectively improving the product qualification rate and reducing resource waste.

[0052] Regarding the problem of insufficient structural strength of the static finger 42 and the moving finger 43, the triangular reinforcing rib 6 at the tooth root of the comb-shaped structure 5 plays a key role. The triangular reinforcing rib 6 enhances the structural strength of the static finger 42, making it not easy to deform or damage when bearing the pressure of crushing large balls, ensuring the stability of the crushing operation, extending the service life of the equipment, and reducing the maintenance cost.

[0053] Regarding the problem of low crushing efficiency of the moving finger 43, the serrated structures 7 on both sides of the moving finger 43 and the design with a double-bevel wedge-shaped head and an included angle of 30° greatly improve the crushing efficiency. This structure can more effectively cut into the inside of the large ball, quickly crush the large ball, reduce the crushing time, and improve the production efficiency.

[0054] Specifically, the implementation of the comb-shaped structure 5 of the static finger 42: The comb-shaped structures 5 on both side walls of the static finger 42 are the key parts to achieve accurate crushing. In terms of material selection, it is necessary to consider that it needs to have high hardness and wear resistance, and high-strength alloy steel is usually selected. This material can ensure that the comb teeth are not easily worn during long-term contact and extrusion with the pellets.

[0055] The size design of the comb teeth should be strictly carried out according to the requirements of the comb tooth gap G. First, it is necessary to accurately measure the diameter d of the qualified ball max and the diameter D of the large ball min , and determine the comb tooth gap according to the formula G = d max + 16mm, and G < D min . The height of the comb teeth is generally between 5 - 10mm, and the width is between 5 - 15mm. The specific dimensions should be adjusted according to the actual pellet size and crushing requirements.

[0056] Implementation of the triangular reinforcing rib 6: The triangular reinforcing rib 6 is set at the tooth root of the comb-shaped structure 5 to enhance the structural strength. The material of the triangular reinforcing rib 6 is the same as that of the main body of the static finger 42 to ensure its connection strength with the comb teeth.

[0057] The size design of the triangular reinforcing rib 6 should be determined according to the size of the comb teeth and the pressure it bears. Generally speaking, the bottom length of the triangular reinforcing rib 6 is between 5 - 10 mm, and the height is between 3 - 10 mm. When processing the triangular reinforcing rib 6, welding or integral molding can be used.

[0058] Implementation of the sawtooth structure 7 on the moving finger 43: Sawtooth structures 7 are provided on the two side walls of the moving finger 43 corresponding to the comb-like structure 5. The material of the sawtooth structure 7 is also selected as high-strength alloy steel to ensure its strength and wear resistance.

[0059] The tip of the sawtooth structure 7 faces the comb-like structure 5, which is convenient for accurately cutting into the interior of the large ball when breaking the large ball. The head of the sawtooth structure 7 is in the shape of a double-bevel wedge with an included angle of 30°. This angle design enables the moving finger 43 to insert into the large ball more smoothly and improves the crushing efficiency.

[0060] The size design of the sawteeth should match the size of the comb teeth. The height of the sawteeth is generally between 20 - 30 mm, the upper diameter of the width is 30 - 35 mm, and the lower diameter is less than 5 mm.

[0061] Overall assembly and debugging: After completing the processing of each part of the static finger 42 and the moving finger 43, they need to be assembled onto the mounting seat 41 of the crushing hand 4. During the assembly process, it is necessary to ensure the accurate relative position of the static finger 42 and the moving finger 43, and the sawtooth structure 7 can accurately insert into the comb tooth gap.

[0062] After the assembly is completed, debugging work needs to be carried out. First, perform no-load debugging to check whether the movement of the static finger 42 and the moving finger 43 is smooth and whether there is any jamming phenomenon. Then, perform load debugging. Put pellets of different sizes and observe whether the qualified balls can pass through the comb tooth gap smoothly and whether the large balls can be effectively broken. According to the debugging results, fine-tune the positions and sizes of the static finger 42 and the moving finger 43 to ensure that the entire crushing device can work normally and efficiently.

[0063] After a period of production statistics, the breakage rate of the qualified balls is controlled within 5%, which is 10% less than before. The crushing rate of the large balls reaches more than 90%, which is 5% higher than before. This shows that the comb-like structure 5 of the static finger 42 and the sawtooth structure 7 of the moving finger 43 are reasonably designed, can accurately break the large balls, and at the same time protect the qualified balls to the greatest extent, improving the quality and efficiency of pelletizing production.

[0064] In another technical solution, the driving assembly includes a cylinder 45 mounted on the mounting base 41, a connecting rod 46 hinged to the piston rod of the cylinder 45, and a mounting rod 47 hinged to the other end of the connecting rod 46. Wherein, the moving finger 43 is fixed on the mounting rod 47. When the piston rod is in an extended state, it is the initial position. In the initial position, the moving finger 43 and the static finger 42 are in an open state.

[0065] In the above technical solution, for the problem of reliably driving the rotation of the moving finger 43 to achieve accurate cooperation with the static finger 42, through the combination of the cylinder 45, the connecting rod 46 and the mounting rod 47, this driving assembly can convert the linear motion of the piston rod of the cylinder 45 into the rotation of the moving finger 43. This mechanical transmission method is stable and reliable, can accurately control the rotation angle and speed of the moving finger 43, make the moving finger 43 and the static finger 42 precisely cooperate, effectively complete the operation of crushing large balls, and improve the success rate of crushing.

[0066] For the problem of avoiding interference with surrounding objects during the movement of the robotic arm, the extended state of the piston rod is set as the initial position. At this time, the moving finger 43 and the static finger 42 are in an open state. In this way, during the process of the robotic arm moving to the designated position, the moving finger 43 and the static finger 42 will not collide or interfere with surrounding objects, ensuring the safety of the equipment operation and reducing the risk of equipment damage.

[0067] Regarding the problem of designing a simple and easy-to-maintain driving method to reduce costs, this driving assembly has a simple structure and is composed of common components such as a cylinder 45, a connecting rod 46 and a mounting rod 47. The manufacturing process is relatively simple, reducing the manufacturing cost of the equipment. At the same time, these components are easy to disassemble and replace, facilitating daily maintenance and repair, and further reducing the maintenance cost.

[0068] Specifically, the installation and debugging of the cylinder 45: The cylinder 45 is the power source of the driving assembly. During installation, it is necessary to ensure that it is firmly fixed on the mounting base 41. Select a suitable installation position to ensure that the moving direction of the piston rod of the cylinder 45 matches the rotation plane of the moving finger 43. During the installation process, use bolts to fix the cylinder 45 at the designated position on the mounting base 41. When tightening the bolts, pay attention to the uniform force to avoid the cylinder 45 being installed obliquely.

[0069] After the installation is completed, the cylinder 45 needs to be debugged. Connect the air source, check the airtightness of the cylinder 45 to ensure that there is no air leakage. By controlling the switch of the air source, test whether the telescopic movement of the piston rod of the cylinder 45 is smooth and whether the stroke meets the design requirements. Tools such as vernier calipers can be used to measure the stroke of the piston rod for adjustment and calibration to ensure that the extension and contraction of the piston rod can accurately drive the movement of the connecting rod 46.

[0070] Installation and adjustment of connecting rod 46: The connecting rod 46 plays a role in transmitting the movement of the piston rod of the cylinder 45, and its two ends are respectively hinged to the piston rod of the cylinder 45 and the mounting rod 47. When installing the connecting rod 46, the flexibility at the hinge should be ensured to avoid jamming. Appropriate lubricating oil can be applied at the hinge to reduce the frictional resistance.

[0071] After installation, the length and angle of the connecting rod 46 need to be adjusted. The length of the connecting rod 46 should be accurately set according to the stroke of the cylinder 45 and the rotation requirements of the moving finger 43 to ensure that the telescopic movement of the piston rod can accurately drive the moving finger 43 to rotate. By adjusting the connection angles of the connecting rod 46 with the piston rod and the mounting rod 47, it is ensured that the moving finger 43 is in an open state with the static finger 42 at the initial position, and can smoothly insert into the gap between the static fingers 42 when the piston rod contracts.

[0072] Installation and fixation of the mounting rod 47: The mounting rod 47 is used to fix the moving finger 43, and during installation, its connection with the moving finger 43 should be ensured to be firm. The moving finger 43 can be fixed to the mounting rod 47 by welding or bolt connection, and the appropriate connection method should be selected according to the structure and size of the moving finger 43. During welding, the welding quality should be ensured to avoid problems such as false welding and cracks; during bolt connection, the bolts should be tightened to prevent the moving finger 43 from loosening.

[0073] The mounting rod 47 is hinged to the connecting rod 46, and the concentricity at the hinge should be ensured during installation so that the mounting rod 47 can rotate flexibly. At the same time, the position of the mounting rod 47 should be adjusted to ensure the matching accuracy between the moving finger 43 and the static finger 42 during rotation. By finely adjusting the position and angle of the mounting rod 47, the moving finger 43 can accurately crush the large ball when inserting into the gap between the static fingers 42.

[0074] Overall debugging and optimization: After installing all components of the drive assembly, overall debugging should be carried out. First, perform no-load debugging. Start the cylinder 45, observe the rotation of the moving finger 43, and check whether the opening and closing actions of the moving finger 43 and the static finger 42 are smooth, and whether there is any jamming or interference.

[0075] Then, perform load debugging. Put in a simulated large ball and test the performance of the drive assembly during actual operation. Observe the crushing effect of the moving finger 43 on the large ball, and optimize and adjust the drive assembly according to the test results. For example, adjust the air pressure of the cylinder 45 to change the movement speed and force of the piston rod; finely adjust the positions and angles of the connecting rod 46 and the mounting rod 47 to improve the matching accuracy between the moving finger 43 and the static finger 42. After multiple debuggings and optimizations, ensure that the drive assembly can work stably and reliably.

[0076] During the actual operation, when the six-degree-of-freedom robotic arm 2 moves to the designated position according to the instructions of the control unit, the piston rod of the cylinder 45 of the driving component is in the extended state, and the moving finger 43 and the static finger 42 are in the open state, without interfering with the surrounding equipment and pellets. When the robotic arm reaches the position of the large ball, the control unit issues an instruction, and the piston rod of the cylinder 45 contracts. The contraction of the piston rod drives the mounting rod 47 to rotate through the connecting rod 46, so that the moving finger 43 rotates quickly and inserts into the gap between the static fingers 42 to break the large ball. During 200 consecutive crushing operations, the action accuracy rate of the moving finger 43 reached 99%, without misoperation or jamming. This fully proves that the driving component has a simple and reliable structure, can accurately control the movement of the moving finger 43, effectively complete the task of breaking large balls, and improve the efficiency and quality of pelletizing production.

[0077] In another technical solution, it further includes a double eccentric wheel vibrator 8, which is installed on the mounting seat 41 through a rubber spring composite shock-absorbing layer 9, and the static finger 42 is installed on the double eccentric wheel vibrator 8. Among them, the vibration direction of the double eccentric wheel vibrator 8 is set to vibrate obliquely at 45° relative to the length direction 45 of the static finger 42. When the moisture content of the qualified balls is less than 10%, the vibration parameters of the double eccentric wheel vibrator 8 are set as follows: the frequency is 20 Hz and the amplitude is 2 mm; When the moisture content of the qualified balls is not less than 10%, the vibration parameters of the double eccentric wheel vibrator 8 are set as follows: the frequency is 12 Hz and the amplitude is 5 mm.

[0078] In the above technical solution, aiming at the problem that qualified balls are easily broken by mistake, the double eccentric wheel vibrator 8 works first before breaking the large balls, and uses different vibration parameters to screen out the qualified balls from the gap of the static fingers 42, only retaining the large balls for breaking, effectively avoiding unnecessary damage to the qualified balls, significantly improving the qualification rate of the products, and reducing resource waste.

[0079] Regarding the impact of vibration on the stability and service life of the equipment, the rubber spring composite shock-absorbing layer 9 plays a good buffering and shock-absorbing role. It can absorb and disperse the vibration energy generated by the double eccentric wheel vibrator 8, greatly reducing the vibration intensity transmitted to the mounting seat 41 and other equipment components, ensuring the stable operation of the equipment, extending the service life of the equipment, and reducing the maintenance cost.

[0080] Considering the influence of different moisture contents of qualified balls on screening, it is very crucial to adjust the vibration parameters according to the moisture content. When the moisture content is low, high frequency and small amplitude are adopted, and when the moisture content is high, low frequency and large amplitude are adopted, which can better adapt to the characteristics of qualified balls with different moisture contents, improve the screening efficiency, and ensure the reliability of the screening effect.

[0081] Connection between the static finger 42 and the double eccentric wheel vibrator 8: The static finger 42 is installed on the double eccentric wheel vibrator 8, and the connection method should ensure that the static finger 42 can fully receive the vibration transmitted by the vibrator.

[0082] Setting and adjustment of vibration parameters: According to the moisture content of the qualified balls, accurately set the vibration parameters of the double eccentric wheel vibrator 8. During the production process, regularly detect the moisture content of the qualified balls. When it is detected that the moisture content of the qualified balls is less than 10%, set the frequency of the double eccentric wheel vibrator 8 to 20 Hz and the amplitude to 2 mm; when it is detected that the moisture content of the qualified balls is not less than 10%, set the frequency to 12 Hz and the amplitude to 5 mm.

[0083] When setting the vibration parameters, the operation can be carried out through the control system of the vibrator. The control system generally has functions of parameter display and adjustment, and the operator can make corresponding settings according to the detection results. After the setting is completed, a trial run should be carried out to observe the screening effect of the qualified balls. If the screening effect is not ideal, the vibration parameters should be further fine-tuned according to the actual situation until the best screening effect is achieved.

[0084] Control of the screening and crushing process: Before crushing the large balls, first start the double eccentric wheel vibrator 8 and screen the qualified balls according to the set vibration parameters. During the screening process, observe the flow condition of the pellets and the screening effect. When the qualified balls are basically screened out, turn off the double eccentric wheel vibrator 8. Then, control the six-degree-of-freedom robotic arm 2 to move the crushing hand 4 to the crushing area. After reaching the crushing area, start the drive assembly to insert the moving finger 43 into the gap between the static fingers 42 to crush the large balls. After the crushing is completed, the robotic arm returns to the initial position and waits for the next screening and crushing task.

[0085] During the entire screening and crushing process, ensure that the actions of each component are coordinated and accurate. An automated operation can be achieved by writing a control program to improve production efficiency and accuracy. At the same time, the operating status of the equipment should be monitored in real time to detect and handle abnormal situations in a timely manner.

[0086] During the production process, regularly detect the moisture content of the qualified balls. When it is detected that the moisture content of the qualified balls is less than 10%, set the frequency of the double eccentric wheel vibrator 8 to 20 Hz and the amplitude to 2 mm. After starting the vibrator, the qualified balls are quickly screened out from the gaps between the static fingers 42 under the action of vibration, while the large balls remain on the static fingers 42. After a period of screening, the proportion of large balls mixed in the screened qualified balls is extremely low, and the effective separation of qualified balls and large balls is basically achieved.

[0087] When the moisture content of the qualified balls rises to no less than 10%, adjust the frequency of the vibrator to 12 Hz and the amplitude to 5 mm. At this time, although the fluidity of the qualified balls becomes worse due to the increase in moisture, under the action of appropriate vibration parameters, they can still be smoothly screened out from the gaps between the static fingers 42.

[0088] After the screening is completed, turn off the double eccentric wheel vibrator 8, and the robotic arm moves the crushing hand 4 to the crushing area. Start the drive assembly, and the moving finger 43 accurately inserts into the gap between the static fingers 42 to crush the large balls. After multiple tests, the crushing rate of the large balls reaches more than 95%, and at the same time, the breakage rate of the qualified balls is controlled within 0.5%.

[0089] In another technical solution, the rubber spring composite shock-absorbing layer 9 includes a bottom plate 91, a telescopic sleeve 92 arranged on the bottom plate 91, a spring 93 sleeved inside the telescopic sleeve 92, and a top plate 94 arranged on the top of the telescopic sleeve 92. Among them, both ends of the spring 93 are fixed to the bottom plate 91 and the top plate 94 respectively. A rubber layer is laid on the outer surface of the spring 93, a buffer layer 95 is laid on the bottom surface and the top surface of the top plate 94, and the double eccentric wheel vibrator 8 is installed on the top plate 94.

[0090] In the above technical solution, the rubber spring composite shock-absorbing layer 9 plays a significant role in solving the problem of vibration damage to the equipment. The spring 93 can absorb and buffer the vibration energy, and the telescopic sleeve 92 ensures the stability of the spring 93 during the telescopic process. The rubber layer further enhances the shock-absorbing effect and disperses and consumes the vibration energy. The buffer layers 95 above and below the top plate 94 can also reduce the impact force, effectively reducing the vibration intensity transmitted to the mounting base 41 and the surrounding equipment, avoiding damage to the equipment components due to vibration, and extending the service life of the equipment.

[0091] Regarding the problem of vibration generating noise, the rubber layer and the buffer layer 95 have good sound insulation and sound absorption properties. They can reduce the transmission of structure-borne noise caused by vibration, lower the noise level in the working environment, create a relatively quiet working environment for the operators, and protect their physical health.

[0092] Regarding the problem of vibration affecting the stability and working accuracy of the equipment, this shock-absorbing layer makes the overall equipment more stable through effective shock absorption. A stable working state helps to ensure the working accuracy of the large ball automatic crushing device, improve the crushing efficiency and product quality, and ensure that the equipment can operate continuously and stably.

[0093] Installation of the rubber spring composite shock-absorbing layer 9 and the double eccentric wheel vibrator 8: Install the assembled rubber spring composite damping layer 9 on the mounting seat 41 through the bottom plate 91 and fix it with bolts. The tightening torque shall meet the design requirements. Then install the double eccentric wheel vibrator 8 on the top plate 94 and also fix it with bolts. During the installation process, ensure that the installation position of the double eccentric wheel vibrator 8 is accurate, with a deviation not exceeding ±1 mm. After installation, check whether the connection between the damping layer and the double eccentric wheel vibrator 8 is tight and there is no loosening.

[0094] In another technical solution, the image analysis algorithm specifically includes the following steps: S1. Convert the received image data into a grayscale image and the HSV color space. Based on the distribution differences of the hue H, saturation S, and brightness V of the sphere and the material layer in the HSV color space, set the three-channel joint threshold range for dynamic segmentation to generate a preliminary binary mask. S2. Extract the brightness channel of the HSV color space, calculate the brightness mean μ and variance σ² of the preliminary segmentation region, and set the variance threshold σ th , and determine the segmentation region that satisfies σ²>σ th as the dynamic material layer, and retain the smooth region where σ²≤σ th as the candidate sphere set. S3. Calculate the image contrast Contrast using the gray level co-occurrence matrix. The texture feature is the contrast. According to the texture difference between the sphere and the material layer, preset the contrast threshold C th , and segment the image to generate a texture feature binary mask. Among them, the region with a contrast higher than the contrast threshold C th is the material layer region, and the region not higher than the contrast threshold C th is the candidate sphere set. Among them, P ( i , j ) is the joint probability of the pixel pair ( i , j ) in the gray level co-occurrence matrix, and N is the number of gray levels; S4. Fuse the color and texture feature masks, and select the pixels with gray values in the range of [μ - 2σ, μ + 2σ] and texture classification as spheres as the initial seed points. Set the dynamic similarity criterion and iteratively merge adjacent similar pixels until convergence to output the refined sphere region segmentation result. Among them, the pixel difference threshold allowed by the dynamic similarity criterion is: gray difference ΔGray≤15, HSV color distance ΔE≤8, and Euclidean distance of texture feature vector ΔF≤0.2; S5. Apply Canny edge detection to the segmented sphere regions to extract closed contours, calculate the contour convex hull and geometric parameters. The geometric parameters include area A, perimeter L, and shape factor SF = 4πA / L², where the area A is the number of pixels. Set the geometric constraint conditions as an area threshold and shape factor SF ≥ 0.80. Contours that meet the geometric constraint conditions are determined to be valid spheres. S6. Based on the sphere projection area and spatial geometric relationship, establish an equivalent diameter model and calculate the equivalent diameter. D eq , perform trajectory association on the detection results of consecutive multiple frames, and use Kalman filtering to track the sphere motion trajectory. The number of stable trajectories is the total number of real-time spheres. Among them, A proj is the sphere region projection area, which is converted by pixel area × single-pixel physical size. k is the ball disk inclination compensation coefficient, which is a pre-calibrated parameter, k = cosθ, where θ is the angle between the disk axis and the camera optical axis. S7. When the detected equivalent diameter D eq is greater than the preset diameter threshold D th , and the number of large balls is greater than the preset number threshold, trigger the large ball coordinate reporting and fragmentation command.

[0095] In the above technical solution, to address the problem that it is difficult to accurately distinguish between spheres and the material layer, this image analysis algorithm converts the image data into the HSV color space, uses the distribution differences of hue, saturation, and brightness for dynamic segmentation, and combines the gray-level co-occurrence matrix to calculate the texture contrast. By integrating color and texture features, it can more accurately distinguish between spheres and the material layer, improve the accuracy of large ball recognition, and provide a reliable basis for subsequent fragmentation operations.

[0096] For the problem of insufficient refinement in sphere region segmentation, the algorithm selects pixels in a specific gray level range and classified as spheres in terms of texture as initial seed points by fusing color and texture feature masks, and sets strict dynamic similarity criteria for iterative merging. This effectively reduces the influence of noise and interference, achieves refined segmentation of the sphere region, avoids over-segmentation and under-segmentation, and makes the judgment of the number and position of large balls more accurate.

[0097] Regarding the problem of sphere motion trajectory tracking, the algorithm establishes an equivalent diameter model based on the projected area of the sphere and the spatial geometric relationship, and uses Kalman filtering to perform trajectory association and tracking on the detection results of multiple consecutive frames. It can stably and accurately track the motion trajectory of the sphere, obtain the number and position information of the large spheres in real time, provide strong support for the real-time control of the large sphere automatic crushing device, and improve the automation degree and working efficiency of the device.

[0098] Specifically, the implementation of step S1: In practical applications, the image data of the area of the pelletizing disc 3 collected by the image acquisition module 1 is first transmitted to the control unit. After receiving the image data, the control unit uses the corresponding image processing functions to convert the image data into a grayscale image and the HSV color space. This conversion process can be achieved by calling functions in image processing libraries such as OpenCV.

[0099] After converting to the HSV color space, it is necessary to analyze the distribution differences between the spheres and the material layer in the three channels of hue H, saturation S, and brightness V. This can be achieved by statistically analyzing a large number of image samples and analyzing the pixel value distribution ranges of the spheres and the material layer in different channels. For example, by analyzing 1000 images containing spheres and the material layer, the pixel value ranges of the spheres and the material layer in the hue, saturation, and brightness channels are determined.

[0100] Based on these analysis results, a reasonable three-channel joint threshold range is set. The setting of this threshold range requires multiple experiments and adjustments to ensure that the spheres and the material layer can be accurately distinguished. For example, through experiments, it is found that when the threshold range of hue H is set to [0, 30], the threshold range of saturation S is set to [50, 150], and the threshold range of brightness V is set to [80, 200], the preliminary segmentation of the spheres and the material layer can be better achieved. After setting the threshold range, the image is dynamically segmented to generate a preliminary binary mask.

[0101] The implementation of step S2: After generating the preliminary binary mask, the brightness channel of the HSV color space is extracted. This operation can also be achieved by using functions in the image processing library. After extracting the brightness channel, the preliminary segmentation area is divided into multiple sub-regions, and the brightness mean μ and variance σ² of each sub-region are calculated.

[0102] When calculating the brightness mean μ, the brightness values of all pixels in the sub-region are added up and then divided by the total number of pixels in the sub-region. When calculating the variance σ², according to the variance calculation formula, the difference between the brightness value of each pixel and the mean is squared and then averaged.

[0103] Set the variance threshold σ th , and the setting of this threshold needs to be adjusted according to the actual situation. For example, through the analysis and experiments of a large number of images, it is found that when the variance threshold σth When set to 50, it can better distinguish the dynamic material layer and the candidate sphere set. For the segmentation region that satisfies σ²>σ th is determined as the dynamic material layer, and the smooth region where σ²≤σ th is reserved as the candidate sphere set.

[0104] Step S3 implementation: The gray-level co-occurrence matrix (GLCM) is used to calculate the image contrast Contrast. First, determine the calculation parameters of the gray-level co-occurrence matrix, such as the number of gray levels N, the offset, etc. The number of gray levels N is generally determined according to the gray range and accuracy requirements of the image, and usually takes values such as 16, 32, or 64. The offset can be selected with different directions and distances according to the actual situation, such as the horizontal direction, the vertical direction, the diagonal direction, etc.

[0105] When calculating the gray-level co-occurrence matrix, count the occurrence frequency of pixel pairs (i,j) that satisfy a specific offset in the image to obtain the gray-level co-occurrence matrix P(i,j). Then, calculate the image contrast Contrast according to the calculation formula of the contrast.

[0106] Preset the contrast threshold C th based on the texture difference between the sphere and the material layer. The setting of this threshold also needs to be determined through the analysis and experiment of a large number of images. For example, through experiments, it is found that when the contrast threshold C th is set to 0.3, it can better distinguish the material layer region and the candidate sphere set. Segment the image to generate a texture feature binary mask, where the contrast is higher than the contrast threshold C th is the material layer region, and the region not higher than the contrast threshold C th is the candidate sphere set.

[0107] Step S4 implementation: Fuse the color and texture feature masks, that is, merge the preliminary binary mask and the texture feature binary mask. The fusion of the masks can be achieved through logical operations, such as AND operation, OR operation, etc.

[0108] Select the pixels with gray values in the interval [μ - 2σ, μ + 2σ] and the texture classification as spheres as the initial seed points. The setting of this interval is based on statistical principles and can effectively select pixel points similar to the sphere characteristics. Set the dynamic similarity criteria, including the gray difference ΔGray≤15, the HSV color distance ΔE≤8, and the Euclidean distance of the texture feature vector ΔF≤0.2.

[0109] According to the dynamic similarity criterion, iteratively merge adjacent similar pixels. In each iteration, check the similarity between adjacent pixels and the initial seed points. If the similarity criterion is met, merge the pixel into the sphere region. Repeat this process until no new pixels can be merged, and output the refined sphere region segmentation result.

[0110] Implementation of step S5: For the segmented sphere region, use Canny edge detection to extract the closed contour. The Canny edge detection algorithm is a classic edge detection algorithm. Through steps such as Gaussian filtering, calculating the gradient magnitude and direction, non-maximum suppression, and double-threshold detection, it can accurately extract the edge information in the image.

[0111] After extracting the closed contour, calculate the convex hull and geometric parameters of the contour. When calculating the area A, count the number of pixels inside the contour; when calculating the perimeter L, traverse the pixel points on the contour, calculate the distance between adjacent pixel points and sum them; calculate the shape factor SF = 4πA / L².

[0112] Set the geometric constraint conditions as the area threshold and the shape factor SF ≥ 0.80. The setting of the area threshold needs to be adjusted according to the actual situation. For example, when the area threshold is set to 100 pixels, it can better screen out the spheres that meet the requirements. The contour that meets the geometric constraint conditions is determined as a valid sphere.

[0113] Implementation of step S6: Based on the sphere projection area and spatial geometric relationship, establish an equivalent diameter model. First, calculate the sphere region projection area Aproj, which is obtained by converting the pixel area × the physical size of a single pixel. The physical size of a single pixel can be determined through camera calibration.

[0114] Determine the ball disk inclination compensation coefficient k, k = cosθ, where θ is the angle between the disk axis and the camera optical axis. This angle can be obtained through measurement or pre-calibration. Calculate the equivalent diameter Deq according to the equivalent diameter model.

[0115] Perform trajectory association on the detection results of consecutive multiple frames. A method based on feature matching can be used, such as matching based on shape features, color features, etc., to associate the spheres in different frames. Use Kalman filtering to track the sphere motion trajectory. Kalman filtering is a commonly used state estimation method. Through the two steps of prediction and update, it can effectively track the sphere motion trajectory. The number of stable trajectories is the total number of real-time spheres.

[0116] Implementation of step S7: Set a preset diameter threshold D th and a preset quantity threshold. The setting of these thresholds needs to be determined according to the actual production requirements and the definition of large balls. For example, when the preset diameter threshold D thWhen it is set to 50 mm and the preset quantity threshold is set to 5, it can meet the demand for large ball detection in production.

[0117] When it is detected that the equivalent diameter Deq is greater than the preset diameter threshold D th , and the number of large balls is greater than the preset quantity threshold, the large ball coordinate reporting and crushing instruction are triggered. The control unit sends the coordinate information of the large ball to the joint drive assembly of the six-degree-of-freedom robotic arm 2, and the robotic arm moves to the specified position according to the coordinate information, and drives the crushing hand 4 to crush the large ball.

[0118] After running for a period of time, the image analysis algorithm can accurately identify large balls, and the recognition accuracy rate of large balls reaches more than 99%, which is 10% higher than before, effectively improving the working efficiency and accuracy of the large ball automatic crushing device, reducing manual intervention, and reducing the labor intensity.

[0119] In another technical solution, the path planning algorithm specifically includes the following steps: Step 1: Taking the pelletizing disc 3 as the center, determine the working space range of the robotic arm, establish a three-dimensional map model, and mark the positions and shape information of the pelletizing disc 3, the feeding chute, the robotic arm base, and the obstacles in the map; Step 2: Taking the initial position of the robotic arm as the starting point, calculate the optimal target point coordinates for the robotic arm to grasp the large ball according to the position of the large ball; Step 3: Adopt a path search algorithm to search for the optimal path from the starting point to the target point for grasping the large ball, then to the target point above the feeding chute, and finally back to the starting point in the constructed map Step 4: Optimize the searched optimal path, remove the redundant nodes in the path, perform smoothing processing, and send the optimized and smoothed path information to the joint drive assembly.

[0120] In the above technical solution, for the problems of determining the working space range of the robotic arm and avoiding collisions, by establishing a three-dimensional map model with the pelletizing disc 3 as the center and marking the positions and shape information of each object, the robotic arm can clearly understand the working environment, plan the movement path in advance, effectively avoid collisions with surrounding objects, ensure the safe operation of the equipment and the safety of personnel, and at the same time reduce the damage and maintenance costs caused by collisions of the equipment.

[0121] For the problem of calculating the optimal target point coordinates for the robotic arm to grasp the large ball, taking the initial position of the robotic arm as the starting point and calculating in combination with the position of the large ball can ensure that the robotic arm grasps the large ball with the optimal posture and position, improve the grasping efficiency and accuracy, reduce the ineffective movement, save time and energy, and make the large ball crushing operation more efficient.

[0122] Regarding the problem of searching for the optimal path to reduce motion time and energy consumption, a path search algorithm is used to search for the path from the starting point to each target point and then back to the starting point in the map. A relatively shortest and most reasonable path can be found, enabling the robotic arm to move along this path, greatly reducing the motion time and energy consumption, improving the work efficiency, and at the same time reducing the operating cost of the equipment.

[0123] Regarding the traditional path algorithm which involves many complex search and evaluation processes, such as traversing and calculating a large number of unnecessary nodes. In this method, the simplified path algorithm, through in-depth analysis of the workspace and production scenario of the robotic arm, preset the starting position, etc. For example, according to the fixed positions and shapes of the pelletizing disc 3, the feeding chute, and the obstacles, the feasible area and the prohibited area for the movement of the robotic arm are determined. During path search, the nodes that are clearly not within the feasible area are directly excluded, reducing the search scope and thus avoiding the calculation of these nodes. At the same time, for some path parts that can be determined by simple geometric relationships or empirical formulas, complex algorithm calculations are no longer performed, but the pre-determined optimal solution is directly adopted, further reducing the amount of calculation.

[0124] Regarding the problem that the path has redundant nodes resulting in unsmooth motion, the optimal path searched is optimized, removing the redundant nodes and performing smoothing processing, making the motion path of the robotic arm more concise and smooth, improving the motion performance of the robotic arm, reducing the jitter and impact during the movement of the robotic arm, and extending the service life of the robotic arm.

[0125] Specifically, step one is implemented: Determine the workspace range of the robotic arm: First, use measuring tools, such as laser rangefinders, total stations, etc., to accurately measure the pelletizing disc 3, the feeding chute, the robotic arm base, and the possible obstacles. The measurement content includes the dimensions (length, width, height) of each object and the relative positions (distance from the center of the pelletizing disc 3, angle, etc.).

[0126] For the robotic arm, through its kinematic model and the actual motion range, determine its maximum extension range and minimum contraction range in space to define the workspace range of the robotic arm. For example, each joint of the six-degree-of-freedom robotic arm 2 has a certain rotation angle range, and by calculating the combination of these angle ranges in space, the reachable space of the robotic arm is obtained.

[0127] Establish a 3D map model: Using professional 3D modeling software such as AutoCAD, SolidWorks, etc., based on the measured data, with the pelletizing disc 3 as the center, establish a 3D map model. During the modeling process, accurately draw the 3D models of the pelletizing disc 3, the feeding chute, the robotic arm base, and the obstacles according to the actual dimensions and positional relationships. For the convenience of subsequent path planning, number and label each object, and record its position and shape information. For example, label information such as the diameter and height of the pelletizing disc 3, the inclination angle and length of the feeding chute, etc.

[0128] Label position and shape information: In the 3D map model, use different colors, lines, or symbols to distinguish different objects, and detail their position and shape information. For example, use red lines to represent obstacles, blue lines to represent the contour of the pelletizing disc 3, and at the same time record the specific dimensions and position coordinates of each object next to the model or in a separate document.

[0129] Implementation of Step Two: Obtain the initial position of the robotic arm: In the control system of the robotic arm, through sensors such as encoders installed at the joints, obtain the angle information of each joint of the robotic arm in real time. According to the kinematic model of the robotic arm, convert this joint angle information into the coordinates of the end of the robotic arm in 3D space, thereby determining the initial position of the robotic arm and taking it as the starting point.

[0130] Obtain the position of the large ball: The image acquisition module 1 acquires the image data of the area of the pelletizing disc 3. After being processed by the image analysis algorithm in the control unit, obtain the position information of the large ball in the image. Then, according to the calibration parameters of the camera, convert the position information in the image into the 3D coordinates in the actual space to determine the position of the large ball.

[0131] Calculate the coordinates of the optimal target point: According to the structure and motion characteristics of the robotic arm, as well as the position of the large ball, use mathematical algorithms to calculate the coordinates of the optimal target point for the robotic arm to grasp the large ball. For example, the inverse kinematics algorithm can be used, combined with the kinematic model of the robotic arm, to calculate the optimal posture and coordinates of the end of the robotic arm reaching the position of the large ball under the condition of meeting the grasping requirements. During the calculation process, factors such as the joint limits and motion ranges of the robotic arm should be considered to ensure that the calculated target point coordinates are feasible.

[0132] Implementation of Step Three: Select a path search algorithm: Common path search algorithms include the A* algorithm, Dijkstra algorithm, etc. Select a suitable path search algorithm according to the actual situation. For example, the A* algorithm combines the advantages of heuristic search and graph search and can find the optimal path from the starting point to the target point in a relatively short time.

[0133] Searching for a path in the map: Input the starting point, the target point for grasping the large ball, the target point above the blanking chute, and the starting point (return point) into the path search algorithm. The algorithm searches in the constructed 3D map model. By continuously expanding nodes, calculating the distances between nodes and the heuristic function values, it gradually finds the path from the starting point to the target point for grasping the large ball, then to the target point above the blanking chute, and finally back to the starting point.

[0134] During the search process, the algorithm will avoid the areas marked as obstacles to ensure that the searched path is feasible. At the same time, record the coordinate information of each node on the searched path.

[0135] Implementation of Step Four: Removing redundant nodes: Analyze the nodes on the searched path to determine which nodes are redundant. For example, if the path between two nodes is a straight line and there are no obstacles, then the intermediate nodes between these two nodes may be redundant. By calculating the distances and directions between nodes, remove these redundant nodes to make the path more concise.

[0136] Performing smoothing processing: Use methods such as curve fitting to smooth the path. For example, spline curves can be used to fit the nodes on the path to make the path smoother and reduce the sudden change in acceleration during the movement of the robotic arm. During the smoothing process, ensure that the path still satisfies the kinematic constraints and workspace limitations of the robotic arm.

[0137] Sending path information: Organize the optimized and smoothed path information, convert it into a format that the joint drive component can recognize, and then send it to the joint drive component. The joint drive component controls the movement of each joint of the robotic arm according to the received path information, enabling the robotic arm to operate along the planned path.

[0138] One of the embodiments: First, use tools such as laser rangefinders and total stations to accurately measure the pelletizing disc 3 (with a diameter of 6 meters and a height of 1.5 meters), the blanking chute (with an inclination angle of 30 degrees and a length of 4 meters), the base of the six-degree-of-freedom robotic arm 2, and the surrounding obstacles. Then, use AutoCAD software to establish a 3D map model centered on the pelletizing disc 3 and detailedly mark the position and shape information of each object.

[0139] When the image acquisition module 1 detects a large ball in the pelletizing disc 3, the control unit obtains the initial position coordinates (1, 1, 1) (unit: meters) of the robotic arm. At the same time, according to the image analysis algorithm, it obtains the position coordinates (3, 3, 0.5) of the large ball. Calculate the optimal target point coordinates (3, 3, 0.6) for the robotic arm to grasp the large ball through the inverse kinematics algorithm to ensure that the robotic arm can stably grasp the large ball.

[0140] Next, the A* algorithm is used to search for a path in the three-dimensional map model, from the starting point (1, 1, 1) to the target point for grasping the large ball (3, 3, 0.6), then to the target point above the blanking chute (4, 2, 1), and finally back to the starting point (1, 1, 1). The algorithm successfully avoids the obstacles marked on the map and finds a feasible path.

[0141] Finally, the searched path is optimized by removing 5 redundant nodes and smoothed using a spline curve. The optimized and smoothed path information is sent to the joint drive component, and the robotic arm accurately moves to the target position according to the planned path, successfully grasping and crushing the large ball, and then safely returning to the starting point. After multiple tests, this path planning algorithm can effectively plan the movement path of the robotic arm, improving the efficiency and safety of large ball crushing.

[0142] Although the embodiments of the present invention have been disclosed as above, it is not limited to the applications listed in the specification and embodiments. It can be fully applied to various fields suitable for the present invention. For those familiar with the field, additional modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and the illustrated and described examples here.

Claims

1. A large ball automatic crushing device for a disc ball making machine, characterized in that: include: An image acquisition module, which is used to acquire image data of the ball-making disk area; a control unit, which is communicatively connected to the image acquisition module and has a built-in image analysis algorithm and a three-dimensional path planning algorithm based on machine vision, wherein the image analysis algorithm is configured to perform multi-target detection on the received image data, identify the contour of the sphere through an edge detection algorithm, and calculate the equivalent diameter according to the projection area of ​​the sphere; when it is detected that the equivalent diameter exceeds a preset threshold and the number of consecutive frames reaches a set value, it is determined that there is a large ball, and the number of large balls is counted; when the number of large balls reaches a preset threshold, the three-dimensional coordinate information of the large ball area is generated; and the three-dimensional path planning algorithm is configured to calculate the obstacle avoidance motion trajectory based on the three-dimensional coordinate information of the large ball area and the preset starting point position and the broken area position; A six-degree-of-freedom robotic arm is arranged near the ball-making disk area, a joint drive assembly of the six-degree-of-freedom robotic arm is connected to the control unit by signal, and performs spatial positioning based on an obstacle avoidance motion trajectory in response to a trigger instruction generated by the control unit; The crushing hand comprises a mounting base mounted on the end of the six-degree-of-freedom robotic arm, a plurality of static fingers mounted on the mounting base, a driving component mounted on the mounting base, and a plurality of dynamic fingers driven to rotate by the driving component, wherein when the driving component drives the plurality of dynamic fingers to rotate, the plurality of dynamic fingers are inserted into the gaps between the plurality of static fingers in a one-to-one correspondence.

2. The large ball automatic crushing device for disc ball making machine according to claim 1, characterized in that: The static finger body is in the shape of a rod, with a bent end, and blocking rods are arranged on both sides of the outermost part.

3. The automatic large ball crushing device for a disc ball making machine as claimed in claim 1, characterized in that: Both side walls of the static finger are provided with a comb-tooth structure, the tooth roots of the comb-tooth structure are provided with triangular reinforcing ribs, and the comb-tooth gap G of the comb-tooth structure is set as: G=d max +16mm, and G<D min , where d max is the qualified ball diameter, D min is the diameter of the large ball; Both side walls of the moving finger are provided with sawtooth structures corresponding to the comb-tooth structure, the tip of the sawtooth structure faces the comb-tooth structure, and the head of the sawtooth structure is in a double-bevel wedge shape with an included angle of 30°.

4. The automatic large ball crushing device for a disc ball making machine as claimed in claim 1, characterized in that: The driving assembly includes a cylinder installed on the mounting seat, a connecting rod hinged to the piston rod of the cylinder, and a mounting rod hinged to the other end of the connecting rod, wherein the moving finger is fixed on the mounting rod, and the piston rod is in an extended state as an initial position. In the initial position, the moving finger and the static finger are in an open state.

5. The automatic large ball crushing device for disc ball making machine as claimed in claim 3, characterized in that: It also includes a double eccentric vibrator, which is mounted on the mounting seat through a rubber spring composite shock-absorbing layer, and the static finger is mounted on the double eccentric vibrator, wherein the vibration direction of the double eccentric vibrator is set to vibrate at an oblique angle of 45° relative to the length direction of the static finger; When the moisture content of the qualified ball is less than 10%, the vibration parameters of the double eccentric vibrator are set as follows: frequency is 20 Hz and amplitude is 2 mm; When the moisture content of the qualified balls is not less than 10%, the vibration parameters of the double eccentric vibrator are set as follows: frequency is 12 Hz and amplitude is 5 mm.

6. The automatic large ball crushing device for a disc ball making machine as claimed in claim 5, characterized in that: The rubber spring composite shock-absorbing layer includes a bottom plate, a telescopic sleeve arranged on the bottom plate, a spring sleeved in the telescopic sleeve, and a top plate arranged on the top of the telescopic sleeve, wherein the two ends of the spring are respectively fixed to the bottom plate and the top plate, the surface of the spring is coated with a rubber layer, the bottom surface of the top plate and the top surface of the top plate are coated with a buffer layer, and the double eccentric wheel vibrator is installed on the top plate.

7. The automatic large ball crushing device for a disc ball making machine as claimed in claim 1, characterized in that: The image analysis algorithm specifically comprises the following steps: S1. Convert the received image data into a grayscale image and HSV color space, and set a three-channel joint threshold range for dynamic segmentation based on the distribution difference of hue H, saturation S, and brightness V between the sphere and the material layer in the HSV color space to generate a preliminary binary mask; S2. Extract the brightness channel of the HSV color space, calculate the brightness mean μ and variance σ² of the preliminary segmented area, and set the variance threshold σ th , for which σ²>σ th The segmented area is determined as a dynamic material layer, and σ²≤σ th The smooth area is taken as the candidate sphere set; S3. Use gray-level co-occurrence matrix to calculate image contrast. The texture feature is contrast. The contrast threshold is preset based on the texture difference between the sphere and the material layer. C th , segment the image and generate a binary mask of texture features, where the contrast is higher than the contrast threshold C th For the material layer area, not higher than the contrast threshold C th is a set of candidate spheres; in, P ( i , j ) is the pixel pair in the gray-level co-occurrence matrix ( i , j ), N is the gray level; S4, fuse the color and texture feature masks, select the pixels whose grayscale value is in the interval [μ-2σ, μ+2σ] and whose texture is classified as a sphere as the initial seed point, set the dynamic similarity criterion, iteratively merge adjacent similar pixels until convergence, and output the refined sphere region segmentation result, where the pixel difference thresholds allowed by the dynamic similarity criterion are: grayscale difference ΔGray≤15, HSV color distance ΔE≤8, and texture feature vector Euclidean distance ΔF≤0.2; S5. Use Canny edge detection to extract closed contours from the segmented sphere area, calculate the contour convex hull and geometric parameters, the geometric parameters include area A, perimeter L, shape factor SF=4πA / L², where area A is the number of pixels, and set geometric constraints as area threshold and shape factor SF≧0.80; the contour that meets the geometric constraints is determined to be a valid sphere; S6. According to the relationship between the projected area of ​​the sphere and the spatial geometry, an equivalent diameter model is established to calculate the equivalent diameter. D eq , the trajectory association of continuous multi-frame detection results is performed, and the Kalman filter is used to track the trajectory of the ball. The number of stable trajectories is the total number of real-time balls; in, A proj is the projected area of ​​the sphere, which is converted by pixel area × single pixel physical size. k is the compensation coefficient of the ball disk tilt angle, is the pre-calibrated parameter, k=cosθ, θ is the angle between the disk axis and the camera optical axis; S7, when the equivalent diameter is detected D eq Greater than the preset diameter threshold D th , and when the number of large balls is greater than the preset number threshold, the large ball coordinate reporting and crushing instructions are triggered.

Images