Full-automatic large-particle gangue sorting system

Abstract

The invention discloses a full-automatic large-particle gangue automatic sorting system. The system comprises a main conveying belt, a feeding buffer unit, a multi-mode recognition unit, a flexible grabbing unit and an ore collecting unit. The feeding buffer unit is used for screening raw coal and conveying large-particle materials to the main conveying belt. The multi-mode recognition unit is erected above the main conveying belt and used for recognizing material attributes and outputting space coordinate information of the material attributes. The flexible grabbing unit stretches across the main conveying belt and is used for grabbing and sorting the target materials according to the output information of the multi-mode recognition unit. The ore collecting unit comprises a gangue collecting bin and a coal collecting bin which are arranged on the two sides of the main conveying belt correspondingly. The method has extremely high sorting precision, and gangue can be recognized in a multi-mode mode; according to the method, the sorting efficiency is remarkably improved, the labor burden can be reduced, meanwhile, the labor cost and the safety risk can be reduced, advanced concepts such as digital twinning and predictive maintenance are integrated, and core data support and decision basis are provided for an intelligent mine.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of mineral sorting, and particularly relates to a full-automatic large-particle gangue automatic sorting system. BACKGROUND

[0002] In the process of coal mining, gangue as the main associated impurities, its efficient separation is the key link to improve the quality of commercial coal, reduce transportation cost and realize green mine construction. At present, the sorting technology for medium and small particle size (<50mm) coal gangue has been relatively mature, such as heavy medium separation, jigging separation and intelligent dry sorting machine based on RGB vision. However, for large-particle gangue with particle size greater than 250mm, the existing technology has obvious deficiencies:

[0003] 1. When heavy medium, jigging and other processes are used to treat large block materials, the equipment is severely worn, the energy consumption is extremely high, the water consumption is large, and the equipment is prone to blockage failure, which is poor in economy.

[0004] 2. The sorting machine based on pneumatic injection or mechanical ejection has difficulty in generating sufficient throwing force on large mass gangue, and has poor sorting effect, and is prone to cause materials to collide and break, resulting in loss of high-quality lump coal.

[0005] 3. A large number of mines still rely on workers to manually select beside the conveyor belt, which has large labor intensity, poor working environment, low sorting efficiency and safety hazards, and does not meet the development trend of modern mines.

[0006] 4. In many areas, the surface color and texture of coal and gangue are very similar after weathering or pollution, and it is easy to misjudge only by traditional RGB vision or single sensing technology, which leads to that the sorting precision cannot meet the requirements.

[0007] Therefore, in view of the above technical problems, it is necessary to provide a full-automatic large-particle gangue automatic sorting system.

[0008] The information disclosed in this BACKGROUND section is only intended to increase an understanding of the general background of the application and should not be construed as a recognition of the prior art against which the application is defined. SUMMARY

[0009] The purpose of the present application is to provide a full-automatic large-particle gangue automatic sorting system which can solve the problems in the background art.

[0010] In order to achieve the above purpose, the technical scheme provided by an embodiment of the present application is as follows:

[0011] The full-automatic large-particle gangue automatic sorting system comprises a main conveying belt, a feeding buffer unit, a multi-modal identification unit, a flexible grabbing unit and a gangue collecting unit. The feeding buffer unit is used for screening raw coal and conveying large-particle materials to the main conveying belt. The multi-modal identification unit is arranged above the main conveying belt and is used for identifying material properties and outputting spatial coordinate information. The flexible grabbing unit is arranged across the main conveying belt and is used for grabbing and sorting target materials according to the output information of the multi-modal identification unit. The gangue collecting unit comprises gangue collecting bins and coal collecting bins arranged on both sides of the main conveying belt.

[0012] In one or more embodiments of the present application, the feeding buffer unit comprises a vibrating feeder and a double-layer screen, and the double-layer screen is connected to the vibrating feeder.

[0013] In one or more embodiments of the present application, the upper screen of the double-layer screen has a mesh size of 250-350 mm, and the lower screen has a mesh size of 40-60 mm.

[0014] In one or more embodiments of the present application, the multi-modal identification unit comprises a spectral imaging module, a 3D contour scanning module and an embedded processing platform. The spectral imaging module is used for acquiring spectral feature information of the material surface. The 3D contour scanning module is used for acquiring real-time three-dimensional point cloud data of the material. The embedded processing platform is internally provided with a pre-trained multi-modal fusion deep learning model, which is used for fusion analysis of the spectral feature information and the three-dimensional point cloud data, and outputs the attribute label of the material and the three-dimensional coordinates in the set coordinate system.

[0015] In one or more embodiments of the present application, the multi-modal fusion deep learning model is a double-branch neural network architecture, wherein the first branch is used for extracting deep features of spectral data, the second branch is used for extracting spatial geometric features of 3D point cloud data, and the features are fused and classified through a feature splicing layer and a full connection layer.

[0016] In one or more embodiments of the present application, the flexible grabbing unit comprises a portal, a high-speed parallel robot and a multi-finger adaptive hand claw. The portal is arranged across the main conveying belt. The high-speed parallel robot is slidable on the portal and can move along the axial direction. The multi-finger adaptive hand claw is connected to the end effector interface of the high-speed parallel robot through a flange.

[0017] In one or more embodiments of the present application, the multi-finger adaptive hand claw has at least three independently driven bionic knuckles, the inside of the bionic knuckles is covered with a variable stiffness friction material, and a multi-dimensional torque sensor is internally arranged, and the multi-finger adaptive hand claw can plan the envelope trajectory and grabbing force of each knuckle in real time according to the 3D contour information of the target material.

[0018] In one or more embodiments of the present application, the flexible grabbing unit is integrated with a dynamic target tracking controller, which calculates and controls the coordinated movement of the gantry, high-speed parallel robot and multi-finger adaptive hand claw based on the material movement information from the multi-modal recognition unit, to achieve predictive synchronous grabbing of dynamic materials.

[0019] In one or more embodiments of the present application, the inner wall of the gangue collecting bin and the coal collecting bin is lined with wear-resistant steel plates, the top is provided with a hydraulic buffer discharge port, and the bottom is provided with a weighing module and a pneumatic discharge gate.

[0020] In one or more embodiments of the present application, the sorting system further comprises an intelligent management and control platform, which is in communication connection with the multi-modal recognition unit, the flexible grabbing unit and the mining unit through an industrial Ethernet, for realizing system state monitoring, production data management, predictive maintenance alarm and remote model updating.

[0021] Compared with the prior art, the full-automatic large-particle gangue automatic sorting system has extremely high sorting precision, can multi-modal recognize gangue, significantly improves sorting efficiency, can reduce artificial burden, and can also reduce artificial cost and safety risk, and integrates advanced concepts such as digital twin and predictive maintenance, provides core data support and decision basis for intelligent mines. BRIEF DESCRIPTION OF DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiment or prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments described in the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0023] Figure 1 The structural block diagram of the full-automatic large-particle gangue automatic sorting system in an embodiment of the present application is shown in the figure.

[0024] Figure 2 The structural block diagram of the feeding buffer unit in an embodiment of the present application is shown in the figure.

[0025] Figure 3 The structural block diagram of the multi-modal recognition unit in an embodiment of the present application is shown in the figure.

[0026] Figure 4 The structural block diagram of the flexible grabbing unit in an embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0027] In order for those skilled in the art to better understand the technical solutions in the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described in the following with reference to the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only a part of the embodiments of the present disclosure, rather than all the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by those of ordinary skill in the art without creative work should fall within the protection scope of the present disclosure.

[0028] As shown in Figures 1 to 4 An embodiment of the automatic large-particle gangue automatic sorting system in the present application includes a main conveying belt, a feeding buffer unit, a multi-modal recognition unit, a flexible grabbing unit, and a gang collecting unit. The feeding buffer unit is used for screening raw coal and conveying large-particle materials to the main conveying belt. The multi-modal recognition unit is erected above the main conveying belt and is used for identifying material properties and outputting spatial coordinate information. The flexible grabbing unit is arranged across the main conveying belt and is used for grabbing and sorting target materials according to the output information of the multi-modal recognition unit. The gang collecting unit includes gangue collecting bins and coal collecting bins arranged on both sides of the main conveying belt.

[0029] The present application performs gradient screening on raw coal through the feeding buffer unit, so that large-particle materials meeting the particle size requirements are distributed in a single layer on the main conveying belt. The multi-modal recognition unit synchronously scans the materials on the main conveying belt, uses an AI fusion model to identify properties and accurately position, and generates sorting instructions. The flexible grabbing unit receives the sorting instructions, controls dynamic target tracking, drives the paw to move synchronously with the materials, and uses a self-adaptive force control strategy to complete grabbing of the target materials. According to the properties of the materials, the grabbed target materials are moved to the corresponding gangue collecting bins or coal collecting bins. The intelligent control platform collects and analyzes the whole process data, realizes production visualization, operation and maintenance intelligentization, and algorithm continuous optimization.

[0030] As shown in Figure 1 The feeding buffer unit includes a vibrating feeder and a double-layer screen, and the double-layer screen is connected to the vibrating feeder. The upper screen of the double-layer screen has a mesh size of 250-350 mm, and the lower screen has a mesh size of 40-60 mm. Through screening, large-particle materials can be targetedly processed, and small and medium-particle materials are prevented from entering the subsequent complex and expensive sorting process, thereby significantly reducing system load and energy consumption and improving overall sorting efficiency and economy.

[0031] As shown in Figure 2As shown, the multi-modal recognition unit includes a spectral imaging module, a 3D profile scanning module, and an embedded processing platform. The spectral imaging module is used to acquire spectral feature information of the material surface. The 3D profile scanning module is used to acquire real-time three-dimensional point cloud data of the material. The embedded processing platform is built-in with a pre-trained multi-modal fusion deep learning model, which is used to perform fusion analysis on the spectral feature information and the three-dimensional point cloud data, and output the attribute label of the material and the three-dimensional coordinates in the set coordinate system.

[0032] The present application combines the chemical composition analysis capability of spectrum and the spatial form analysis capability of 3D structure light, overcomes the problem of low recognition rate of single visual technology when the surface color and texture of coal and gangue are similar, greatly improves the accuracy and reliability of sorting, and is especially suitable for coal and gangue sorting under complex geological conditions.

[0033] Among them, the multi-modal fusion deep learning model is a double-branch neural network architecture, in which the first branch is used to extract deep features of spectral data, and the second branch is used to extract spatial geometric features of 3D point cloud data, and fusion and classification are performed through a feature splicing layer and a full connection layer. By using the powerful feature extraction and classification capability of deep learning, it can continuously adapt to the characteristics of raw coal in different mining areas and different coal types, has strong adaptability and generalization ability, and reduces the parameter workload of re-tuning for different sites.

[0034] As shown in Figure 4 As shown, the flexible grasping unit includes a portal frame, a high-speed parallel robot, and a multi-finger adaptive hand claw. The portal frame spans the main conveying belt. The high-speed parallel robot can slide on the portal frame and can move along its axial direction. The multi-finger adaptive hand claw is connected to the end effector interface of the high-speed parallel robot through a flange. The present application can realize fast tracking and accurate positioning of dynamic targets on a high-speed running conveying belt, meeting the high rhythm requirement in large-scale production.

[0035] Among them, the multi-finger adaptive hand claw has at least three independently driven bionic knuckles, the inside of the bionic knuckle is covered with a variable stiffness friction material, and a multi-dimensional torque sensor is built-in, the multi-finger adaptive hand claw can plan the envelope trajectory and grasping force of each knuckle according to the 3D profile information of the target material in real time. The adaptive hand claw can gently and firmly grasp various irregularly shaped large particle gangue or coal blocks, avoiding material breakage caused by rigid grasping or sliding caused by insufficient grasping force, significantly reducing the damage rate of the material, and ensuring the integrity and value of high-quality coal.

[0036] In addition, the flexible grabbing unit is integrated with a dynamic target tracking controller, which calculates and controls the coordinated movement of the gantry, the high-speed parallel robot and the multi-finger adaptive hand claw based on the material movement information from the multi-modal recognition unit, so as to realize predictive synchronous grabbing of dynamic materials. In this way, real dynamic grabbing is realized, the end of the mechanical arm and the material are relatively static at the moment of grabbing, the success rate of grabbing is greatly improved, impact, collision or grabbing failure caused by relative speed is avoided, and the system is still stable and reliable under the working condition of a high-speed conveying belt.

[0037] As shown in Figures 1 to 4 The inner walls of the gangue collecting bin and the coal collecting bin are lined with wear-resistant steel plates, the top is provided with a hydraulic buffer discharge port, and the bottom is provided with a weighing module and a pneumatic discharge gate. The wear-resistant steel plate reduces the risk of material crushing when falling. The weighing module is used to count the sorting output in real time, providing data support for production management. The pneumatic discharge gate facilitates connection with the subsequent transportation system to realize continuous operation.

[0038] As shown in Figure 1 The sorting system further includes an intelligent management and control platform, which is in communication connection with the multi-modal recognition unit, the flexible grabbing unit and the collecting unit through an industrial Ethernet, and is used for realizing system state monitoring, production data management, predictive maintenance alarm and model remote updating. The present application can realize intelligent operation and maintenance and digital management of the system. Remote monitoring and diagnosis function reduces the on-site maintenance demand and reduces the operation and maintenance cost; data recording and analysis provides a basis for optimizing the production process and predictive maintenance; the intelligent management and control platform enables the algorithm to be continuously iterated and upgraded, keeping the system advanced in technology.

[0039] In specific use, the raw coal is first conveyed to the feeding buffer unit by the belt conveyor. The vibrating feeder uniformly throws the material to the double-layer screen. After screening, the material with a particle size greater than 300mm (large block coal or gangue) is sent to the main conveying belt; the material with a particle size between 50mm and 300mm is shunted to the middle block sorting system; and the pulverized coal and fine coal with a particle size less than 50mm directly fall into the lower layer and are conveyed away.

[0040] The main conveyor carries the materials through the scanning area of the multi-modal recognition unit at a speed of about 1.2 m / s. The spectral imaging module acquires the continuous spectral images of the materials in the wavelength range of 400-1000 nm in a line-scan manner. At the same time, the 3D profile scanning module projects a laser line and generates a 3D point cloud with millimeter-level precision using triangulation. The dual-branch neural network model embedded in the processing platform processes these two types of data simultaneously. One branch of the model processes the spectral data cube to extract the chemical "fingerprint" features of the material, and the other branch processes the 3D point cloud to extract the geometric features such as size, shape, and surface concave-convex of the material. Subsequently, the feature vectors of the two branches are spliced in the fusion layer, and the final classification confidence and category (coal / gangue) are output through the fully connected layer and the Softmax function. At the same time, the system calculates the three-dimensional spatial coordinates (X, Y, Z) and velocity vector of the target material in the direction of motion of the main conveyor based on the 3D point cloud data.

[0041] Once a certain material is identified as gangue and the confidence is higher than the set threshold (such as 98%), the system generates a sorting instruction package containing its spatial coordinates and motion vector, and sends it to the flexible grabbing unit through high-speed Ethernet.

[0042] After receiving the instruction, the dynamic target tracking controller of the flexible grabbing unit starts working. The controller calculates an optimal motion trajectory in real time based on the current pose and velocity of the material and the kinematic parameters of the grabbing execution mechanism itself. The trajectory controls the high-speed parallel robot to slide on the gantry, and the joints of the high-speed parallel robot and the multi-finger adaptive hand claw move cooperatively, so that the hand claw can achieve nearly zero relative speed with the material at the predetermined grabbing point.

[0043] In the process of approaching the grabbing point, the multi-finger adaptive hand claw adjusts the opening and closing angles of the three bionic knuckles in advance based on the pre-acquired 3D profile of the material, forming a preliminary envelope pose. When the hand claw contacts the material, the multi-dimensional torque sensor inside the knuckle feedbacks the force in real time. The force control system of the hand claw adjusts the output torque of each knuckle motor dynamically based on the pre-set grabbing force curve, realizing stable grabbing. For example, a larger force is used to stabilize the material quickly first, and then the force is reduced to maintain the force to avoid crushing. After successful grabbing, the high-speed parallel robot cooperates with the gantry to quickly and smoothly move the gangue to the top of the gangue collection bin on one side of the main conveyor, releasing the material. If the material is identified as coal, the flexible grabbing unit does not intervene, allowing it to naturally fall into the coal collection bin at the end.

[0044] The collection bin of the mining unit is lined with wear-resistant steel plates to withstand the impact of large pieces of material. The material enters the bin through a hydraulic buffer discharge port, reducing the impact force of falling. The weighing module at the bottom records the collection amount of each bin in real time and uploads the data. When the bin reaches a certain level, the pneumatic discharge gate automatically opens to discharge the material to the downstream transportation system.

[0045] The intelligent management and control platform, as the brain of the system, is deployed on the cloud and local servers. It communicates with all underlying devices through industrial protocols such as OPCUA and builds a digital twin model of the system. On the web visualization interface, operators can monitor the device status, sorting yield, recognition accuracy, energy consumption, and other key performance indicators (KPIs) in real time. The platform's big data analysis module can analyze historical sorting data to identify bottlenecks affecting efficiency. Its machine learning algorithm can also predict potential failures of key components based on sensor data such as motor current and vibration, enabling predictive maintenance. In addition, the platform supports the return of difficult sample (low confidence sample) data packets collected on site to the cloud for expert annotation and incremental learning update of the edge AI model to optimize the system.

[0046] Those skilled in the art will appreciate that embodiments of the application can be provided as methods, systems, or computer program products. Accordingly, the application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) having computer-usable program code embodied therein.

[0047] The present application is described in reference to the flowcharts and / or block diagrams of the methods, apparatus (systems) and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as combinations of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing apparatus to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing apparatus produce a means for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 The means for implementing each flow or multiple flows and / or blocks Figure 1 The means for implementing each flow or multiple flows and / or blocks

[0048] These computer program instructions can also be stored in a computer-readable memory that can direct the computer or other programmable data processing apparatus to work in a specific manner, so that the instructions stored in the computer-readable memory produce a product including instruction means, which implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The means for implementing each flow or multiple flows and / or blocks Figure 1 The means for implementing each flow or multiple flows and / or blocks

[0049] These computer program instructions can also be loaded into a computer or other programmable data processing devices, so that a series of operational steps are performed on the computer or other programmable data processing devices to generate a computer implemented process, so that the instructions executed on the computer or other programmable data processing devices provide the function of realizing the processes specified in the flowchart Figure 1 one flow or multiple flows and / or the functions specified in the block ​ one block or multiple blocks.

[0050] It is apparent for those skilled in the art that the present disclosure is not limited to the details of the above-described exemplary embodiments, and the present disclosure can be implemented in other specific forms without departing from the spirit or essential characteristics of the present disclosure. Therefore, the embodiments should be considered in all aspects as illustrative and not restrictive, and the scope of the present disclosure is defined by the appended claims rather than the above description, and it is intended to encompass all changes falling within the meaning and range of equivalents of the elements of the claims. Any reference signs in the claims should not be considered as limiting the claims involved.

[0051] Furthermore, it should be understood that, although the present specification is described in terms of embodiments, not every embodiment contains only one independent technical solution, and the present specification is described in this way only for the sake of clarity, and those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that those skilled in the art can understand.

Claims

1. A fully automatic large-particle gangue sorting system, characterized in that, include: Main conveyor belt; The feeding buffer unit is used to screen the raw coal and convey large particles to the main conveyor belt; A multimodal recognition unit, mounted above the main conveyor belt, is used to identify material properties and output their spatial coordinate information; A flexible gripping unit is set across the main conveyor belt and is used to grip and sort target materials according to the output information of the multimodal recognition unit; The ore collection unit includes a gangue collection bin and a coal collection bin located on both sides of the main conveyor belt.

2. The fully automatic large-particle gangue sorting system according to claim 1, characterized in that, The feeding buffer unit includes a vibrating feeder and a double-layer screen, wherein the double-layer screen is connected to the vibrating feeder.

3. The fully automatic large-particle gangue sorting system according to claim 2, characterized in that, The upper mesh of the double-layer screen has a mesh size of 250-350mm, and the lower mesh has a mesh size of 40-60mm.

4. The fully automatic large-particle gangue sorting system according to claim 1, characterized in that, The multimodal recognition unit includes: The spectral imaging module is used to acquire spectral feature information of the material surface; The 3D contour scanning module is used to acquire real-time three-dimensional point cloud data of materials. The embedded processing platform has a pre-trained multimodal fusion deep learning model built in, which is used to fuse and analyze the spectral feature information and the three-dimensional point cloud data, and output the material's attribute labels and three-dimensional coordinates in a set coordinate system.

5. The fully automatic large-particle gangue sorting system according to claim 4, characterized in that, The multimodal fusion deep learning model is a dual-branch neural network architecture, in which the first branch is used to extract deep features of spectral data, and the second branch is used to extract spatial geometric features of 3D point cloud data. The features are then fused and classified through a feature splicing layer and a fully connected layer.

6. The fully automated large-particle gangue sorting system according to claim 5, characterized in that, The flexible grasping unit includes: A gantry spans across the main conveyor belt; A high-speed parallel robot can slide on the gantry and move along its axis; The multi-fingered adaptive gripper is connected to the end effector interface of the high-speed parallel robot via a flange.

7. The fully automatic large-particle gangue sorting system according to claim 6, characterized in that, The multi-finger adaptive gripper has at least three independently driven bionic phalanges. The inner side of each bionic phalange is covered with a variable stiffness friction material and has a built-in multi-dimensional torque sensor. The multi-finger adaptive gripper can plan the envelope trajectory and grasping force of each phalange in real time based on the 3D contour information of the target material.

8. The fully automatic large-particle gangue sorting system according to claim 7, characterized in that, The flexible gripping unit integrates a dynamic target tracking controller. Based on the material motion information from the multimodal recognition unit, the dynamic target tracking controller calculates and controls the gantry, the high-speed parallel robot, and the multi-fingered adaptive gripper to perform coordinated movements, so as to achieve predictive synchronous gripping of dynamic materials.

9. The fully automatic large-particle gangue sorting system according to claim 1, characterized in that, The inner walls of the gangue collection bin and the coal collection bin are lined with wear-resistant steel plates, the top is equipped with a hydraulic buffer unloading port, and the bottom is equipped with a weighing module and a pneumatic discharge gate.

10. The fully automatic large-particle gangue sorting system according to claim 1, characterized in that, The sorting system also includes an intelligent control platform, which is connected to the multimodal recognition unit, flexible grasping unit and ore collection unit via industrial Ethernet to realize system status monitoring, production data management, predictive maintenance alarms and remote model updates.

Images