System device for intelligently sorting coal gangue

Abstract

The invention provides a system device for intelligently sorting coal gangue. The system device comprises a feeding unit, a machine vision recognition unit, a main control unit and a sorting execution unit which are connected in sequence, the feeding unit comprises a vibrating feeder and a refining and shaping mechanism; the machine vision identification unit comprises a high-speed industrial camera and a near infrared spectrum camera; a coal gangue recognition model based on deep learning is arranged in the main control unit; the sorting execution unit comprises a plurality of high-pressure electromagnetic air valve nozzles which are evenly distributed in the transverse direction of the conveying belt. The vibration feeding mechanism and the material uniformizing and shaping mechanism are combined, so that single-layer uniform distribution of materials is ensured, and overlapping and shielding of the materials are effectively avoided; machine vision and near infrared spectrum multi-modal information fusion is adopted, and a deep learning algorithm is combined, so that the accuracy and adaptability of coal gangue recognition are remarkably improved; and through the time delay compensation and accurate positioning technology, it is ensured that the sorting execution mechanism accurately hits the target, and the sorting efficiency is improved.

Description

Technical Field

[0001] This invention belongs to the field of coal mine equipment technology and relates to a system device for intelligent sorting of coal gangue. Background Technology

[0002] Coal gangue is a solid waste generated during coal mining and washing. Its main components include carbonaceous shale and argillaceous sandstone, characterized by low calorific value and high hardness. During coal production, the proportion of coal gangue mixed in raw coal can reach 15% to 30%. If it cannot be effectively separated, it will lead to a series of problems: gangue reduces the overall calorific value of marketable coal, affecting coal quality and price; gangue increases transportation costs and boiler wear; and large stockpiles of coal gangue not only occupy land but also release harmful gases due to spontaneous combustion or pollute groundwater resources through rainwater leaching, posing a serious threat to the ecological environment. Therefore, achieving efficient separation of coal and gangue is one of the key links in the clean utilization of coal and the construction of green mines.

[0003] Traditional coal gangue sorting methods mainly fall into two categories: manual sorting and mechanical wet sorting. Manual sorting relies on the visual identification and physical labor of sorting workers, who pick out large pieces of gangue next to the conveyor belt. This method is not only extremely labor-intensive and involves harsh working conditions, but also has low sorting efficiency. The sorting accuracy fluctuates significantly due to the worker's experience and fatigue level, making it difficult to meet the needs of modern, large-scale coal production. Mechanical wet sorting, such as jigging and heavy media separation, although capable of handling large volumes, generally suffers from problems such as huge water consumption, the need for complex coal slurry water treatment systems, high construction and operating costs, and the potential for secondary pollution. In particular, for water-scarce regions such as western my country, the severe scarcity of water resources severely restricts the application of wet sorting technology.

[0004] In recent years, dry sorting technologies based on radiation detection have been developed, such as dual-energy transmission identification technology using X-rays or gamma rays, and identification technology based on visible and near-infrared spectroscopy. These technologies achieve sorting by analyzing the differences in physical properties or spectral characteristics between coal and gangue. However, these technologies still face challenges in practical applications: X-ray or gamma-ray devices have radiation protection requirements, high equipment costs, complex maintenance, and poor sorting effect on some gangue with similar density to coal; visible and near-infrared spectroscopy technologies are easily affected by factors such as coal dust, surface moisture, and lighting conditions, and their identification stability and reliability need to be improved. In addition, the above systems are often complex in structure and densely arranged with sensors, resulting in high initial investment and maintenance costs, which limits their widespread application in small and medium-sized coal mines.

[0005] With advancements in artificial intelligence, machine vision, and high-speed pneumatic actuators, intelligent dry sorting has emerged as a new direction for coal and gangue sorting. Existing intelligent sorting systems typically consist of three units: feeding, detection and identification, and sorting execution. The basic principle is that after the material is evenly distributed by the feeding system, its characteristic information is acquired through a sensing detection zone. A computer algorithm then determines whether it is coal or gangue, and finally, the actuator is controlled to blow the gangue away. However, existing systems still suffer from several pain points in engineering applications. Insufficient uniformity and stability of feeding lead to material accumulation or uneven distribution on the conveyor belt, resulting in material overlap and obstruction in the identification area, severely impacting identification accuracy. Existing identification models are poorly adaptable to complex and changing production conditions (such as changes in material surface humidity, dust interference, and fluctuations in ambient light), easily leading to misjudgments and missed judgments. Poor control of system response delay from identification to execution, coupled with the high-speed operation of the conveyor belt causing inaccurate pulse positioning and reduced sorting efficiency, further exacerbates the problem. Systems often lack effective self-learning and adaptive capabilities, failing to dynamically optimize model parameters based on changes in coal quality and sorting results. Fifth, the existing equipment's structural design does not adequately consider harsh mining conditions, and has deficiencies in dust prevention, shock resistance, and ease of maintenance, affecting the long-term stable operation of the equipment.

[0006] Therefore, there is an urgent need in this field for an intelligent coal gangue sorting system that has high sorting accuracy, fast response speed, good environmental adaptability, compact and reliable structure, and moderate cost, in order to solve the problems of uneven feeding, poor recognition stability, execution delay, lack of adaptive ability, and insufficient environmental tolerance in the existing technology. Summary of the Invention

[0007] To address the shortcomings of existing technologies, the present invention aims to provide a system device for intelligent coal gangue sorting. This invention combines vibratory feeding with a material leveling and shaping mechanism to ensure uniform single-layer distribution of materials, effectively avoiding material overlap and obstruction, thus laying the foundation for accurate identification. It employs machine vision and near-infrared spectral multimodal information fusion, combined with deep learning algorithms, to significantly improve the accuracy and adaptability of coal gangue identification. Through delay compensation and precise positioning technology, it ensures that the sorting execution mechanism accurately hits the target, improving sorting efficiency. It possesses self-learning and adaptive capabilities, continuously optimizing the sorting effect based on production data. The overall structural design fully considers the harsh working conditions of mines, featuring dustproof, shockproof, and easy-to-maintain characteristics, ensuring long-term stable operation of the equipment.

[0008] To achieve this objective, the present invention adopts the following technical solution:

[0009] This invention provides a system for intelligent sorting of coal gangue. The system includes a feeding unit, a machine vision recognition unit, a main control unit, and a sorting execution unit connected in sequence. The feeding unit includes a vibrating feeder and a material leveling and shaping mechanism for uniformly distributing the material in a single layer on the conveyor belt. The machine vision recognition unit includes a high-speed industrial camera and a near-infrared spectral camera for acquiring image and spectral information of the material. The main control unit is equipped with a deep learning-based coal gangue recognition model for fusing and processing image and spectral information and generating sorting instructions. The sorting execution unit includes several high-pressure electromagnetic valve nozzles evenly distributed laterally along the conveyor belt for blowing the gangue away from the discharge stream according to the sorting instructions.

[0010] This invention combines vibration feeding with a material leveling and shaping mechanism to ensure uniform single-layer material distribution, effectively avoiding material overlap and occlusion, thus laying the foundation for accurate identification. It employs machine vision and near-infrared spectral multimodal information fusion, combined with deep learning algorithms, to significantly improve the accuracy and adaptability of coal and gangue identification. Through delay compensation and precise positioning technology, it ensures the sorting execution mechanism accurately hits the target, improving sorting efficiency. It possesses self-learning and adaptive capabilities, continuously optimizing sorting performance based on production data. The overall structural design fully considers the harsh working conditions of mines, featuring dustproof, shockproof, and easy-to-maintain characteristics, ensuring long-term stable operation of the equipment.

[0011] Optionally, the material leveling and shaping mechanism is disposed between the discharge port of the vibrating feeder and the conveyor belt.

[0012] Optionally, the material leveling and shaping mechanism includes a guide plate disposed opposite to the discharge end of the vibrating feeder, and a scraper plate disposed between the guide plate and the transmission belt.

[0013] It should be noted that the feeding unit in this invention serves as the initial stage of the entire sorting system, ensuring that the material enters the subsequent identification and sorting stages in a single layer and with a uniform distribution. This unit achieves this goal through the coordinated operation of a vibrating feeder and a material leveling and shaping mechanism. After being continuously and stably fed out by the vibrating feeder, the material is first constrained by the lateral guide plate and limited to a preset width range. Subsequently, the material flow is forcibly scraped into a thin, uniform, single-layer material layer below the height-adjustable scraper, and finally spread evenly on the conveyor belt, entering the machine vision recognition area at a uniform speed.

[0014] Optionally, the machine vision recognition unit further includes an illumination housing, which is sealed and has a light-absorbing material on its inner wall. The high-speed industrial camera and the near-infrared spectral camera are mounted on the top of the illumination housing.

[0015] It should be noted that in this invention, the machine vision recognition unit acts as the system's eyes, accurately distinguishing between coal and gangue through multimodal information fusion. This function is achieved collaboratively by a high-speed industrial camera and a near-infrared spectral camera. The target's spatial location and contour information extracted by the high-speed industrial camera is used to delineate the region of interest (ROI) for the near-infrared spectral data, thereby achieving a precise correspondence between spectral information and specific materials. The fusion algorithm within the main control unit concatenates the visual features (shape, texture) and spectral features (composition) of each material into a high-dimensional feature vector. This feature vector is input into a pre-trained deep learning model (such as a hybrid model of convolutional neural network (CNN) and support vector machine (SVM)) for comprehensive decision-making. This multimodal fusion method overcomes the limitations of a single sensor in specific working conditions (such as coal surface contamination or gangue color similar to coal), greatly improving the accuracy and robustness of classification. Both cameras are integrated into a sealed lighting enclosure. The inner wall of the chamber is made of black light-absorbing material and equipped with an adjustable brightness and spectrally stable LED surface light source, providing constant and uniform lighting conditions for imaging, completely eliminating the interference of changes in ambient natural light, and ensuring the consistency of the collected data.

[0016] Optionally, the deep learning-based coal gangue identification model is a hybrid model combining convolutional neural networks and support vector machines, used as input for texture and shape features extracted by a high-speed industrial camera and spectral features extracted by a near-infrared spectral camera.

[0017] It should be noted that the hybrid model combining convolutional neural networks (CNNs) and support vector machines (SVMs) in this invention consists of two core stages: feature extraction and classification decision-making. CNNs replace the traditional manual feature design process (such as HOG and LBP), automatically learning superior and more abstract feature representations from raw images and spectral data, overcoming the blindness and limitations of manual feature design. SVMs are generally stronger than classifiers like Softmax in finding the optimal classification surface with small to medium sample sizes. Submitting the deep features extracted by CNNs to SVMs for classification yields more generalized and robust classification boundaries. This hybrid structure reduces reliance on large-scale data annotation and is more adaptable to common issues in industrial production such as imbalanced samples and noise interference. This architecture naturally supports early or mid-stage fusion of visual and spectral information, achieving information complementarity and effectively solving the classification problem of difficult samples with similar shapes but different textures (such as coal-coated gangue) or similar textures but different shapes (such as broken coal blocks).

[0018] Optionally, the main control unit further includes a data storage and learning module for recording historical sorting data and corresponding sensor information, and periodically retraining the coal gangue identification model to optimize the identification parameters.

[0019] Optionally, the main control unit further includes a delay compensation module, which is used to dynamically calculate and correct the delay time from recognition to blowing based on the conveyor belt speed and the physical distance between the recognition unit and the execution unit.

[0020] It should be noted that the delay compensation module in this invention is the core algorithm component that ensures the accuracy of sorting execution. It is responsible for solving the time delay problem between material identification and the execution of the sorting action. The delay compensation module needs to acquire the following system parameters in real time: Belt speed (v): measured in real time by an encoder installed on the drive roller, in m / s. Fixed distance (L) from the identification unit to the execution unit: a fixed value determined by the system's mechanical structure, in meters. System inherent delay (t_fixed): a fixed value obtained through experimental pre-calibration, including the sum of t1, t2, and t3, in seconds.

[0021] Optionally, the main control unit is electrically connected to a touch screen human-machine interface for real-time display of sorting status, recognition accuracy, equipment operating parameters, and receiving manual control commands.

[0022] Optionally, the high-pressure solenoid valve nozzle in the sorting execution unit is fixedly connected above the unloading section of the conveyor belt via a mounting bracket. A laser positioner is provided on the mounting bracket, and the laser positioner is used to calibrate the nozzle position.

[0023] Optionally, the blowing pressure and duration of the high-pressure electromagnetic valve nozzle are adjusted in stages by the main control unit according to the identified gangue size and location.

[0024] It should be noted that the high-pressure electromagnetic valve nozzle in this invention is the core operating component of the sorting execution unit, responsible for converting the identification signal into a physical sorting action. The high-pressure electromagnetic valve nozzle is an integrated component combining a solenoid valve and a nozzle, mainly including a pilot-operated solenoid valve with a direct-acting structure and a response time of less than 10ms; a high-efficiency nozzle using a Laval nozzle design, with an outlet airflow velocity reaching Mach number above 0.8; a pressure regulating module using an integrated pressure regulating valve, with an adjustable working pressure range of 0.4~1.0MPa; and a quick connector for connecting the high-pressure air source pipeline and the electrical control circuit. When not energized, the solenoid valve is closed under the action of the spring, and the high-pressure gas is sealed in the valve cavity. Upon receiving the blowing command from the main control unit, the electromagnetic coil is instantaneously energized (24V DC, 1.5A), and the electromagnetic force drives the valve core to move, overcoming the spring pressure to open the valve channel. The high-pressure gas instantly passes through the Laval nozzle, forming a high-speed airflow jet. After blowing is completed, the power is immediately cut off, and the valve core resets and closes under the action of the spring. It solves the technical problems of slow response, high energy consumption and poor accuracy in traditional pneumatic sorting systems, and provides a reliable execution guarantee for intelligent sorting of coal gangue.

[0025] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description.

[0026] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0027] This invention combines vibration feeding with a material leveling and shaping mechanism to ensure uniform single-layer material distribution, effectively avoiding material overlap and occlusion, thus laying the foundation for accurate identification. It employs machine vision and near-infrared spectral multimodal information fusion, combined with deep learning algorithms, to significantly improve the accuracy and adaptability of coal and gangue identification. Through delay compensation and precise positioning technology, it ensures the sorting execution mechanism accurately hits the target, improving sorting efficiency. It possesses self-learning and adaptive capabilities, continuously optimizing sorting performance based on production data. The overall structural design fully considers the harsh working conditions of mines, featuring dustproof, shockproof, and easy-to-maintain characteristics, ensuring long-term stable operation of the equipment. Attached Figure Description

[0028] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0029] Figure 1 A schematic diagram of the structure of a coal gangue intelligent sorting system device provided for a specific embodiment of the present invention;

[0030] Figure 2 A schematic diagram of the feeding unit in a coal gangue intelligent sorting system device provided in a specific embodiment of the present invention;

[0031] Among them, 1-feeding unit; 2-machine vision recognition unit; 3-main control unit; 4-sorting execution unit;

[0032] 11-Vibrating feeder; 12-Equalizing and shaping mechanism; 13-Conveyor belt. Detailed Implementation

[0033] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0034] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0035] It should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "set," "connected," and "linked" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0036] The technical solution of the present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0037] In one specific embodiment, the present invention provides a system device for intelligent sorting of coal gangue, such as... Figure 1 and Figure 2 As shown, the system includes a feeding unit 1, a machine vision recognition unit 2, a main control unit 3, and a sorting execution unit 4 connected in sequence. The feeding unit 1 includes a vibrating feeder 11 and a material leveling and shaping mechanism 12, which are used to distribute the material in a single layer and evenly on the conveyor belt 13. The machine vision recognition unit 2 includes a high-speed industrial camera and a near-infrared spectral camera, which are used to collect image information and spectral information of the material. The main control unit 3 is equipped with a coal gangue recognition model based on deep learning, which is used to fuse and process image and spectral information and generate sorting instructions. The sorting execution unit 4 includes several high-pressure electromagnetic valve nozzles evenly distributed laterally along the conveyor belt 13, which are used to blow the gangue away from the discharge stream according to the sorting instructions.

[0038] This invention combines vibratory feeding with a uniform material shaping mechanism 12 to ensure a single-layer, uniform distribution of materials, effectively avoiding material overlap and obstruction, thus laying the foundation for accurate identification. It employs machine vision and near-infrared spectral multimodal information fusion, combined with deep learning algorithms, to significantly improve the accuracy and adaptability of coal gangue identification. Through delay compensation and precise positioning technology, it ensures that the sorting execution mechanism accurately hits the target, improving sorting efficiency. It possesses self-learning and adaptive capabilities, continuously optimizing sorting performance based on production data. The overall structural design fully considers the harsh working conditions of mines, featuring dustproof, shockproof, and easy-to-maintain characteristics, ensuring long-term stable operation of the equipment.

[0039] In some embodiments, the material leveling and shaping mechanism 12 is disposed between the discharge port of the vibrating feeder 11 and the conveyor belt 13.

[0040] In some embodiments, the material leveling and shaping mechanism 12 includes a guide plate disposed opposite to the discharge port end of the vibrating feeder 11, and a scraper plate disposed between the guide plate and the drive belt.

[0041] It should be noted that, in this invention, the feeding unit 1 serves as the starting point of the entire sorting system, ensuring that the material enters the subsequent identification and sorting stages in a single layer and uniformly distributed state. This unit achieves this goal through the coordinated operation of the vibrating feeder 11 and the material shaping mechanism 12. After being continuously and stably fed out by the vibrating feeder 11, the material is first constrained by the lateral guide plate and limited to a preset width range. Subsequently, the material flow passes under the height-adjustable scraper plate, where it is forcibly scraped into a thin layer of material of uniform thickness and single layer, and finally spread evenly on the conveyor belt 13, entering the machine vision recognition area at a uniform speed.

[0042] In some embodiments, the machine vision recognition unit 2 further includes an illumination housing, which is sealed and has a light-absorbing material on its inner wall. A high-speed industrial camera and a near-infrared spectral camera are mounted on the top of the illumination housing.

[0043] It should be noted that in this invention, the machine vision recognition unit 2 acts as the system's eyes, accurately distinguishing between coal and gangue through multimodal information fusion. This function is achieved collaboratively by a high-speed industrial camera and a near-infrared spectral camera. The target spatial location and contour information extracted by the high-speed industrial camera is used to delineate the region of interest (ROI) for the near-infrared spectral data, thereby achieving a precise correspondence between spectral information and specific materials. The fusion algorithm within the main control unit 3 concatenates the visual features (shape, texture) and spectral features (composition) of each material into a high-dimensional feature vector. This feature vector is input into a pre-trained deep learning model (such as a hybrid model of convolutional neural network (CNN) and support vector machine (SVM)) for comprehensive decision-making. This multimodal fusion method overcomes the limitations of a single sensor in specific working conditions (such as coal surface contamination or gangue color similar to coal), greatly improving the accuracy and robustness of classification. Both cameras are integrated into a sealed lighting enclosure. The inner wall of the chamber is made of black light-absorbing material and equipped with an adjustable brightness and spectrally stable LED surface light source, providing constant and uniform lighting conditions for imaging, completely eliminating the interference of changes in ambient natural light, and ensuring the consistency of the collected data.

[0044] In some implementations, the deep learning-based coal gangue identification model is a hybrid model combining convolutional neural networks and support vector machines, used as input for texture and shape features extracted by a high-speed industrial camera and spectral features extracted by a near-infrared spectral camera.

[0045] It should be noted that the hybrid model combining convolutional neural networks (CNNs) and support vector machines (SVMs) in this invention consists of two core stages: feature extraction and classification decision-making. CNNs replace the traditional manual feature design process (such as HOG and LBP), automatically learning superior and more abstract feature representations from raw images and spectral data, overcoming the blindness and limitations of manual feature design. SVMs are generally stronger than classifiers like Softmax in finding the optimal classification surface with small to medium sample sizes. Submitting the deep features extracted by CNNs to SVMs for classification yields more generalized and robust classification boundaries. This hybrid structure reduces reliance on large-scale data annotation and is more adaptable to common issues in industrial production such as imbalanced samples and noise interference. This architecture naturally supports early or mid-stage fusion of visual and spectral information, achieving information complementarity and effectively solving the classification problem of difficult samples with similar shapes but different textures (such as coal-coated gangue) or similar textures but different shapes (such as broken coal blocks).

[0046] In some implementations, the main control unit 3 also includes a data storage and learning module for recording historical sorting data and corresponding sensor information, and periodically retraining the coal gangue identification model to optimize the identification parameters.

[0047] In some implementations, the main control unit 3 also includes a delay compensation module, which dynamically calculates and corrects the delay time from recognition to spraying based on the speed of the conveyor belt 13 and the physical distance between the recognition unit and the execution unit.

[0048] It should be noted that the delay compensation module in this invention is the core algorithm component that ensures the accuracy of sorting execution. It is responsible for solving the time delay problem between material identification and the execution of the sorting action. The delay compensation module needs to acquire the following system parameters in real time: Belt speed (v): measured in real time by an encoder installed on the drive roller, in m / s. Fixed distance (L) from the identification unit to the execution unit: a fixed value determined by the system's mechanical structure, in meters. System inherent delay (t_fixed): a fixed value obtained through experimental pre-calibration, including the sum of t1, t2, and t3, in seconds.

[0049] In some implementations, the main control unit 3 is electrically connected to a touch screen human-machine interface for real-time display of sorting status, recognition accuracy, equipment operating parameters, and receiving manual control commands.

[0050] In some embodiments, the high-pressure solenoid valve nozzle in the sorting execution unit 4 is fixedly connected above the unloading section of the conveyor belt 13 by a mounting bracket, and a laser positioner is provided on the mounting bracket for calibrating the nozzle position.

[0051] In some implementations, the blowing pressure and duration of the high-pressure solenoid valve nozzle are adjusted in stages by the main control unit 3 according to the identified gangue size and location.

[0052] It should be noted that the high-pressure electromagnetic valve nozzle in this invention is the core operating component of the sorting execution unit 4, responsible for converting the identification signal into a physical sorting action. The high-pressure electromagnetic valve nozzle is an integrated component combining a solenoid valve and a nozzle, mainly including a pilot-operated solenoid valve with a direct-acting structure and a response time of less than 10ms; a high-efficiency nozzle with a Laval nozzle design, and an outlet airflow velocity of over Mach 0.8; a pressure regulating module with an integrated pressure regulating valve, adjustable in working pressure range of 0.4~1.0MPa; and a quick connector for connecting the high-pressure gas source pipeline and the electrical control circuit. When not energized, the solenoid valve is closed under the action of the spring, and the high-pressure gas is sealed in the valve cavity. Upon receiving the blowing command from the main control unit 3, the electromagnetic coil is instantaneously energized (24V DC, 1.5A), and the electromagnetic force drives the valve core to move, overcoming the spring pressure to open the valve channel. The high-pressure gas instantly passes through the Laval nozzle, forming a high-speed airflow jet. After blowing is completed, the power is immediately cut off, and the valve core resets and closes under the action of the spring. It solves the technical problems of slow response, high energy consumption and poor accuracy in traditional pneumatic sorting systems, and provides a reliable execution guarantee for intelligent sorting of coal gangue.

[0053] Example 1

[0054] This embodiment provides a system device for intelligent sorting of coal gangue, wherein:

[0055] The system includes a feeding unit 1, a machine vision recognition unit 2, a main control unit 3, and a sorting execution unit 4 connected in sequence. The feeding unit 1 includes a vibrating feeder 11 and a material leveling and shaping mechanism 12, which are used to distribute the material in a single layer and evenly on the conveyor belt 13. The machine vision recognition unit 2 includes a high-speed industrial camera and a near-infrared spectral camera, which are used to collect image information and spectral information of the material. The main control unit 3 is equipped with a coal gangue recognition model based on deep learning, which is used to fuse and process image and spectral information and generate sorting instructions. The sorting execution unit 4 includes several high-pressure electromagnetic valve nozzles evenly distributed laterally along the conveyor belt 13, which are used to blow the gangue away from the discharge stream according to the sorting instructions.

[0056] The material leveling and shaping mechanism 12 is disposed between the discharge port of the vibrating feeder 11 and the conveyor belt 13. The material leveling and shaping mechanism 12 includes a guide plate disposed opposite to the end of the discharge port of the vibrating feeder 11, and a scraper plate disposed between the guide plate and the conveyor belt.

[0057] The machine vision recognition unit 2 also includes an illumination housing, which is sealed and has light-absorbing material on its inner wall. A high-speed industrial camera and a near-infrared spectral camera are mounted on the top of the illumination housing.

[0058] The deep learning-based coal gangue identification model is a hybrid model combining convolutional neural networks and support vector machines. It is used as input for texture and shape features extracted by a high-speed industrial camera and spectral features extracted by a near-infrared spectral camera.

[0059] The main control unit 3 also includes a data storage and learning module, used to record historical sorting data and corresponding sensor information, and periodically retrain the coal gangue identification model to optimize identification parameters. The main control unit 3 also includes a delay compensation module, used to dynamically calculate and correct the delay time from identification to injection based on the speed of the conveyor belt 13 and the physical distance between the identification unit and the execution unit. The main control unit 3 is electrically connected to a touch screen human-machine interface, used to display the sorting status, identification accuracy, equipment operating parameters in real time, and receive manual control commands.

[0060] The high-pressure solenoid valve nozzle in the sorting execution unit 4 is fixedly connected to the top of the unloading section of the conveyor belt 13 via a mounting bracket. A laser positioner is installed on the mounting bracket to calibrate the nozzle position. The blowing pressure and duration of the high-pressure solenoid valve nozzle are adjusted by the main control unit 3 according to the size and position of the identified gangue.

[0061] In summary, this invention combines vibratory feeding with a uniform material shaping mechanism 12 to ensure a single-layer, uniform distribution of materials, effectively avoiding material overlap and occlusion, thus laying the foundation for accurate identification. The use of machine vision and near-infrared spectral multimodal information fusion, combined with deep learning algorithms, significantly improves the accuracy and adaptability of coal gangue identification. Delay compensation and precise positioning technologies ensure that the sorting execution mechanism accurately hits the target, improving sorting efficiency. It possesses self-learning and adaptive capabilities, continuously optimizing sorting performance based on production data. The overall structural design fully considers the harsh working conditions of mines, featuring dustproof, shockproof, and easy-to-maintain characteristics, ensuring long-term stable operation of the equipment.

[0062] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A system device for intelligent sorting of coal gangue, characterized in that, The system device includes a feeding unit, a machine vision recognition unit, a main control unit, and a sorting execution unit connected in sequence; The feeding unit includes a vibrating feeder and a material leveling and shaping mechanism, which are used to distribute the material in a single layer and evenly on the conveyor belt. The machine vision recognition unit includes a high-speed industrial camera and a near-infrared spectral camera, used to acquire image and spectral information of the material. The main control unit is equipped with a deep learning-based coal gangue identification model, which is used to fuse and process image and spectral information and generate sorting instructions. The sorting execution unit includes several high-pressure electromagnetic valve nozzles evenly distributed laterally along the conveyor belt, used to blow gangue away from the discharge stream according to sorting instructions.

2. The system apparatus according to claim 1, characterized in that, The material leveling and shaping mechanism is located between the discharge port of the vibrating feeder and the conveyor belt.

3. The system apparatus according to claim 2, characterized in that, The material leveling and shaping mechanism includes a guide plate disposed opposite to the discharge end of the vibrating feeder, and a scraper plate disposed between the guide plate and the transmission belt.

4. The system apparatus according to claim 1, characterized in that, The machine vision recognition unit also includes an illumination housing, which is sealed and has a light-absorbing material on its inner wall. The high-speed industrial camera and the near-infrared spectral camera are mounted on the top of the illumination housing.

5. The system apparatus according to any one of claims 1-4, characterized in that, The deep learning-based coal gangue identification model is a hybrid model combining convolutional neural networks and support vector machines, used as input for texture and shape features extracted by a high-speed industrial camera and spectral features extracted by a near-infrared spectral camera.

6. The system apparatus according to claim 1, characterized in that, The main control unit also includes a data storage and learning module, which is used to record historical sorting data and corresponding sensor information, and periodically retrain the coal gangue identification model to optimize the identification parameters.

7. The system apparatus according to claim 6, characterized in that, The main control unit also includes a delay compensation module, which is used to dynamically calculate and correct the delay time from recognition to blowing based on the conveyor belt speed and the physical distance between the recognition unit and the execution unit.

8. The system apparatus according to claim 7, characterized in that, The main control unit is electrically connected to a touch screen human-machine interface, which is used to display the sorting status, recognition accuracy, and equipment operating parameters in real time, and to receive manual control commands.

9. The system apparatus according to claim 1, characterized in that, The high-pressure electromagnetic valve nozzle in the sorting execution unit is fixedly connected to the top of the conveyor belt unloading section via a mounting bracket. A laser positioner is installed on the mounting bracket, which is used to calibrate the nozzle position.

10. The system apparatus according to claim 9, characterized in that, The blowing pressure and duration of the high-pressure electromagnetic valve nozzle are adjusted in stages by the main control unit according to the size and location of the identified gangue.

Images