An intelligent impurity sorting system and method for clean coal of a coal preparation plant vibrating screen

Abstract

The application discloses a clean coal impurity intelligent sorting system and method for a coal preparation plant vibrating screen, and relates to the technical field of industrial automation and intelligent sorting. The system comprises a super-vision perception system, an intelligent control core and a high-precision execution system. The super-vision perception system is used for collecting and identifying impurities on the vibrating screen in a vibrating and dusty environment. The intelligent control core is used for receiving the identification result and generating a sorting instruction. The high-precision execution system is used for sorting the identified impurities according to the identification instruction. The super-vision perception system, the intelligent control core and the high-precision execution system are integrated on a mechanical platform with anti-vibration and dustproof design. The clean coal impurity sorting scheme has high adaptability, improves the impurity sorting precision and efficiency, overcomes the strong vibration of the vibrating screen and the dusty environment of the coal preparation plant, and realizes stable operation in harsh working conditions.

Description

Technical Field

[0001] This invention relates to the field of industrial automation and intelligent sorting technology, specifically to an intelligent sorting system and method for clean coal and impurities in a vibrating screen of a coal preparation plant. Background Technology

[0002] In the coal washing and processing stage, during the vibrating screen grading operation, the clean coal conveyed on the screen surface is commonly mixed with foreign objects such as gangue, wood, gas pipes, and rope-plastic debris. The presence of these impurities in the finished clean coal not only causes excessive ash content and a decline in coal quality, reducing the economic value of the coal, but the hard impurities can also easily impact and wear down subsequent production equipment such as screening and crushing equipment, causing equipment failures and hindering the stable operation of the production line. In the coal preparation process, the clean coal on the vibrating screen is often mixed with gangue, wood, gas pipes, and other impurities, seriously affecting the quality of the clean coal. Currently, the main method relies on manual identification and sorting on both sides of the vibrating screen, which has certain drawbacks, such as the harsh working environment. Coal plant sites are characterized by high levels of dust, noise, and vibration, posing significant health risks to workers. Sorting efficiency and quality are inconsistent; manual sorting is affected by subjective factors and physical exertion, making it difficult to maintain high efficiency and consistency, and prone to missed inspections. Labor costs are high, with companies facing difficulties in recruitment and continuously rising management costs. Existing automated sorting solutions are mostly designed for stationary or stable materials; when directly applied to vibrating screens subjected to continuous and intense vibration, problems such as blurred images, inaccurate positioning, and mechanical resonance lead to a sharp decline in recognition and sorting rates, failing to meet industrial production requirements. Therefore, there is an urgent need for a dedicated intelligent sorting solution that can adapt to the special working conditions of vibrating screens. Summary of the Invention

[0003] The purpose of this invention is to provide an intelligent sorting system and method for clean coal impurities in a vibrating screen of a coal preparation plant, so as to solve the problems raised in the prior art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: an intelligent sorting system for clean coal impurities in a vibrating screen of a coal preparation plant, the system comprising: an ultra-visual perception system, an intelligent control center, and a high-precision execution system; The super-visual perception system is used to collect and identify debris on a vibrating screen in a vibrating and dusty environment; The intelligent control center is used to receive the identification results and generate sorting instructions; The high-precision execution system is used to sort the identified debris according to the identification instructions; The super-vision perception system, intelligent control center and high-precision execution system are integrated into a mechanical platform with vibration resistance and dustproof design.

[0005] Furthermore, the ultra-visual perception system includes a dustproof and waterproof industrial camera, an image processing unit, and an inertial measurement unit; The dustproof and waterproof industrial camera is equipped with a shock-resistant lens and an adaptive lighting system; The image processing unit is built on a Python environment and integrates the YOLOv5 object detection algorithm; The inertial measurement unit is used for vibration compensation.

[0006] Furthermore, the YOLOv5 object detection algorithm employs a multi-scale training strategy and is trained using an augmented dataset simulating a coal preparation plant environment to achieve high-precision identification of impurities in clean coal. The core of the super-visual perception system lies in the recognition module built based on Python and the YOLOv5 algorithm. This recognition module has undergone targeted optimization, such as the introduction of attention mechanisms and multi-scale training, and can effectively distinguish the subtle features of clean coal blocks from impurities.

[0007] Furthermore, the intelligent control center uses the Hechuan Q series high-performance PLC as its hardware platform; it incorporates a material trajectory prediction algorithm and a vibration compensation control algorithm based on Kalman filtering; it communicates with the high-precision execution system via EtherCAT bus and with the ultra-visual perception system via TCP / IP protocol.

[0008] A method for intelligent sorting of impurities in clean coal using a vibrating screen in a coal preparation plant, comprising the following steps: S1. Real-time acquisition of material image data of the vibrating screen through the ultra-visual perception system; S2. Use the optimized YOLOv5 algorithm to process the material image, identify debris and obtain the location information of the debris; S3. Set the material trajectory prediction algorithm to calculate the motion trajectory of the debris by combining the location information of the identified debris, the operating parameters of the vibrating screen and the vibration data fed back by the IMU; S4. Use the intelligent control center to generate sorting instructions and send them to the high-precision execution system via the EtherGAT bus; S5. A vibration compensation control algorithm is set to drive the high-precision execution system to complete the sorting action of debris at the estimated position; the high-precision execution system includes a servo system and a mechanical gantry manipulator; the servo system drives the mechanical gantry manipulator to remove debris from the clean coal stream at precise time and position to complete the sorting.

[0009] Furthermore, in step S1: the material image data is collected and analyzed using a super-visual perception system. Based on the differences in edge contour, size, shape, and color features of the extracted coal blocks and impurities, convolutional neural network technology is used to classify and distinguish the extracted features, thereby achieving the distinction between coal blocks and impurities. The classification and distinction of extracted features using convolutional neural network technology includes the following steps: automatically extracting the contour, texture, grayscale, and shape features of clean coal and impurities through convolutional layers; using pooling layers to filter effective features, remove redundant information, and enhance anti-interference ability; fusing and clustering the extracted multidimensional features to form a feature model specific to various materials; and comparing feature differences through fully connected layers to complete the classification and discrimination of clean coal and various impurities.

[0010] Furthermore, in step S2: the material image is processed, and the position information is determined based on the center point of the object recognition box. Using the four-point transformation perspective principle, the vibrating screen area is selected, with the upper left point of the four-point box as the origin of the coordinate system. With the origin of the coordinate system as the reference, the upper edge of the four-point box to the left is the positive Y-axis, and the lower left edge of the four-point box is the positive X-axis, thus establishing a plane rectangular coordinate system. At the same time, the vibrating screen area is defined, and the view distortion and near-large-far-small phenomenon caused by the change of the camera installation position are eliminated. In addition, the recognition area is further subdivided into horizontal areas, dividing the horizontal area into N small areas. The X-coordinate of each independent small area is independently corrected. When performing independent X-coordinate correction, different compensation amounts can be set for the X-coordinate values ​​in the N independent areas, so that the transformed coordinate X obtained in the Nth area plus the compensation amount of the Nth area is closer to the actual coordinate X, thereby improving the accuracy of the recognition position coordinate.

[0011] Furthermore, in step S3: the material trajectory prediction algorithm sets the sampling period. When setting the sampling period, the sampling period is variable and can be set to different lengths. However, once the sampling period is set, the system's sampling period will remain constant each time. To adjust the sampling period, the sampling frequency per second can be modified in the program to adjust the size of the sampling period. For example, the sampling period Δt = 1 / sampling frequency per second; the location of the debris collected in the previous sampling period is recorded as P. t-1 The location of the debris collected during the current sampling period is denoted as P. t The trajectory of the debris is predicted using second-order kinematic equations; the current velocity of the debris is calculated using the following formula: ; where v t The velocity of the debris during the current sampling period is represented by the following formula; the acceleration of the debris during the current sampling period is calculated using the following formula: ; where a t v represents the acceleration of the debris during the current sampling period. t-1This indicates the velocity of the debris in the previous sampling period; v t This represents the velocity of debris within the current sampling period; based on the calculation results and the current position P... t The location of the debris in the next sampling period is predicted based on the calculation results of the following formula: P t+1 =P t +v t · + ·a t ·(Δt) 2 Due to inertia, this algorithm has high accuracy. The material trajectory prediction algorithm combines the kinematic model of the material on the vibrating screen and real-time vibration data to dynamically correct the trajectory of the debris. First, the position of the debris on the horizontal plane of the screen is predicted and analyzed by the second-order kinematic equation. Then, the accurate position and relatively stable grasping time of the debris are obtained by the vibration compensation mechanism.

[0012] Furthermore, in step S5: the vibration compensation control algorithm includes both open-loop control and closed-loop control. It utilizes the vibration compensation mechanism to determine the peak or trough time, thereby enabling debris grabbing at the peak or trough. Because the debris is relatively stationary in the vertical direction at the peak or trough, setting the grabbing time at the peak or trough improves debris sorting efficiency and accuracy. The open-loop control can directly calculate the peak or trough time based on the known vibration frequency f and initial phase φ0, using the following formula: , where t k φ0 represents the peak or trough moment; k is an integer; ω represents the angular velocity of the vibrating motor; since the motion of the vibrating screen is ideally a sine wave, this formula represents the peak calculation formula for a sine wave; whether the peak or trough moment is obtained depends on the phase, for example, when... At that time, it is the peak moment; when When the pendulum reaches its highest point, the time is the trough. Closed-loop control, on the other hand, uses an encoder or photoelectric sensor switch installed at the vibrating motor to send a signal when the pendulum swings to its highest point, thus determining whether the debris is at the peak or trough of the wave.

[0013] Compared with the prior art, the beneficial effects of the present invention are: The proposed coal impurity sorting scheme is highly adaptable and improves sorting accuracy and efficiency. Through hardware vibration resistance design and software compensation algorithms, it effectively overcomes the strong vibrations of vibrating screens and the dusty environment of coal preparation plants, achieving stable operation under harsh conditions. The optimized YOLOv5 algorithm ensures a recognition rate of up to 95%. Combined with advanced trajectory prediction and real-time control, the sorting success rate reaches 90%, far exceeding the efficiency of manual sorting. By deeply integrating the Python open-source vision ecosystem, the CODESYS automation platform, and the domestically produced Hechuan PLC, it creatively solves the pain points of specific industrial scenarios, providing a replicable technical path for the intelligent upgrading of traditional industries. Attached Figure Description

[0014] Figure 1 This is a block diagram of the overall architecture of an intelligent sorting system for clean coal impurities in a vibrating screen of a coal preparation plant according to the present invention. Figure 2 This is a schematic diagram of the process steps of an intelligent sorting method for clean coal impurities in a vibrating screen of a coal preparation plant according to the present invention. Detailed Implementation

[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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 are within the scope of protection of the present invention.

[0016] like Figure 1 As shown, the present invention provides a technical solution: an intelligent sorting system for clean coal impurities in a vibrating screen of a coal preparation plant. The system includes an ultra-visual perception system, an intelligent control center, and a high-precision execution system. The ultra-visual sensing system is used to collect and identify debris on vibrating screens in vibrating and dusty environments; The intelligent control center is used to receive the identification results and generate sorting instructions; The high-precision execution system is used to sort identified debris according to identification instructions; The super vision perception system, intelligent control center and high-precision execution system are integrated into a mechanical platform with vibration resistance and dustproof design.

[0017] The ultra-visual perception system includes a dustproof and waterproof industrial camera, an image processing unit, and an inertial measurement unit; Dustproof and waterproof industrial cameras are equipped with anti-vibration lenses and adaptive lighting systems; The image processing unit is built on a Python environment and integrates the YOLOv5 object detection algorithm; Inertial measurement units are used for vibration compensation.

[0018] The YOLOv5 object detection algorithm employs a multi-scale training strategy and is trained using an augmented dataset simulating a coal preparation plant environment to achieve high-precision identification of impurities in clean coal.

[0019] The intelligent control center uses Hechuan Q series high-performance PLC as its hardware platform; it has built-in material trajectory prediction algorithm and vibration compensation control algorithm based on Kalman filtering; it communicates with the high-precision execution system through EtherCAT bus and with the ultra-vision perception system through TCP / IP protocol.

[0020] like Figure 2 As shown, an intelligent sorting method for clean coal impurities in a vibrating screen of a coal preparation plant includes the following steps: S1. Real-time acquisition of material image data of the vibrating screen through the ultra-visual perception system; S2. Use the optimized YOLOv5 algorithm to process the material image, identify debris and obtain the location information of the debris; S3. Set the material trajectory prediction algorithm to calculate the motion trajectory of the debris by combining the location information of the identified debris, the operating parameters of the vibrating screen and the vibration data fed back by the IMU; S4. Use the intelligent control center to generate sorting instructions and send them to the high-precision execution system via the EtherGAT bus; S5. Set a vibration compensation control algorithm to drive the high-precision execution system to complete the sorting action of debris at the estimated position.

[0021] In step S1: The material image data is collected and analyzed using a super-visual perception system. Based on the differences in edge contour, size, shape, and color features of the extracted coal blocks and impurities, convolutional neural network technology is used to classify and distinguish the extracted features, thereby achieving the distinction between coal blocks and impurities. The classification and distinction of extracted features using convolutional neural network technology includes the following steps: automatically extracting the contour, texture, grayscale, and shape features of clean coal and impurities through convolutional layers; using pooling layers to filter effective features, remove redundant information, and enhance anti-interference ability; fusing and clustering the extracted multidimensional features to form a feature model specific to each type of material; and comparing feature differences through fully connected layers to complete the classification and discrimination of clean coal and various types of impurities.

[0022] In step S2: The material image is processed, and the position information is determined based on the center point of the object recognition frame. Using the four-point transformation perspective principle, the vibrating screen area is selected, with the upper left point of the four-point frame as the origin of the coordinate system. With the origin of the coordinate system as the reference, the upper edge of the four-point frame to the left is the positive Y-axis, and the lower left edge of the four-point frame is the positive X-axis, thus establishing a plane rectangular coordinate system. This simultaneously achieves the framing of the vibrating screen area and eliminates the view distortion and near-large-far-small phenomenon caused by changes in the camera's installation position. In addition, the recognition area is further subdivided into N smaller areas. The X-coordinate of each independent small area is independently corrected. When independently correcting the X-coordinate, different compensation amounts can be set for the X-coordinate values ​​in the N independent areas, so that the transformed coordinate X obtained in the Nth area plus the compensation amount of the Nth area is closer to the actual coordinate X, thus improving the accuracy of the recognition position coordinates.

[0023] In step S3: The material trajectory prediction algorithm is set with a fixed but variable sampling period. The sampling period is variable and can be set to different lengths. However, once a sampling period is set, it will remain constant for each system sampling cycle. To adjust the sampling period, the sampling frequency per second can be modified in the program. For example, the sampling period Δt = 1 / sampling frequency per second. The location of the debris collected in the previous sampling period is recorded as P. t-1 The location of the debris collected during the current sampling period is denoted as P. t The trajectory of the debris is predicted using second-order kinematic equations; the current velocity of the debris is calculated using the following formula: ; where v t The velocity of the debris during the current sampling period is represented by the following formula; the acceleration of the debris during the current sampling period is calculated using the following formula: ; where a t v represents the acceleration of the debris during the current sampling period. t-1 This represents the velocity of debris in the previous sampling period; based on the calculation results and the current position P... t The location of the debris in the next sampling period is predicted based on the calculation results of the following formula: P t+1 =P t +v t · + ·a t ·(Δt) 2 Due to inertia, this algorithm has high accuracy. The material trajectory prediction algorithm combines the kinematic model of the material on the vibrating screen and real-time vibration data to dynamically correct the trajectory of the debris. First, the position of the debris on the horizontal plane of the screen is predicted and analyzed by the second-order kinematic equation. Then, the vibration compensation mechanism is used to obtain the accurate position of the debris and the relatively stable grasping time.

[0024] In step S5: The vibration compensation control algorithm includes both open-loop and closed-loop control. Debris is grabbed at wave crests or troughs, and at these times, the debris is relatively stationary in the vertical direction. Open-loop control can directly calculate the timing of the wave crest or trough based on the known vibration frequency f and initial phase φ0. ;t k φ0 represents the peak or trough moment; k is an integer; ω represents the angular velocity of the vibrating motor; since the motion of the vibrating screen is ideally a sine wave, this formula represents the peak calculation formula for a sine wave; whether the peak or trough moment is obtained depends on the phase, for example, when... At that time, it is the peak moment; when When the pendulum swings to the peak position, it is at the trough position; closed-loop control determines whether the debris is at the peak or trough position by adding an encoder or photoelectric sensor switch to the vibrating motor and sending a signal when the pendulum swings to the peak position.

[0025] Example 1:

[0026] In this embodiment, the system is installed next to the main vibrating screen of the coal preparation plant; the industrial camera of the ultra-vision perception system faces the discharge end of the vibrating screen to capture moving materials; after the system is started, the camera continuously acquires images and runs the YOLOv5 model through the Python vision platform for real-time recognition; once debris is detected, its pixel coordinates, category confidence, and other information are sent to the Hechuan PLC via the TCP / IP protocol; after receiving the data, the CODESYS runtime environment in the PLC immediately starts the trajectory prediction algorithm; this algorithm combines the screen speed fed back by the encoder and the IMU measurement... Based on real-time vibration acceleration and a preset material motion model, the system calculates the precise three-dimensional coordinates (X, Y, Z) and arrival time T of the debris within the next few hundred milliseconds. Simultaneously, the vibration compensation algorithm fine-tunes the actuator's trajectory according to the current vibration state. Finally, the PLC synchronously sends instructions containing the target point and compensation amount to all servo drives via the EtherCAT bus. The servo system drives the mechanical gantry robot to remove the debris from the clean coal stream at precise times and locations, completing the sorting process. The entire system has been successfully tested on-site at a vibration acceleration of 2g and a dust concentration of 50mg / m³. 3 In this environment, it has been running continuously and stably for more than 1,000 hours, achieving a 95% recognition rate and a 90% sorting success rate.

[0027] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. An intelligent sorting system for clean coal impurities in a vibrating screen of a coal preparation plant, characterized in that: The system includes: Ultra-visual perception system, intelligent control center, and high-precision execution system; The super-visual perception system is used to collect and identify debris on a vibrating screen in a vibrating and dusty environment; The intelligent control center is used to receive the identification results and generate sorting instructions; The high-precision execution system is used to sort the identified debris according to the identification instructions; The super-vision perception system, intelligent control center and high-precision execution system are integrated into a mechanical platform with vibration resistance and dustproof design.

2. The intelligent sorting system for clean coal impurities in a vibrating screen of a coal preparation plant according to claim 1, characterized in that: The ultra-visual perception system includes a dustproof and waterproof industrial camera, an image processing unit, and an inertial measurement unit. The dustproof and waterproof industrial camera is equipped with a shock-resistant lens and an adaptive lighting system; The image processing unit is built on a Python environment and integrates the YOLOv5 object detection algorithm; The inertial measurement unit is used for vibration compensation.

3. The intelligent sorting system for clean coal impurities in a vibrating screen of a coal preparation plant according to claim 2, characterized in that: The YOLOv5 object detection algorithm employs a multi-scale training strategy and is trained using an augmented dataset simulating a coal preparation plant environment to achieve high-precision identification of impurities in clean coal.

4. The intelligent sorting system for clean coal impurities in a vibrating screen of a coal preparation plant according to claim 1, characterized in that: The intelligent control center uses a PLC as its hardware platform; it incorporates a material trajectory prediction algorithm and a vibration compensation control algorithm based on Kalman filtering; it communicates with the high-precision execution system via EtherCAT bus and with the ultra-visual perception system via TCP / IP protocol.

5. The intelligent sorting method for clean coal impurities in a vibrating screen of a coal preparation plant according to claim 1, characterized in that: The method includes the following steps: S1. Real-time acquisition of material image data of the vibrating screen through the ultra-visual perception system; S2. Use the optimized YOLOv5 algorithm to process the material image, identify debris and obtain the location information of the debris; S3. Set the material trajectory prediction algorithm to calculate the motion trajectory of the debris by combining the location information of the identified debris, the operating parameters of the vibrating screen and the vibration data fed back by the IMU; S4. Use the intelligent control center to generate sorting instructions and send them to the high-precision execution system via the EtherGAT bus; S5. Set a vibration compensation control algorithm to drive the high-precision execution system to complete the sorting action of debris at the estimated position.

6. The intelligent sorting method for clean coal impurities in a vibrating screen of a coal preparation plant according to claim 5, characterized in that: In step S1: The material image data is collected and analyzed using a super vision perception system. Based on the differences in edge contour, size, shape and color features of the extracted coal blocks and debris, the extracted features are classified and distinguished using convolutional neural network technology, thereby realizing the distinction between coal blocks and debris.

7. The intelligent sorting method for clean coal impurities in a vibrating screen of a coal preparation plant according to claim 5, characterized in that: In step S2: The material image is processed, and the position information is determined based on the center point of the object recognition box. Using the four-point perspective transformation principle, the vibrating screen area is selected, with the upper left point of the four-point box as the origin of the coordinate system. With the origin of the coordinate system as the reference, the upper edge of the four-point box to the left is the positive Y-axis, and the lower left edge of the four-point box is the positive X-axis, thus establishing a plane rectangular coordinate system. At the same time, the vibrating screen area is defined, and the view distortion and near-large-far-small phenomenon caused by the change of the camera installation position are eliminated. In addition, the recognition area is further subdivided into horizontal areas, dividing the horizontal area into N small areas, and the X-coordinate of each independent small area is independently corrected.

8. The intelligent sorting method for clean coal impurities in a vibrating screen of a coal preparation plant according to claim 5, characterized in that: In step S3: the material trajectory prediction algorithm sets the sampling period. The location of the debris collected in the previous sampling period is recorded as P. t-1 The location of the debris collected during the current sampling period is denoted as P. t The trajectory of debris is predicted using second-order kinematic equations. The material trajectory prediction algorithm combines the kinematic model of the material on the vibrating screen and real-time vibration data to dynamically correct the trajectory of the debris. First, the position of the debris on the horizontal plane of the screen is predicted and analyzed by the second-order kinematic equation. Then, the position and grasping time of the debris are obtained by the vibration compensation mechanism.

9. A method for intelligent sorting of clean coal impurities in a vibrating screen of a coal preparation plant according to claim 8, characterized in that: In step S5: the vibration compensation control algorithm includes both open-loop and closed-loop control. It utilizes a vibration compensation mechanism to determine the peak or trough times, thereby enabling debris grabbing at those times. The open-loop control can directly calculate the peak or trough time t based on the known vibration frequency f and initial phase φ0. k Closed-loop control, on the other hand, involves installing an encoder or photoelectric sensor switch at the vibrating motor. When the pendulum swings to its highest point, a signal is sent to determine whether the debris is at the crest or trough of the wave.

Images