A coal beneficiation method, device, equipment and medium

Abstract

The present application relates to a coal beneficiation method, device, equipment and medium, and relates to the technical field of coal processing and intelligent sorting, comprising: collecting real-time images of coal to be sorted and performing image preprocessing; extracting image features to identify the surface texture and gray difference of coal and gangue; determining the real-time coordinates of gangue on the conveyor belt based on the constructed coal-gangue recognition positioning network; extracting the motion speed of the conveyor belt and calculating the delay time of the action of the blowing valve; when the delay time arrives, controlling the blowing valve corresponding to the coordinates to spray high-pressure gas to separate the gangue. The present application realizes efficient and accurate automatic sorting of coal and gangue by high-precision image recognition combined with precise timing control of the blowing gas flow, improves the purity and efficiency of coal washing, and reduces the labor intensity and sorting cost.

Description

Technical Field

[0001] This invention relates to the field of coal processing and intelligent sorting technology, and in particular to a coal beneficiation device. Background Technology

[0002] Coal is a crucial basic energy source in my country, and its mining and processing technologies directly impact energy efficiency and environmental protection. During coal mining, raw coal often contains a large amount of gangue (i.e., coal-related waste). To improve coal combustion efficiency and reduce the emission of harmful gases after combustion, it is essential to effectively separate coal from gangue through mineral processing before use.

[0003] Traditional coal beneficiation mainly relies on manual sorting, heavy media separation, and jigging. Manual sorting is not only labor-intensive and involves harsh working conditions, but also has extremely low sorting efficiency and accuracy, making it difficult to meet the needs of large-scale industrial production. While wet beneficiation technologies such as heavy media separation and jigging have higher sorting efficiency, they consume large amounts of beneficiation water resources and generate a large amount of coal slurry water in subsequent processing, which not only wastes water resources but also easily leads to secondary environmental pollution problems.

[0004] With the development of computer vision and automation control technology, dry separation technology combining photoelectric detection and pneumatic injection has gradually become a research hotspot. However, in the actual coal transportation process, the surfaces of coal and gangue are often covered with coal dust, resulting in extremely similar grayscale and texture characteristics, making it difficult for existing image recognition methods to quickly and accurately distinguish them. At the same time, due to the slight fluctuations in the conveyor belt speed and the varying sizes of coal and gangue particles, the response timing of the pneumatic injection valve is not precise enough when the located gangue reaches the injection area, easily leading to premature or delayed injection, resulting in mis-injection and missed injection, which seriously affects the purity and recovery rate of dry intelligent mineral processing. Summary of the Invention

[0005] To address the aforementioned problems, this application provides a coal beneficiation apparatus.

[0006] This application provides a coal beneficiation method, apparatus, equipment, and medium, employing the following technical solution: In a first aspect, this application provides a method for beneficiating coal, comprising the following steps: Step S1: Obtain the original image of the coal material on the conveyor belt through the image acquisition module, and perform bilateral filtering denoising and histogram equalization enhancement on the original image to obtain a preprocessed image; Step S2: Input the preprocessed image into a preset coal gangue identification and positioning network, extract the shallow texture features and deep grayscale features of the preprocessed image through the coal gangue identification and positioning network, and fuse them to obtain a coal gangue feature map; Step S3: Based on the coal gangue feature map, perform target detection to identify the gangue region and extract the real-time center coordinates of the gangue region mapped onto the conveyor belt. ; Step S4: Obtain the real-time operating speed of the conveyor belt. And based on the real-time center coordinates With respect to the preset horizontal baseline of the blow valve assembly Calculate the physical distance of the gangue to the injection area; Step S5: Based on the physical distance and the real-time running speed Calculate the delay time of the jet valve action When the timer reaches the aforementioned delay time At that time, control the corresponding horizontal coordinate. The injection valve at the point is opened, and high-pressure gas is used to separate the gangue from the coal material.

[0007] By adopting the above technical solution, the system can first perform noise reduction and contrast enhancement processing on the coal material image on the conveyor belt to eliminate the impact of coal dust on image quality; then, it can extract deep and shallow features through the coal gangue identification and positioning network to achieve accurate positioning and coordinate extraction of gangue; finally, it can calculate the accurate injection delay time based on the real-time running speed of the conveyor belt and execute the injection, thus realizing automatic, efficient and high-precision dry sorting of coal gangue.

[0008] Preferably, in step S1, the bilateral filtering denoising involves removing pixels from the original image. The grayscale value is replaced by the weighted average of the neighboring pixels, and the weights are determined by the spatial domain kernel function and the value domain kernel function.

[0009] By adopting the above technical solution, bilateral filtering can effectively remove high-frequency noise in the image (such as noise caused by coal dust) while preserving the edge information of coal and gangue, avoiding edge blurring, and laying a good foundation for subsequent feature extraction and localization.

[0010] Preferably, in step S2, the coal gangue identification and localization network adopts an improved YOLOv5 network, in which an attention mechanism module is introduced into the feature extraction backbone network to enhance the ability to extract weak surface texture features of gangue.

[0011] By adopting the above technical solution, the improved YOLOv5 network can achieve fast forward reasoning of images. The introduction of the attention mechanism enables the network to focus on the differences between coal and gangue in subtle textures such as reflectivity and roughness, which significantly improves the recognition accuracy in complex backgrounds.

[0012] Preferably, in step S4, the real-time running speed It is calculated by real-time acquisition of pulse signals by a rotary encoder installed on the drive roller of the conveyor belt.

[0013] By adopting the above technical solution, the real-time speed of the conveyor belt can be obtained through a rotary encoder, which can eliminate speed errors caused by voltage fluctuations, load changes or belt slippage of the conveyor belt motor, and ensure the accuracy of the mapping from physical distance to the time domain.

[0014] Preferably, in step S5, the pressure of the high-pressure gas is controlled between 0.6 MPa and 0.8 MPa, and the duration of opening the blow valve is proportional to the geometric dimensions of the gangue along the conveyor belt running direction.

[0015] By adopting the above technical solution, stabilizing the gas pressure at 0.6MPa~0.8MPa (e.g. 0.7MPa) can ensure that sufficient momentum is generated to blow heavy gangue away from its original trajectory without affecting adjacent coal. At the same time, the injection duration is dynamically adjusted according to the size of the gangue, which not only ensures that large pieces of gangue are completely blown away, but also avoids wasting the lifespan of multiple components and high-pressure gas when injecting small pieces of gangue.

[0016] Secondly, this application provides a coal beneficiation apparatus, comprising: The image acquisition module is used to acquire raw images of the coal material on the conveyor belt; The image preprocessing module is used to perform bilateral filtering for noise reduction and histogram equalization enhancement on the original image to obtain a preprocessed image; The coal gangue identification and positioning module is used to input the preprocessed image into a preset coal gangue identification and positioning network, extract the shallow texture features and deep grayscale features of the preprocessed image to fuse them to obtain a coal gangue feature map, identify the gangue region based on the coal gangue feature map, and extract the real-time center coordinates of the gangue region mapped on the conveyor belt. ; The delay control module is used to obtain the real-time operating speed of the conveyor belt. Calculate the physical distance of the gangue to the injection zone, and base the calculation on the physical distance and the real-time operating speed. Calculate the delay time of the jet valve action ; The sorting module is used to execute the sorting module when the timer reaches the delay time. At that time, control the corresponding horizontal coordinate. The injection valve at the point is opened, and high-pressure gas is used to separate the gangue from the coal material.

[0017] By adopting the above technical solutions, the various modules work together to form a complete closed-loop control system from image acquisition, recognition, positioning, timing calculation to physical sorting, realizing the integrated and intelligent operation of dry coal preparation equipment.

[0018] Preferably, the sorting module includes a plurality of high-frequency electromagnetic jet valves arranged at equal intervals along the width of the conveyor belt, and an air source distributor connected to the high-frequency electromagnetic jet valves.

[0019] By adopting the above technical solution, the equally spaced jet valve array can cover all transverse areas of the conveyor belt, and the air source distributor ensures the stability and balance of air pressure when each jet valve is working, thereby improving the overall execution accuracy of the sorting system.

[0020] Preferably, the image acquisition module includes an industrial camera mounted above the conveyor belt and strip LED auxiliary light sources symmetrically arranged on both sides of the industrial camera.

[0021] By adopting the above technical solution, the symmetrically arranged strip LED light source provides a uniform and high-brightness lighting environment, eliminating the interference of shadows on the image features of coal gangue, which is beneficial for industrial cameras to capture high-contrast images of the coal gangue surface.

[0022] Thirdly, this application provides a coal beneficiation device, including a processor and a memory, wherein the memory stores a computer program, and when the computer program is executed by the processor, it implements the steps of a coal beneficiation method.

[0023] By adopting the above technical solution and using dedicated hardware to carry the control algorithm, it can be easily deployed on the production line site to achieve industrial-grade coal and gangue sorting control.

[0024] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of a coal beneficiation method.

[0025] By adopting the above technical solution, a software carrier for the mineral processing method is provided, which facilitates program updates, porting, and distribution.

[0026] In summary, this application includes at least one of the following beneficial technical effects: 1. By employing bilateral filtering for noise reduction and histogram equalization enhancement, combined with a deep learning-based coal gangue identification and positioning network, the system can automatically extract the grayscale and texture features of coal and gangue at different depths. This effectively overcomes the problems of low efficiency in traditional manual sorting and poor identification accuracy caused by coal dust interference, achieving high-precision non-contact dry sorting. 2. Real-time running speed is adopted. With real-time center coordinates Dynamic calculation of jetting delay time Because it eliminates the interference of instantaneous speed fluctuations of the conveyor belt on the positioning distance, the opening and closing action of the high-frequency injection valve is precisely aligned with the instant the gangue passes through the nozzle, which significantly improves the injection hit rate and reduces the coal mis-injection rate during the sorting process. Attached Figure Description

[0027] Figure 1 This is a flowchart of a coal beneficiation method.

[0028] Figure 2 This is a flowchart and data flow diagram of coal gangue identification and location in a coal beneficiation method.

[0029] Figure 3 This is a schematic diagram of the overall structure of a coal beneficiation device.

[0030] Figure 4 This is a hardware connection block diagram of the control system for a coal beneficiation equipment. Detailed Implementation

[0031] The present application will be further described in detail below with reference to all the accompanying drawings.

[0032] This application discloses a coal beneficiation apparatus.

[0033] Reference Figure 1 and Figure 2 A coal beneficiation method, comprising: Step S1: Acquire the original image of the coal material on the conveyor belt using the image acquisition module, and perform bilateral filtering for noise reduction and histogram equalization enhancement on the original image to obtain a preprocessed image. In actual mineral processing operations, raw coal is conveyed forward by a conveyor belt at a certain speed. The image acquisition module is fixedly installed above the conveyor belt, with its lens pointing vertically downwards at the surface of the conveyor belt. Due to the slight vibration and coal dust flying during the operation of the conveyor belt, the acquired original image often contains a large number of high-frequency noise points caused by coal dust particles, and the lighting may be uneven.

[0034] To preserve the edge contours of the coal gangue while denoising, in this embodiment, bilateral filtering is used to filter the pixels in the original image. The grayscale value is replaced by a weighted average of neighboring pixels, with the weights determined by both the spatial domain kernel function and the range kernel function. The specific calculation formula is as follows: ; in, Represents the target pixel after filtering and denoising. The output grayscale value; This represents the original grayscale value of the neighboring pixels within the sliding window centered on the target pixel; ωω represents the radius of the sliding window, which is set to 5 here. The weighting coefficients are calculated using the following formula: ; in, The spatial kernel function is used to measure the influence of spatial distance on the weights, and is defined as follows: ; The standard deviation of the spatial domain is used to control the attenuation range of the spatial neighborhood; in this embodiment, the value is 3.0. This is the range kernel function, used to measure the influence of pixel grayscale similarity on the weights. Its definition is: ; The value is the standard deviation of the range, used to control the sensitivity of grayscale difference. In this embodiment, the value is set to 0.1.

[0035] After denoising by bilateral filtering, histogram equalization is used to stretch the grayscale range of the image to 0-255, thereby enhancing the contrast between coal and gangue in the image and obtaining a clear and stable preprocessed image.

[0036] Step S2: The preprocessed image is input into a preset coal gangue identification and localization network. The network extracts the shallow texture features and deep grayscale features of the preprocessed image and fuses them to obtain a coal gangue feature map. In this embodiment, combined with... Figure 2 The coal gangue identification and localization network adopts an improved YOLOv5 network. Its feature extraction backbone is based on the CSPDarknet53 architecture. To enhance the extraction capability of weak surface texture features of gangue, an SE (Squeeze-and-Excitation) attention mechanism module is introduced between the last C3 module and the SPPF (Spatial Pyramid Pooling) module in the backbone network. The SE attention mechanism module reweights the feature channels in two steps: first, a Squeeze operation is performed, which uses global average pooling to reduce the spatial dimension of the feature map. When compressed to 1×1, the calculation formula is: ; in, This represents the global spatial eigenvalue of channel c. This represents the c-th channel of the input feature map, where H and W represent the height and width of the feature map, respectively. Then, two fully connected layers and an activation function are used to capture the correlation between channels. The calculation formula is as follows: ; Where s is the calculated channel weight vector. This represents the Sigmoid activation function. Represents the ReLU activation function. and These are the weight matrices for the first and second fully connected layers, respectively. Finally, by multiplying each component of the weight vector s back onto the corresponding channel of the feature map u, the feature map is reweighted to generate a feature map with prominent coal gangue characteristics. The coal gangue identification and localization network was pre-trained on a computer. The training dataset contained 5000 images collected from the raw coal conveyor line, including 2500 images of gangue and 2500 images of pure coal. LabelImg was used for annotation to generate corresponding category labels (coal, gangue) and bounding box information for each target. During training, the dataset was divided into training, validation, and test sets in an 8:1:1 ratio. The loss function for model training was: ; in, The CIOU loss function is used to calculate the bounding box regression loss; The confidence loss is calculated using the binary cross-entropy loss (BCE Loss); The classification loss is calculated using BCE Loss; These are the corresponding weight coefficients, with values ​​of 0.05, 1.0, and 0.5 respectively. The number of training iterations is set to 300, the batch size is 16, and a stochastic gradient descent (SGD) optimizer with momentum is used, with momentum set to 0.937 and the initial learning rate set to 0.01.

[0037] Step S3: Perform target detection based on the coal gangue feature map, identify the gangue region, and extract the real-time center coordinates of the gangue region mapped onto the conveyor belt. The detection bounding box output by the localization network includes the center point of the target in pixel coordinates. The pixel coordinates are: width w and height h. Using a pre-calibrated rotation matrix R and translation vector T, the pixel coordinates are... Convert to real-time center coordinates in the physical world coordinate system of the conveyor belt ,in This represents the physical position of the gangue in the lateral direction of the conveyor belt. This represents the real-time physical position of the gangue along the longitudinal direction of the conveyor belt.

[0038] Step S4: Obtain the real-time operating speed of the conveyor belt And based on real-time center coordinates With respect to the preset horizontal baseline of the blow valve assembly The physical distance from the waste rock to the injection zone is calculated. A rotary encoder is installed on the drive shaft of the conveyor belt. By reading the pulse count of the PLC per unit time, the actual linear speed of the conveyor belt is calculated in real time. The physical distance SS from the current position of the waste rock to the nozzle row below is calculated using the following formula: ; In the formula, The preset physical baseline coordinates of the jet valve row in the longitudinal direction; The vertical real-time physical coordinates of the gangue; The compensation amount is for the inherent mechanical installation deviation of the system, and the unit is millimeters (mm).

[0039] Step S5: Based on physical distance and real-time running speed Calculate the delay time of the jet valve action When the timer reaches the delay time At that time, control the corresponding horizontal coordinate. The injection valve at the point is opened, using high-pressure gas to separate gangue from the coal material.

[0040] Calculate the delay time of the jet valve action The calculation formula is: ; In the formula, The delay time for the action of the blow valve, in seconds (s); The real-time operating speed of the conveyor belt is expressed in millimeters per second (mm / s). This is the inherent response delay time of the electromagnetic jet valve from receiving the control electrical signal to the airflow being ejected to the nozzle. In this embodiment, the value is taken as 0.005s.

[0041] To ensure sorting accuracy and conserve air, the duration of the blow valve being open is... It is proportional to the geometric dimensions of the gangue along the conveyor belt's direction of travel. The calculation formula is: ; In the formula, , is the opening duration of the high-frequency electromagnetic jet valve, in seconds (s); D is the physical geometric dimension of the gangue along the conveyor belt running direction, in millimeters (mm), which is obtained by multiplying the height h of the target detection box by the physical-to-pixel conversion ratio in image recognition; The real-time operating speed of the conveyor belt is expressed in millimeters per second (mm / s); k is a dimensionless control proportional coefficient, which is set to 1.2 in this case. The base blowing time, used to ensure that even the smallest pieces of gangue have sufficient momentum, is set to 0.02s.

[0042] The working principle of a coal beneficiation device according to an embodiment of this application is as follows: When raw coal mixed with gangue moves forward on a conveyor belt and passes through the image acquisition area, an industrial camera captures the image. The preprocessing algorithm quickly uses bilateral filtering to remove coal dust noise and enhance image contrast. Subsequently, a coal gangue identification and positioning network based on improved YOLOv5 extracts multi-dimensional features of coal gangue in terms of grayscale, reflectivity, and roughness from the image. The SE attention module in the positioning network enhances the response to weak coal gangue edges, accurately delineates the position of the gangue, and calculates its physical coordinates on the conveyor belt. The control system, combined with the real-time belt speed of the conveyor belt measured by the encoder, accurately calculates the physical time required for the gangue to move to the lower nozzle, and subtracts the hardware response time of the solenoid valve to obtain the precise delay time. When the timer returns to zero, the control system sends an opening signal to the jet valve of the corresponding transverse channel. The opening duration is dynamically determined by the size of the gangue and the belt speed. The high-pressure airflow that is sprayed instantaneously accurately knocks down the moving gangue, realizing automatic separation of coal and gangue.

[0043] Reference Figure 3 This application also discloses a coal beneficiation apparatus, comprising: The image acquisition module is used to acquire the original image of the coal material on the conveyor belt. In this embodiment, the image acquisition module uses an industrial area array camera with a resolution of 1920×1200 and a low-distortion industrial lens with a focal length of 8mm. The strip LED auxiliary light source is symmetrically and tilted on both sides of the industrial camera. In this embodiment, the light source has a brightness of 10000 Lux and a color temperature of 6000K. The image preprocessing module, which is electrically connected to the image acquisition module, is used to receive and perform bilateral filtering for noise reduction and histogram equalization enhancement on the original image to obtain a preprocessed image.

[0044] The coal gangue identification and localization module, connected to the image preprocessing module, is used to input the preprocessed image into a preset coal gangue identification and localization network, extract the shallow texture features and deep grayscale features of the preprocessed image to fuse them into a coal gangue feature map, identify the gangue region based on the coal gangue feature map, and extract the real-time center coordinates of the gangue region mapped on the conveyor belt. .

[0045] The delay control module, whose input is connected to the coal and gangue identification and positioning module and the rotary encoder, is used to obtain the real-time operating speed of the conveyor belt. Calculate the physical distance of the gangue to the injection zone, and base the calculation on the physical distance and real-time operating speed. Calculate the delay time of the jet valve action The sorting module is electrically connected to the control output of the delay control module, and is used to determine the time when the delay period is reached. At that time, control the corresponding horizontal coordinate. The injection valves at the point are opened, using high-pressure gas to separate gangue from the coal material. The sorting module includes 32 high-frequency electromagnetic injection valves arranged horizontally at equal intervals along the width of the conveyor belt. The center distance between adjacent injection valves is 25mm. The air inlets of all injection valves are connected to a common air source distributor to ensure that the air pressure of each nozzle is constant at 0.7MPa when it sprays.

[0046] As coal material moves along the conveyor belt, an industrial camera continuously captures images of the material surface under the illumination of a strip LED auxiliary light source. The image signals are then input to an image preprocessing module for bilateral filtering and noise reduction. The denoised images are transmitted to a coal gangue identification and positioning module for target detection and coordinate transformation, outputting the physical coordinates of the gangue. A delay control module obtains pulses from the rotary encoder on the conveyor belt drive shaft via the PLC's high-speed counter interface, calculates the belt speed, and, combined with the coordinates, calculates the delay time and starts an internal timer. When the timer reaches its set value, the main control board sends a high-speed drive signal to the drive circuit board corresponding to the lateral coordinate in the sorting module. This drives the high-frequency electromagnetic jet valve to open, and high-pressure air from the air source distributor is ejected from the corresponding nozzle, instantly forming a high-speed airflow that blows the passing gangue off its original throwing trajectory, achieving automatic sorting.

[0047] Reference Figure 4 This application also discloses a coal beneficiation device, including a processor and a memory. The memory stores a computer program, which, when executed by the processor, implements steps S1 to S5 of a coal beneficiation method.

[0048] Specifically, the processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices. In this embodiment, the processor specifically employs a combination of an industrial-grade GPU (such as the NVIDIA Jetson Xavier NX embedded processor) and a microcontroller unit (MCU, such as the STM32F407). The GPU is used to run deep learning algorithms, specifically performing forward inference calculations of the coal gangue identification and localization network to extract coal gangue feature maps and output center coordinates; the MCU is used to receive pulse input from the rotary encoder to calculate the real-time operating speed. And calculate the delay time based on physical distance and speed. This generates high-precision PWM control pulses to drive the high-frequency solenoid valve.

[0049] The memory, as a non-volatile computer-readable storage medium carrying computer programs, may include random access memory (RAM) and read-only memory (ROM), and may also include flash memory or solid-state drive (SSD). The memory stores a bootloader for system booting, an operating system (such as an embedded Linux system), an image processing library (such as OpenCV), a deep learning inference engine (such as TensorRT), and an application program that implements the core mineral processing algorithm of this invention.

[0050] This application also discloses a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements steps S1 to S5 of the coal beneficiation method as described in Embodiment 1.

[0051] The storage medium can be a computer-readable signal medium or a computer-readable storage medium. In this embodiment, the computer-readable storage medium is a non-volatile readable storage medium, specifically a USB flash drive, a portable hard drive, a read-only optical disc (CD-ROM), an integrated circuit memory card (CF card, SD card), or a server-side network storage device. The computer program stored on the storage medium is divided into multiple logical modules, including an image acquisition driving module, a bilateral filtering denoising algorithm module, a histogram equalization module, a coal gangue convolutional neural network inference module, a coordinate mapping transformation module, an encoder speed calculation module, a timing logic control module, and a jet valve drive output module.

[0052] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A method for beneficiating coal, characterized in that, Includes the following steps: Step S1: Obtain the original image of the coal material on the conveyor belt through the image acquisition module, and perform bilateral filtering denoising and histogram equalization enhancement on the original image to obtain a preprocessed image; Step S2: Input the preprocessed image into a preset coal gangue identification and positioning network, extract the shallow texture features and deep grayscale features of the preprocessed image through the coal gangue identification and positioning network, and fuse them to obtain a coal gangue feature map; Step S3: Based on the coal gangue feature map, perform target detection to identify the gangue region and extract the real-time center coordinates of the gangue region mapped onto the conveyor belt. ; Step S4: Obtain the real-time operating speed of the conveyor belt. And based on the real-time center coordinates With respect to the preset horizontal baseline of the blow valve assembly Calculate the physical distance of the gangue to the injection area; Step S5: Based on the physical distance and the real-time running speed Calculate the delay time of the jet valve action. When the timer reaches the aforementioned delay time At that time, control the corresponding horizontal coordinate. The injection valve at the point is opened, and high-pressure gas is used to separate the gangue from the coal material.

2. The coal beneficiation method according to claim 1, characterized in that, In step S1, the bilateral filtering denoising is to remove pixels from the original image. The grayscale value is replaced by the weighted average of the neighboring pixels, and the weights are determined by the spatial domain kernel function and the value domain kernel function.

3. The coal beneficiation method according to claim 1, characterized in that, In step S2, the coal gangue identification and positioning network adopts an improved YOLOv5 network. An attention mechanism module was introduced into its feature extraction backbone network to enhance the ability to extract weak surface texture features of gangue.

4. The coal beneficiation method according to claim 1, characterized in that, In step S4, the real-time running speed It is calculated by real-time acquisition of pulse signals by a rotary encoder installed on the drive roller of the conveyor belt.

5. A coal beneficiation method according to claim 1, characterized in that, In step S5, the pressure of the high-pressure gas is controlled between 0.6 MPa and 0.8 MPa, and the duration of opening the blow valve is proportional to the geometric dimensions of the gangue along the conveyor belt running direction.

6. A coal beneficiation apparatus, characterized in that, include: The image acquisition module is used to acquire raw images of the coal material on the conveyor belt; The image preprocessing module is used to perform bilateral filtering for noise reduction and histogram equalization enhancement on the original image to obtain a preprocessed image; The coal gangue identification and positioning module is used to input the preprocessed image into a preset coal gangue identification and positioning network, extract the shallow texture features and deep grayscale features of the preprocessed image to fuse them to obtain a coal gangue feature map, identify the gangue region based on the coal gangue feature map, and extract the real-time center coordinates of the gangue region mapped on the conveyor belt. ; The delay control module is used to obtain the real-time operating speed of the conveyor belt. Calculate the physical distance of the gangue to the injection area, and base the calculation on the physical distance and the real-time operating speed. Calculate the delay time of the jet valve action. ; The sorting module is used to execute the sorting module when the timer reaches the delay time. At that time, control the corresponding horizontal coordinate. The injection valve at the point is opened, and high-pressure gas is used to separate the gangue from the coal material.

7. A coal beneficiation apparatus according to claim 6, characterized in that, The sorting module includes a plurality of high-frequency electromagnetic jet valves arranged at equal intervals along the width of the conveyor belt, and an air source distributor connected to the high-frequency electromagnetic jet valves.

8. A coal beneficiation apparatus according to claim 6, characterized in that, The image acquisition module includes an industrial camera mounted above the conveyor belt and strip LED auxiliary light sources symmetrically arranged on both sides of the industrial camera.

9. A coal beneficiation equipment, characterized in that, It includes a processor and a memory, wherein the memory stores a computer program that, when executed by the processor, implements the steps of the coal beneficiation method as described in any one of claims 1 to 5.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the coal beneficiation method as described in any one of claims 1 to 5.

Images