A Coal Flow Monitoring and Control Method and System Based on Visual Intelligent Recognition

Abstract

This application discloses a coal flow monitoring and control method and system based on visual intelligent recognition, belonging to the field of industrial automation control technology. The method includes: real-time acquisition of visual image data of coal at the buffer bin outlet and gate area, as well as multi-source sensor data such as buffer bin material level, belt coal flow, gate opening, and feeder speed; extraction of coal state characteristics based on visual image data, combined with multi-source sensor data, and real-time analysis and prediction of coal flow conditions through a pre-trained prediction model, including three states: coal bridging warning, material blockage, and insufficient coal; and execution of corresponding hierarchical control strategies according to different conditions: during coal bridging warning, air cannons are activated for preventative clearing and gate opening is adjusted; during material blockage, air cannon operation is enhanced; and during insufficient coal, the feeding capacity of the feeder and belt conveyor is increased. This application enables intelligent monitoring and precise control of coal flow, improving production efficiency and equipment operational stability.

Description

Technical Field

[0001] This application relates to a method for monitoring and controlling coal flow, and more specifically, to a method for real-time monitoring and automatic control of coal flow based on visual intelligent recognition technology combined with multi-source sensor data. Background Technology

[0002] In railway coal loading operations, a meticulous "replenishment" operation is required at the end of the loading process to ensure that the loading weight accurately meets the standards. This stage typically involves reducing the opening of the replenishment gate to precisely control the coal flow. However, reducing the gate opening can lead to poor coal flow at the buffer bin outlet, increasing the risk of "coal bridging" (i.e., coal forming an arch at the outlet) or even complete "blockage." Once blockage occurs, the coal flow will be interrupted, severely impacting loading efficiency and operational continuity.

[0003] Currently, the industry generally adopts a "post-event handling" approach to address material blockage issues. This involves using air cannons installed on the buffer silo walls to clear the blockage after detecting an interruption in coal flow (material blockage). This method has significant time lag, cannot prevent material blockages from occurring, and the process itself causes operational interruptions. Furthermore, existing systems often struggle to accurately distinguish between "coal slab blockage" and simply "insufficient coal in the buffer silo." Accidentally triggering the air cannon when there is insufficient coal not only wastes energy but may also damage the silo structure due to unnecessary impact. Existing technologies largely rely on single flow or level sensors for judgment, lacking the ability to perceive the real-time accumulation state of coal and the intuitive physical characteristics of particle movement, thus failing to achieve proactive risk prediction and accurate differentiation of operating conditions.

[0004] Therefore, there is an urgent need for a solution that can integrate visual perception and multi-source data to achieve early warning of coal shed risks, accurate identification of working conditions, and intelligent hierarchical control, so as to fundamentally improve the stability, efficiency, and economy of the feeding stage. Summary of the Invention

[0005] In view of this, this application aims to provide a coal flow monitoring and control method and system based on visual intelligent recognition, so as to solve the problems of lagging material blockage treatment, unclear working conditions, and crude control in the prior art.

[0006] To achieve the above objectives, this application adopts the following technical solution:

[0007] In a first aspect, this application provides a coal flow monitoring and control method based on visual intelligent recognition, comprising the following steps:

[0008] Real-time acquisition of visual image data of coal at the outlet of the buffer bin and the gate area, as well as multi-source sensor data including the material level in the buffer bin, the coal flow rate on the conveyor belt, the gate opening and the speed of the coal feeder;

[0009] Based on the visual image data, coal condition features are extracted, and combined with the multi-source sensor data, the working conditions of the coal flow are analyzed and predicted in real time through a pre-trained prediction model. The working conditions include at least coal slab warning, material blockage, and insufficient coal.

[0010] According to different working conditions, corresponding graded control strategies are implemented. Specifically, when a coal dam warning is predicted, air cannons are activated for preventative clearing and gate opening is adjusted. When a blockage is predicted, air cannon operation is enhanced and gate opening is adjusted. When a coal shortage is predicted, the feeding capacity of the coal feeder and / or belt is increased, and air cannons are not activated.

[0011] Preferably, the step of extracting coal state features based on the visual image data includes: performing noise reduction and enhancement preprocessing on the visual image data; and extracting feature parameters related to coal flowability based on the preprocessed visual image data, wherein the feature parameters include at least coal accumulation morphology features (such as accumulation angle and contour) and particle motion features (such as flow velocity and trajectory).

[0012] Preferably, the prediction model is a machine learning-based fusion model, whose input includes at least feature parameters extracted from visual image data and the multi-source sensor data.

[0013] Preferably, the graded control strategy includes: for the coal damming warning condition, controlling the air cannon to operate in a first mode (e.g., low frequency, weak vibration) and controlling the gate opening to increase by a first opening value (e.g., 5%-8%); for the material blockage condition, controlling the air cannon to operate in a second mode (e.g., high frequency, strong vibration) with a frequency higher than the first mode, and controlling the gate opening to increase by a second opening value (e.g., 10%-15%), wherein the second opening value is greater than the first opening value; for the insufficient coal condition, controlling the increase of the feeder speed and / or belt conveyor speed.

[0014] Preferably, the method further includes: identifying the physical characteristics of the coal (such as particle size and moisture content) based on the visual image data, and dynamically adjusting the parameters of the prediction model and / or the grading control strategy according to the identified physical characteristics to adapt to the differences in fluidity of different coal types.

[0015] Secondly, this application provides a coal flow monitoring and control system based on visual intelligent recognition, used to implement the method described in any one of the first aspects, comprising:

[0016] The visual acquisition unit is used to acquire real-time visual images of coal at the outlet of the buffer bin and the gate area.

[0017] The multi-source sensing unit is used to collect multi-source sensor data, including buffer bin material level, belt coal flow rate, gate opening and coal feeder speed data.

[0018] The processing and prediction unit is used to receive and process the visual images and multi-source sensor data, and output the coal flow condition judgment result through the prediction model;

[0019] The control execution unit is used to execute corresponding control operations based on the working condition judgment result. The control operations include adjusting one or more of the following: the air cannon, the gate opening, the coal feeder speed, and the belt conveyor speed.

[0020] Preferably, the vision acquisition unit includes an industrial camera, which has protective performance to adapt to dusty environments and is installed in a position that can cover the gate and the material drop area.

[0021] Preferably, the control execution unit includes an air cannon controller that supports multi-level intensity adjustment, an actuator for precisely adjusting the gate opening, and a frequency converter for adjusting the speed of the coal feeder and belt.

[0022] Thirdly, this application provides a coal flow monitoring and control system based on visual intelligent recognition, including the device described in any of the second aspects, as well as a centralized control module and a human-machine interface. The centralized control module is used to coordinate the collaborative work and data interaction of various units within the device, forming a closed loop of "perception-prediction-control-feedback". The human-machine interface is used for parameter setting, status monitoring, and manual intervention.

[0023] Compared with the prior art, this application has the following beneficial effects:

[0024] Through visual intelligent analysis, early warnings can be issued 3-5 seconds before coal blockage actually occurs, and preventive measures can be initiated, nipping the risk of material blockage in the bud and virtually eliminating loading interruptions caused by material blockage, thus improving operational continuity. In addition, by integrating visual and multi-source data, it can clearly distinguish between "material blockage risk" and "insufficient coal," avoiding the accidental activation of air cannons when there is insufficient coal, significantly reducing energy consumption (air cannon energy consumption can be reduced by more than 40%) and equipment wear. Attached Figure Description

[0025] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application. Furthermore, these drawings and textual descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concepts of this application to those skilled in the art through reference to specific embodiments.

[0026] Figure 1 A schematic diagram of the architecture of a coal flow monitoring and control system based on visual intelligent recognition provided in an embodiment of this application;

[0027] Figure 2A schematic diagram of the installation structure of a coal flow monitoring and control system based on visual intelligent recognition provided in an embodiment of this application;

[0028] Figure 3 A flowchart illustrating a coal flow monitoring and control method based on visual intelligent recognition, provided for an embodiment of this application;

[0029] Figure 4 This is a flowchart illustrating a hierarchical control strategy provided in an embodiment of this application. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the embodiments of this application. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0031] Example 1

[0032] This embodiment provides a coal flow monitoring and control method based on visual intelligent recognition. This method achieves precise control of coal flow through multi-source data fusion and intelligent prediction.

[0033] This method is based on Figure 1 The system architecture implementation shown is as follows: Figure 1 As shown, the system mainly includes a perception layer, a data processing layer, a control execution layer, and a centralized control layer.

[0034] The perception layer comprises a visual acquisition unit and a multi-source sensing unit, responsible for real-time acquisition of coal images and various operating condition data. The visual acquisition unit can be an industrial camera, while the multi-source sensing unit can include level sensors, flow sensors, opening sensors, and speed sensors. In practice, a 20-megapixel industrial high-definition camera can be installed 1.8m to the side of the buffer silo outlet, with the lens at a 45° angle to the gate center, ensuring the acquisition range covers the full opening area of ​​the gate and the material drop point below. An ultrasonic radar dual-mode level sensor is installed on the top of the buffer silo to collect buffer silo level data; an opening sensor is installed at the bottom outlet of the buffer silo to collect gate opening data; a weighing flow sensor is installed on the belt conveyor to collect belt coal flow data; and a speed sensor is installed on the feeder motor shaft to collect feeder speed data. Furthermore, industrial Ethernet is used to connect each sensor to the data processing unit, with data transmission latency controlled within 50ms.

[0035] The data processing layer includes edge computing devices, comprising an image preprocessing module, a prediction model, a control strategy module, and a storage module. Internally, it performs image preprocessing, feature extraction, prediction model (such as a CNN-GBT fusion model), and control strategy algorithms, responsible for real-time analysis of the perception layer data and generating control commands. The control execution layer includes air cannon controllers, gate electric actuators, and frequency converters for coal feeders and belt conveyors, responsible for executing specific control actions. The centralized control layer includes a PLC centralized control module, a human-machine interface (HMI), and an operator console, responsible for system integration, process coordination, data communication, and human-machine interaction, forming a complete closed-loop control system.

[0036] Please see Figure 2 This is a typical installation layout for the system on site. An industrial camera is installed approximately 1.5-2 meters to the side of the buffer silo outlet, with the lens angled towards the feeding gate and the material drop area below to ensure complete capture of coal accumulation and flow images. An ultrasonic or radar level gauge is installed on top of the buffer silo. A weighing belt scale is installed on the belt conveyor below the gate. Air cannons are evenly installed on the lower inner wall of the buffer silo. The feeder's rotational speed is monitored by a sensor on its motor shaft. All equipment is connected to the processing unit and PLC in the central control room via an industrial network.

[0037] Based on the above system architecture, this method is executed when the loading operation enters the tail-end replenishment stage.

[0038] like Figure 3 As shown, the method includes the following steps:

[0039] Step S1: Real-time acquisition of visual image data of coal at the outlet of the buffer bin and the gate area, as well as multi-source sensor data including the material level of the buffer bin, the coal flow rate of the conveyor belt, the gate opening and the speed of the coal feeder.

[0040] The visual image data can be acquired through a visual acquisition unit, specifically an industrial camera. Sensor data such as buffer bin level, belt coal flow, gate opening, and feeder speed are acquired sequentially through a level sensor, flow sensor, opening sensor, and speed sensor.

[0041] Next, proceed to step S2: extract coal condition features based on visual image data, and combine multi-source sensor data to analyze and predict the working conditions of coal flow in real time through a pre-trained prediction model.

[0042] Specifically, after collecting the relevant data, it cannot be applied directly; the coal condition characteristics need to be extracted first.

[0043] In some embodiments, step S2, extracting coal state features based on visual image data, may specifically include: performing noise reduction and enhancement preprocessing on the visual image data; and extracting feature parameters related to coal flowability based on the preprocessed visual image data. The feature parameters include at least coal packing morphology features (e.g., packing angle and amount of adhesion) and particle motion features (e.g., particle velocity).

[0044] Specifically, denoising and enhancement preprocessing of visual image data can remove image noise and improve image quality. In some embodiments, the acquired images are preprocessed with Gaussian filtering for noise reduction and histogram equalization for enhancement to eliminate dust and lighting interference.

[0045] Based on preprocessed visual image data, feature parameters related to coal flowability are extracted. These parameters include coal packing morphology and particle motion characteristics. Coal packing morphology is analyzed by examining the coal packing angle to assess blockage risk, while particle motion characteristics are monitored by monitoring changes in particle velocity to evaluate flow status. Specifically, edge detection algorithms can be used to extract the contours of the coal against the bin walls and gate edges, and the coal packing angle in key areas can be calculated. Furthermore, algorithms such as optical flow are used to analyze consecutive frames of images to calculate the average velocity of coal particles. Simultaneously, features such as the coal adhesion area around the gate can be statistically analyzed.

[0046] After extracting the feature data, the feature parameters extracted from the image (pile angle, particle velocity, adhesion amount, etc.) are fused with the real-time multi-source sensor data (material level, flow rate, opening degree, etc.) and input into the pre-trained prediction model.

[0047] The prediction model employs a machine learning-based fusion model, specifically a CNN-GBT fusion prediction model. Its inputs include feature parameters extracted from visual image data and multi-source sensor data. The prediction model is trained using 800 sets of historical operating condition data, covering scenarios with dry coal moisture content of 5%-10%, wet coal moisture content of 15%-25%, fine particle size less than 5mm, coarse particle size of 5-20mm, and different gate opening degrees of 10%-30%. A 35° stacking angle warning threshold, a 45° blockage threshold, and a 20% insufficient material level threshold are set. The operating condition prediction results include a risk-free normal state (stacking angle <35°, normal particle velocity, stable flow rate), a coal bridging warning (stacking angle 35°-45°, decreased particle velocity, slight flow rate fluctuation), a blockage state (stacking angle ≥45°, particle velocity approaching 0, sudden flow rate drop ≥80%), and an insufficient coal level (material level below 20% of the preset threshold, stable flow rate, no visual stacking anomalies).

[0048] Taking the CNN-GBT fusion prediction model as an example, the following supplementary explanations are provided for specific details:

[0049] This model employs a cascaded fusion architecture. First, the preprocessed visual image of the coal is input into a lightweight CNN subnetwork (e.g., containing 3-4 convolutional layers, pooling layers, and fully connected layers). This CNN subnetwork is responsible for automatically learning and extracting deeper, more abstract feature maps from the image. These features go beyond manually extracted basic parameters such as stacking angle and flow velocity, and may contain complex information such as texture variations and local motion patterns. The CNN subnetwork outputs a high-dimensional feature vector (visual depth features).

[0050] Meanwhile, coal state features such as coal accumulation angle and particle flow rate extracted from the image using traditional algorithms, along with multi-source sensing data from sensors (buffer bin level, real-time coal flow rate, gate opening, and feeder speed), together form a set of structured feature vectors.

[0051] Next, the visual depth feature vector output by the CNN is concatenated with the above structured feature vector to form a comprehensive feature vector that integrates visual depth information, visual basic physical information, and multi-source sensor information.

[0052] The combined feature vector is then input into a Gradient Boosting Tree (GBT) classifier for final condition determination. GBT is a powerful ensemble learning algorithm that excels at handling heterogeneous features (such as a mixture of image and numerical features) and effectively captures complex nonlinear relationships between features. It has advantages in distinguishing subtle differences in operating conditions such as "coal shed warning," "blocked material," and "insufficient coal."

[0053] For model training, the fusion model is trained offline using a large amount of historical operating condition data. The training samples need to cover various typical scenarios, including different coal moisture content (dry coal, wet coal), different particle sizes (fine pulverized coal, lump coal), different gate openings (10%-30%), and corresponding label data for "no risk," "coal shed warning," "blocked," and "insufficient coal." Through supervised learning, the parameters of CNN and GBT are optimized, enabling the model to accurately establish the mapping relationship from fused features to operating condition categories. After training, the model is deployed on edge computing devices for online real-time inference.

[0054] Ultimately, the CNN-GBT fusion model outputs the probability or discrimination results for four types of working conditions: "No Risk", "Coal Shed Warning", "Material Blockage", and "Insufficient Coal". Its advantages are: 1. It utilizes CNN to enhance the system's ability to autonomously learn key signs from raw images, without relying entirely on manually designed features; 2. It uses GBT to efficiently fuse heterogeneous data, resulting in strong comprehensive judgment capabilities; 3. It can achieve high accuracy (e.g., accuracy ≥ 98%) and early warning (e.g., 3-5 seconds).

[0055] Then, proceed to step S3: execute the corresponding hierarchical control strategy according to different operating conditions.

[0056] The graded control strategy includes three modes, corresponding to three abnormal states: coal slab warning, material blockage, and insufficient coal.

[0057] When a coal blockage warning is predicted, the air cannon is activated for preventative clearing and the gate opening is adjusted; when a blockage is predicted, the air cannon operation is enhanced and the gate opening is adjusted; when a coal shortage is predicted, the feeding capacity of the coal feeder and / or belt is increased, and the air cannon is not activated.

[0058] More specifically, the hierarchical control strategy includes:

[0059] In response to the coal shed warning situation, the air cannon is controlled to operate in the first mode and the opening of the gate is increased to the first opening value; this mode aims to restore the fluidity of the coal by slightly disturbing it with the air cannon.

[0060] For situations where material blockage has occurred, the air cannon is controlled to operate in a second mode at a higher frequency than the first mode, and the gate opening is controlled to increase to a second opening value, which is greater than the first opening value; this mode is designed to solve the material blockage problem.

[0061] For situations where there is insufficient coal, control and increase the speed of the coal feeder and / or the belt conveyor speed to accelerate the replenishment of coal to the buffer bin.

[0062] Specifically, such as Figure 4 As shown, after determining the operating condition type, if it is "no risk": keep the air cannon closed, keep the gate opening at its initial value (20%-30%), and keep the coal feeder / belt speed at its current set value, that is, keep the system's current operating parameters unchanged.

[0063] If it is a "coal shed warning": control the air cannon to enter the low-frequency prevention mode (such as a frequency of 1-2 times / second, lasting for 2-3 seconds), and at the same time slightly increase the gate opening by 5%-8% to slightly disturb the coal and restore its fluidity.

[0064] If the blockage is "material blockage": Control the air cannon to enter high-frequency strong vibration mode (e.g., frequency 3-4 times / second) and continue operation until the blockage is cleared. At the same time, significantly increase the gate opening by 10%-15% and temporarily reduce the belt speed by about 10% to mitigate the impact caused by the sudden drop of a large amount of coal after the blockage is cleared.

[0065] If the error message is "insufficient coal": Do not start the air cannon. Increase the feeder speed by 15%-20% and the belt speed by 10%-15% using the frequency converter to accelerate the replenishment of coal into the buffer bin until the material level is restored to above the safe threshold.

[0066] Furthermore, after the control action is executed, the system continuously monitors changes in coal flow rate, material level, and visual characteristics, and uses this feedback data to evaluate the control effect. This data is also recorded and can be used for subsequent iterative optimization of the prediction model parameters, enabling the system to have self-learning and adaptive capabilities. In particular, newly generated "intervention success" or "intervention failure" cases can serve as new training samples, periodically incrementally training or fine-tuning the CNN-GBT model to continuously adapt to changes in on-site conditions and maintain prediction accuracy.

[0067] In addition, in some embodiments, the above method may further include: identifying the physical characteristics of coal based on visual image data, and dynamically adjusting the parameters of the prediction model and / or the graded control strategy according to the identified physical characteristics; wherein, the physical characteristics include the particle size and moisture content of the coal.

[0068] Specifically, image analysis technology can identify the particle size distribution and surface moisture level of coal, thereby adjusting the warning threshold and control parameters according to different coal characteristics, improving the system's adaptability and accuracy. For example, by analyzing the texture and color of visual images, the moisture content (dry / wet) and particle size (coarse / fine) of the coal can be preliminarily determined. For easily agglomerated coal types such as wet coal or fine coal powder, the system will automatically lower the stacking angle threshold for "coal shed warning" and appropriately increase the intensity or frequency of air cannon prevention operations, thus achieving precise control "tailored to the coal type." This adaptive adjustment can also be achieved by updating the weights of features related to coal characteristics in the prediction model (such as the GBT classifier) ​​online.

[0069] In a specific implementation case, this system was applied at a loading station. During system operation, when the visual recognition module detected that the coal accumulation angle gradually increased to 36° and the particle velocity dropped to 28% of the initial value, the predictive model output a coal slag warning signal. The control execution module immediately activated the air cannon for low-frequency operation for 2.5 seconds, while simultaneously fine-tuning the gate opening from 25% to 29%, while maintaining the feeder and belt speeds unchanged. One second after intervention, the coal accumulation angle dropped to 32°, the particle velocity recovered to 75% of the initial value, and the warning was lifted. When the level sensor showed that the material level had dropped to 18%, the model determined that there was insufficient coal. The system immediately increased the feeder speed by 18% and the belt speed by 12%, without activating the air cannon, and restored the initial feeding parameters after the material level rose back to 30%.

[0070] This method achieves accurate prediction and graded control of coal flow through visual intelligent recognition and multi-source data fusion, effectively preventing coal blockage and improving the stability and accuracy of coal flow control. Coal flow fluctuations are controlled within ±3.2%, significantly improving the efficiency and safety of coal loading operations.

[0071] Example 2

[0072] This embodiment provides a coal flow monitoring and control system based on visual intelligent recognition, which is used to implement the monitoring and control method described in Embodiment 1. The system achieves intelligent monitoring and precise control of coal flow through the coordinated cooperation of multiple functional units.

[0073] The system mainly consists of four core parts: a visual acquisition unit, a multi-source sensing unit, a processing and prediction unit, and a control and execution unit.

[0074] The visual acquisition unit is responsible for acquiring real-time visual images of the coal at the buffer bin outlet and gate area. In some embodiments, this unit includes an industrial camera with protective features to withstand dusty environments, and is installed in a position that covers the gate and the material drop area. As described in Embodiment 1, the industrial camera uses a 20-megapixel high-definition sensor and is installed 1.8m to the side of the buffer bin outlet, with the lens at a 45° angle to the center of the gate, ensuring that the acquisition range covers the fully open area of ​​the gate and the material drop point below. The industrial camera is equipped with a dustproof and shockproof housing, enabling it to operate stably in the harsh environment of coal loading and continuously acquire clear images of coal flow, providing a reliable data foundation for subsequent intelligent identification and analysis.

[0075] The multi-source sensing unit is used to collect multi-source sensor data, including buffer silo level, belt coal flow, gate opening, and feeder speed. This unit integrates various sensor devices, including an ultrasonic radar dual-mode level sensor installed on the top of the buffer silo, an opening sensor at the bottom outlet, a weighing flow sensor on the belt conveyor, and a speed sensor at the feeder motor shaft end. These sensors are connected to the data processing unit via industrial Ethernet to achieve real-time data acquisition and transmission, with transmission latency controlled within 50ms, ensuring the timeliness and accuracy of the system response.

[0076] The processing and prediction unit is the core intelligent module of the device, used to receive and process visual images and multi-source sensor data, and output coal flow condition judgment results through a prediction model. This unit adopts a CNN-GBT fusion prediction model, as described in detail in Example 1, and is trained by importing 800 sets of historical operating condition data, covering scenarios with different humidity, particle size, and gate opening. The processing and prediction unit first performs noise reduction and enhancement preprocessing on the visual image data, and then extracts feature parameters related to coal flowability, including coal accumulation morphology and particle movement characteristics. It combines multi-source sensor data for comprehensive analysis to judge the coal flow condition status in real time, including three abnormal situations: coal bridging warning, material blockage, and insufficient coal.

[0077] The control execution unit executes corresponding control operations based on the working condition judgment results. These operations include adjusting one or more of the following: the air cannon, the gate opening, the feeder speed, and the belt conveyor speed. This unit includes an air cannon controller supporting multi-level intensity adjustment, an actuator for precisely adjusting the gate opening, and a variable frequency speed control device for adjusting the feeder and belt speeds. The air cannon controller can switch between low-frequency and high-frequency modes. The low-frequency mode operates at 1.5 times / second for 2.5 seconds, while the high-frequency mode operates at 3.5 times / second, employing appropriate clearing intensities based on different levels of material blockage. The gate opening actuator enables precise fine-tuning control with an adjustment step of 0.5%, ensuring accurate regulation of coal flow. The variable frequency speed control device can promptly adjust the feeder speed and belt conveyor speed when there is insufficient coal, improving feeding capacity.

[0078] The various functional units communicate and coordinate control via industrial Ethernet. When the vision acquisition unit detects an abnormal increase in the coal accumulation angle, the processing and prediction unit immediately analyzes and determines the operating condition, and the control execution unit executes corresponding control actions according to the preset hierarchical control strategy. The entire process achieves closed-loop management from data acquisition and intelligent analysis to precise control, effectively preventing coal bridging and blockage.

[0079] The device also has a dynamic parameter adjustment function, which can identify the physical characteristics of coal, including particle size and moisture, based on visual image data, and dynamically adjust the parameters of the prediction model and the graded control strategy according to the identified physical characteristics, thereby improving the system's adaptability to different coal characteristics.

[0080] Through the coordinated operation of each unit, the device achieves intelligent monitoring and precise control of coal flow, keeping coal flow fluctuations within ±3.2%, significantly improving the efficiency and safety of coal loading operations, and providing reliable technical support for the automation upgrade of the coal industry.

[0081] It is understood that the same or similar parts in the above embodiments can be referred to each other, and the contents not described in detail in some embodiments can be referred to the same or similar contents in other embodiments.

[0082] It should be noted that in the description of this application, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance. Furthermore, in the description of this application, unless otherwise stated, "a plurality of" means at least two.

[0083] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the function involved, as will be understood by those skilled in the art to which embodiments of this application pertain.

[0084] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0085] Those skilled in the art will understand that all or part of the steps of the methods implementing the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0086] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc.

[0087] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0088] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A coal flow monitoring and control method based on visual intelligent recognition, characterized in that, Includes the following steps: Real-time acquisition of visual image data of coal at the outlet of the buffer bin and the gate area, as well as multi-source sensor data including the material level in the buffer bin, the coal flow rate on the conveyor belt, the gate opening and the speed of the coal feeder; The visual image data is preprocessed for denoising and enhancement; based on the preprocessed visual image data, feature parameters related to coal flowability are extracted, the feature parameters including at least coal accumulation morphology features and particle movement features; Combining the multi-source sensor data, the working conditions of coal flow are analyzed and predicted in real time through a pre-trained prediction model. The prediction model is a fusion model based on machine learning, and its input includes at least feature parameters extracted from visual image data and the multi-source sensor data. The working conditions include at least coal shed warning, material blockage, and insufficient coal. Based on different operating conditions, corresponding hierarchical control strategies are implemented. Specifically, when a coal sluice gate is predicted to be blocked, the air cannon is activated in the first mode to prevent blockage and the gate opening is increased by a first value. When a blockage is predicted, the air cannon is controlled to operate in the second mode at a higher frequency than the first mode and the gate opening is increased by a second value, which is greater than the first value. When a coal shortage is predicted, the feeder speed and / or belt conveyor speed are controlled, and the air cannon is not activated. The physical characteristics of the coal are identified based on the visual image data, and the parameters of the prediction model and / or the graded control strategy are dynamically adjusted according to the identified physical characteristics; wherein, the physical characteristics include the particle size and moisture content of the coal.

2. A coal flow monitoring and control system based on visual intelligent recognition, characterized in that, To implement the method of claim 1, the method comprises: The visual acquisition unit is used to acquire real-time visual images of coal at the outlet of the buffer bin and the gate area. The multi-source sensing unit is used to collect multi-source sensor data, including buffer bin material level, belt coal flow rate, gate opening and coal feeder speed data. The processing and prediction unit is used to receive and process the visual images and multi-source sensor data, and output the coal flow condition judgment result through the prediction model; The control execution unit is used to execute corresponding control operations based on the working condition judgment result. The control operations include adjusting one or more of the following: the air cannon, the gate opening, the coal feeder speed, and the belt conveyor speed.

3. The system according to claim 2, characterized in that, The vision acquisition unit includes an industrial camera, which has protective performance to adapt to dusty environments and is installed in a position that can cover the gate and the material drop area.

4. The system according to claim 2, characterized in that, The control execution unit includes an air cannon controller that supports multi-level intensity adjustment, an actuator for precisely adjusting the gate opening, and a frequency converter for adjusting the speed of the coal feeder and belt.

Images