Intelligent control method and system for coal separation under mine based on mine-hung system

Abstract

The application relates to the technical field of coal separation, intelligent control and mineral processing, in particular to a coal separation intelligent control method and system based on a mine system. Physical property measurement values and motion state parameters of edge particles in a separation device are acquired; the edge particles are particles with spatial positions in a separation area and current motion trajectories tending to or having entered a preset critical area between different collection channels; whether the separation behavior of the edge particles deviates from a preset target is determined based on the physical property measurement values and the motion state parameters; when it is determined that the preset target is deviated, micro correction instructions for the edge particles are generated; wherein the micro correction instructions include a physical intervention direction, an action point and an action force intensity; and the edge particles are physically intervened by a driving execution mechanism based on the micro correction instructions. The application can effectively separate the edge particles in coal.

Description

Technical Field

[0001] This application relates to the technical fields of coal sorting, intelligent control and mineral processing, and in particular to an intelligent control method and system for coal sorting in underground mines based on a mining system. Background Technology

[0002] Coal, as a vital energy resource, plays a crucial role in ensuring energy supply and reducing environmental pollution through efficient sorting. With the advancement of intelligent mining, intelligent sorting technology based on an industrial internet operating system has become an industry trend. Huawei's Mining Hong system, a dedicated industrial internet operating system for the mining industry, achieves unified interconnection of underground equipment through distributed soft bus technology, providing a new technological platform for intelligent control of the coal sorting process.

[0003] Currently, traditional coal sorting technologies mainly rely on differences in the macroscopic physical properties of particles, achieving separation through methods such as jigging, air separation, or heavy media separation. Automated control systems typically employ multi-sensor data acquisition, combined with data analysis modules to predict coal quality, and use control algorithms to adjust the parameters of the sorting equipment. In the application environment of the Kuanghong system, some mines have achieved unified equipment access and basic data acquisition, but the refined control of the sorting process still relies on traditional methods, failing to fully leverage the advantages of the Kuanghong system's distributed computing and real-time communication.

[0004] However, traditional coal sorting technology has significant drawbacks when dealing with "edge particles": when the physical properties of the particles approach the separation threshold between coal and gangue, their trajectory is easily affected by minute disturbances within the equipment, leading to uncertainty. Even in a mining system environment, traditional sorting algorithms still use fixed threshold judgment rules, lacking real-time perception and dynamic intervention capabilities for the microscopic motion of particles, resulting in low sorting accuracy for edge particles and causing loss of clean coal or contamination with gangue. Summary of the Invention

[0005] Therefore, it is necessary to provide an intelligent control method and system for coal sorting in mines based on a mining system, which can effectively sort edge particles, to address the above-mentioned technical problems.

[0006] Firstly, this application provides an intelligent control method for coal sorting in underground mines based on a mining system, the method comprising:

[0007] Obtain the physical property measurements and motion state parameters of edge particles in the sorting equipment; edge particles are those whose spatial location is within the diversion area and whose current motion trajectory tends to or has entered the preset critical area between different collection channels;

[0008] Based on physical property measurements and motion state parameters, it is determined whether the divergence behavior of edge particles deviates from the preset target.

[0009] When a deviation from the preset target is detected, a micro-correction instruction is generated for the edge particles; the micro-correction instruction includes the physical intervention direction, point of application, and force intensity.

[0010] Based on microscopic correction commands, the actuator is driven to physically intervene in edge particles.

[0011] In one embodiment, obtaining physical property measurements of the edge particles includes:

[0012] Obtain the original measurements of the intrinsic material properties of the edge particles;

[0013] The surface condition of edge particles is determined using a vision system;

[0014] When the edge particle is a single particle, the particle independence verification result of the edge particle is determined by the vision system, and when the edge particle is a particle unit, the consistency index of the particle unit is determined.

[0015] Based on the particle surface state, as well as the particle independence verification results or particle unit consistency index, the original measured values ​​of the intrinsic properties of the material are corrected to obtain the key physical property measured values ​​of the edge particles.

[0016] In one embodiment, obtaining the motion state parameters of the edge particles includes:

[0017] The motion information of edge particles collected by visual sensors and the motion information of edge particles collected synchronously by millimeter-wave radar are fused to obtain multi-sensor fusion information.

[0018] The motion state characteristics can be obtained by analyzing a single particle using a particle identity prediction and tracking algorithm, or by analyzing particle units using a grid statistical method.

[0019] Extract the motion state parameters of edge particles from the motion state features.

[0020] In one embodiment, determining whether the divergence behavior of edge particles deviates from a preset target based on physical property measurements and motion state parameters includes:

[0021] Based on physical property measurements and motion state parameters, generate particle classification features and current motion state features;

[0022] Based on particle classification features, the desired collection channels for edge particles are determined;

[0023] Select the set of target boundaries corresponding to the desired collection channel from the preset set of ideal space-motion state boundaries;

[0024] The current motion state features are compared with the target boundary set to obtain the comparison results;

[0025] Based on the comparison results, it is determined whether the flow separation behavior of edge particles deviates from the preset target.

[0026] In one embodiment, based on the comparison results, determining whether the divergence behavior of edge particles deviates from a preset target includes:

[0027] Obtain the environmental fluid disturbance parameters and the response characteristics of the actuator in the diversion zone;

[0028] Based on the perturbation parameters and response characteristics, the deviation threshold for determining whether edge particles deviate from the expected trajectory is dynamically adjusted;

[0029] If the comparison result exceeds the deviation threshold, it is determined that the flow of edge particles deviates from the preset target.

[0030] In one embodiment, generating micro-correction instructions for edge particles includes:

[0031] Based on the comparison results, the physical deviation vector between the edge particles and the target boundary set is quantized;

[0032] Based on the physical deviation vector, microscopic correction instructions are generated for edge particles.

[0033] In one embodiment, after the driving actuator physically intervenes in the edge particles, the method further includes:

[0034] Obtain the motion state of edge particles after physical intervention is applied;

[0035] The correction error is determined by comparing the motion state of the edge particles with the expected correction effect corresponding to the micro-correction command.

[0036] Based on the correction error, the physical correction response mapping rules are adaptively adjusted.

[0037] In one embodiment, generating micro-correction instructions for edge particles includes:

[0038] The energy consumption limits of the actuator, the overall processing capacity of the sorting equipment, and the interaction effects between multiple edge particles are obtained.

[0039] Based on energy consumption constraints, overall processing capacity, interaction effects, and physical deviation vectors, a multi-objective optimization algorithm is used to generate microscopic correction instructions for edge particles.

[0040] In one embodiment, based on microscopic correction commands, the actuator is driven to physically intervene in edge particles, including:

[0041] Obtain operational performance data of the actuator;

[0042] Compare the operational performance data with the expected performance corresponding to the micro-correction instructions to determine whether there is a performance deviation.

[0043] When performance deviations exist, the parameters of the micro-correction command are adaptively adjusted;

[0044] The adjusted micro-correction command is sent to the control interface of the actuator to drive the actuator to physically intervene in the edge particles.

[0045] Secondly, this application also provides an intelligent control system for coal sorting in underground mines based on a mining system, the system comprising:

[0046] The parameter acquisition module is used to acquire the physical property measurement values ​​and motion state parameters of edge particles in the sorting equipment; edge particles are particles whose spatial location is within the diversion area and whose current motion trajectory tends to or has entered the preset critical area between different collection channels;

[0047] The deviation judgment module is used to determine whether the divergence behavior of edge particles deviates from the preset target based on physical property measurement values ​​and motion state parameters.

[0048] The instruction generation module is used to generate micro-correction instructions for edge particles when it is determined that the particle deviates from the preset target; the micro-correction instructions include the physical intervention direction, the point of application, and the intensity of the force.

[0049] The intervention execution module is used to drive the actuator to physically intervene in edge particles based on micro-correction instructions.

[0050] The beneficial effects of this application are as follows: The above-mentioned intelligent control method and system for coal sorting in mines based on the mining system achieves accurate capture of transient microscopic physical interaction signals of particles in the diversion critical region by acquiring the physical property measurement values ​​and motion state parameters of edge particles; it determines whether the diversion behavior deviates from the preset target based on these dynamic parameters, rather than relying on static thresholds; when a deviation is detected, it generates microscopic correction instructions containing the physical intervention direction, point of action, and force intensity, driving the actuator to perform precise physical intervention on the edge particles, realizing refined and proactive guidance of particle trajectories. Therefore, it can effectively identify the deviation trend of the diversion behavior of edge particles, thereby significantly improving the coal sorting accuracy, reducing clean coal loss, and enhancing the system's adaptive control capability in complex underground mining environments. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 This is a flowchart illustrating an intelligent control method for coal sorting in underground mines based on a mining system, as shown in one embodiment.

[0053] Figure 2 This is a flowchart of the steps for obtaining physical property measurement values ​​in one embodiment;

[0054] Figure 3 A flowchart of the motion state parameter acquisition steps in one embodiment;

[0055] Figure 4 This is a flowchart of the deviation determination step for flow diversion behavior in one embodiment;

[0056] Figure 5 This is a flowchart of the dynamic adjustment of the deviation threshold step in one embodiment;

[0057] Figure 6 A flowchart of the micro-correction instruction generation steps in one embodiment;

[0058] Figure 7 This is a flowchart of the physical correction response mapping rules and adaptive adjustment steps in one embodiment;

[0059] Figure 8 A flowchart of the micro-correction instruction generation steps for multi-objective optimization in one embodiment;

[0060] Figure 9 This is a flowchart of the actuator performance monitoring and adjustment steps in one embodiment;

[0061] Figure 10 This is a schematic diagram of the structure of an intelligent control system for coal sorting in mines based on a mining system, as shown in one embodiment. Detailed Implementation

[0062] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0063] In underground coal sorting, when particle trajectories approach the critical line of the diversion zone, some particles deviate from the expected diversion path due to environmental disturbances and particle characteristics, leading to a decrease in sorting accuracy. Related technologies typically mitigate this problem by widening the diversion channel, but this reduces sorting accuracy, creating a technical contradiction. This application provides an intelligent control method for underground coal sorting based on a mining system, which improves sorting accuracy by identifying edge particles within the critical zone and implementing regional intervention.

[0064] To facilitate understanding of this embodiment, the key terms involved are defined as follows:

[0065] Edge particles: Particles or groups of particles whose spatial location is within the diversion area and whose current trajectory tends to or has entered the preset critical area between different collection channels;

[0066] Preset critical region: A specific width region defined by the physical structure of the diversion device;

[0067] Microscopic correction instructions: These are instantaneous control instructions that include the direction, area, and intensity of physical intervention.

[0068] In one exemplary embodiment, such as Figure 1 As shown, an intelligent control method for coal sorting in underground mines based on a mining system is provided. This method includes:

[0069] In step S101, the physical property measurements and motion state parameters of the edge particles in the sorting device are obtained.

[0070] Optionally, an industrial vision system is installed above the diversion area to capture particle images. Continuously acquired images are processed in real time, including image enhancement, target segmentation, and feature extraction. The industrial vision system features a dust-resistant design to ensure stable operation in underground mining environments.

[0071] First, the diversion area was divided into 10mm × 10mm grid cells, and particle information within each grid was collected. Physical property measurements were obtained through image analysis.

[0072] Particle size characteristics are calculated as follows: Particle equivalent diameter = 2 × √(Particle projected area / π); Apparent density characteristics are reflected by the gray-scale statistical properties of the particle region.

[0073] Motion state parameters are calculated using the rate of change of particle positions across consecutive image frames. Let the particle position in the nth frame be (x... n y n If the time interval is Δt, then the instantaneous velocity is calculated as: Instantaneous velocity = √[(x n -x n-1 ²+(y n -yn-1 )²] / Δt; Acceleration is calculated as: Acceleration = (Current velocity - Previous velocity) / Δt.

[0074] For cases where a mesh contains multiple particles, the mesh is considered as an "edge particle unit," and its physical properties and motion state parameters are taken as the statistical average of the particles in the mesh: average particle unit size = (1 / N) × Σ (particle size); apparent particle unit density = (1 / N) × Σ (particle gray value); average particle unit velocity = (1 / N) × Σ (particle velocity); where N is the number of particles in the mesh.

[0075] In step S102, based on the acquired physical property measurements and motion state parameters, it is determined whether the divergence behavior of the edge particles deviates from the preset target.

[0076] Optionally, the current position of the particle or particle unit can be compared with a preset diversion boundary. The preset boundary is set according to the sorting target and represents the boundary of the collection channel into which the particle should enter.

[0077] The judgment logic employs a spatial position comparison method: when the position of a particle or particle unit exceeds a preset boundary by a certain range and its movement trend continues to deviate, it is judged as a deviation in flow behavior. The judgment process considers the current motion state of the particle to avoid misjudgment caused by instantaneous disturbances.

[0078] For particle units, when the number of particles within the grid is ≥3 and the standard deviation of their size is ≤0.25 × average size, the grid center position is used as the criterion: if the grid center position is >+5mm, it is determined to be deviated from the clean coal channel; if the grid center position is <-5mm, it is determined to be deviated from the gangue channel; otherwise, it is determined to be not deviated. Through simple and effective spatial position comparison, edge particles requiring intervention can be quickly identified, ensuring timely response.

[0079] In step S103, when it is determined that the deviation is from the preset target, a micro-correction instruction for the edge particles is generated.

[0080] Optionally, the required intervention parameters are calculated based on the degree of deviation and motion state of the particles or particle units. The intervention direction is perpendicular to the particle motion direction and points towards the target collection channel; the area of ​​action is centered on the particle location or the grid center; the intensity of the action is proportional to the deviation distance.

[0081] The intensity of action is calculated as follows: Intensity of action = proportionality coefficient × |current position - target boundary position|; where the proportionality coefficient is preset according to particle characteristics and environmental conditions.

[0082] For particle units, the action area covers the entire grid (10mm × 10mm), and the action intensity takes into account the number of particles within the grid: Action Intensity = 1.5N / mm × |Grid Center Position - Target Boundary| × (Number of Particles in the Grid / 5); The deviation judgment result is converted into specific physical intervention commands, providing clear operating parameters for the actuator. The command generation process considers the response characteristics of the actuator to ensure that the command is executable.

[0083] In step S104, based on the generated micro-correction instructions, the actuator is driven to physically intervene in the edge particles.

[0084] Optionally, the actuator employs a miniature airflow jet device, installed on both sides of the diversion area. The coverage and intensity of the airflow jet are controlled and adjusted according to the intervention direction, area of ​​action, and intensity specified in the command. The airflow jet is completed in a short time, ensuring that it affects the target particle without interfering with other areas. For particle units, a grid coverage strategy is adopted: area of ​​action = corresponding grid area (10mm × 10mm); action time = 15ms (ensuring coverage of the entire particle passage process); airflow pressure = value corresponding to the intensity of action.

[0085] The trajectory deviation of a particle after being subjected to force is calculated as: Trajectory deviation = (Intensity of action × Time of action²) / (2 × Equivalent mass of particle). By reasonably controlling the intensity and time of action, sufficient deviation of the particle is achieved before it passes through the critical region. Through precisely controlled physical action, the deviated particles are guided to the correct channel, thereby improving sorting accuracy. The micro airflow jet device has a fast response speed and can intervene before the particle passes through the critical region.

[0086] When particles are subjected to a vertical force, their trajectory will deviate. By dividing the diversion area into a 10mm × 10mm grid and implementing regional intervention, the engineering implementation difficulty is significantly reduced while maintaining high sorting accuracy. Compared with traditional methods, this embodiment, through gridded area determination and intervention strategies, relaxes the positioning accuracy requirement of the actuator from ±0.5mm to ±10mm and the response time requirement from ≤5ms to ≤20ms, allowing existing industrial-grade airflow nozzles to meet the requirements and significantly improving engineering feasibility.

[0087] In one exemplary embodiment, this method is implemented based on Huawei's Mining Hong Industrial Internet Operating System. Mining Hong is a dedicated industrial internet operating system developed by Huawei for the mining industry. It achieves unified interconnection and collaborative control of underground equipment through distributed soft bus technology, providing deterministic latency guarantees, distributed data management, and edge computing capabilities. This invention fully utilizes the platform characteristics of Mining Hong, deploying core functional modules such as edge particle physical property measurement, motion state parameter acquisition, flow deviation judgment, and micro-correction command generation as Mining Hong service components. Industrial vision sensors and millimeter-wave radars are accessed through Mining Hong's unified equipment management framework, utilizing their distributed data services to achieve millisecond-level synchronous acquisition and fusion processing of multi-source heterogeneous data. Simultaneously, leveraging Mining Hong's real-time task scheduling mechanism, it ensures that micro-correction commands are transmitted and executed end-to-end before particles pass through the critical region, with control latency stably controlled within 15ms. This implementation based on Mining Hong effectively solves the technical challenge of insufficient edge particle processing capabilities in complex underground mining environments using traditional sorting systems.

[0088] It is understandable that the surface condition and independence of edge particles significantly affect the accuracy of physical property measurements. Related technologies typically use raw sensor measurements directly for sorting decisions; however, issues such as dust adhering to particle surfaces and particle adhesion in underground mining environments can distort measurements, leading to increased misjudgment rates. This application provides a physical property measurement correction method that improves the measurement accuracy of key physical properties by considering both particle surface condition and particle independence. The physical property measurement method provided in this application will be described in conjunction with the steps shown in this embodiment. Figure 2 As shown:

[0089] In step S201, the original measurements of the intrinsic material properties of the edge particles are obtained.

[0090] Optionally, image data of edge particles can be captured using industrial vision, and raw measurements can be extracted using image processing algorithms. The raw measurements include particle size parameters and apparent density parameters.

[0091] Size parameters are obtained by calculating the particle's projected area: Particle equivalent diameter = 2 × √(particle projected area / π);

[0092] The apparent density parameter is obtained by analyzing the gray-level distribution of the particle region: average gray-level value = (1 / N) × ΣI(x,y); where I(x,y) is the gray-level value of each pixel in the particle region, and N is the total number of pixels in the particle region.

[0093] In step S202, the surface state of the edge particles is determined visually.

[0094] Optionally, edge detection and texture analysis can be performed on the particle images to assess the surface condition. Surface condition assessment includes the following aspects:

[0095] Surface cleanliness: determined by calculating the sharpness index of particle edges; for example, surface cleanliness is calculated as: surface cleanliness = average gradient magnitude of particle edge pixels / preset threshold; when the surface cleanliness is less than 0.7, it is judged that the surface is heavily dusty.

[0096] Surface roughness: determined by analyzing the texture characteristics of the particle surface; for example, surface roughness is calculated as: surface roughness = (particle edge length / 2πr) - 1; where r is the equivalent radius of the particle. A surface roughness greater than 0.15 is considered rough.

[0097] Attachment status: determined by detecting irregular protrusions at the particle edges; for example, attachment detection is achieved by calculating the rate of change of curvature at the particle edges; when the rate of change of curvature exceeds a threshold, it is determined that attachments exist; attachment area ratio = number of pixels in the attachment area / total number of pixels at the particle edges.

[0098] In step S203, when the edge particle is a single particle, the particle independence verification result of the edge particle is determined by the vision system, and when the edge particle is a particle unit, the consistency index of the particle unit is determined.

[0099] Optionally, the particle shape characteristics and surrounding environment can be analyzed to determine the particle's independence status.

[0100] The particle independence state includes two cases:

[0101] (1) Whether the particles stick together (for single-particle scenarios);

[0102] (2) Consistency of physical properties of particles within a particle unit (for particle unit scenarios);

[0103] The particle adhesion determination is achieved through the following steps: Calculate the roundness index of the particle profile: Roundness = 4π × (particle area) / (particle perimeter)²; When the roundness is less than 0.8, it is determined that the particle may be adhered to other particles, and the minimum proximity distance is calculated: min{√[(x i -x j )²+(y i -y j When the minimum proximity distance is less than 0.8 times the average particle diameter, particle adhesion is confirmed.

[0104] The consistency assessment of the physical properties of particle units is achieved through the following steps: dividing the flow distribution region into a 10mm × 10mm grid; counting the number of particles in the grid (Ngrid); calculating the standard deviation of particle size (σD) and the standard deviation of grayscale (σG); consistency index = 1 - (σD / D) avg +σG / G avg The consistency index is ≥0.75, indicating that the physical properties are consistent. Through multi-dimensional analysis, the independence status of particles is accurately determined, providing a key basis for subsequent correction. The independence status assessment considers common particle adhesion problems and particle unit characteristics in the underground environment.

[0105] In step S204, based on the particle surface state and the particle independence verification results or particle unit consistency index, the original measured values ​​of the intrinsic properties of the material are corrected to obtain the key physical property measured values ​​of the edge particles.

[0106] Optionally, the original measurements can be adjusted using different correction coefficients depending on the surface condition and independence status.

[0107] The formula for correcting the dimensional parameters is: Corrected dimension = Original dimension × (1 + α × Surface cleanliness deviation + β × Independence factor); Where: Surface cleanliness deviation = 1 - Surface cleanliness; Independence factor = 0 (single particle and no adhesion) or 0.15 (particle adhesion) or 0.05 × (1 - Consistency index) (particle unit); α = 0.2, β = 0.1 (empirical coefficient);

[0108] The formula for correcting the apparent density parameter is: Corrected density = Original density × (1 + γ × Surface roughness + δ × Independence factor); where: γ = 0.3, δ = 0.2 (empirical coefficients);

[0109] The above correction process considers the influence of surface condition and independence on the measured values, balancing the weights of different factors through empirical coefficients. The correction coefficients are determined based on historical data statistics to ensure the reliability of the correction results. This correction method effectively eliminates the influence of surface condition and particle adhesion on the measured values, improving the accuracy of key physical properties. The corrected measured values ​​better reflect the true physical characteristics of the particles.

[0110] It is understandable that accurately acquiring the motion state parameters of edge particles is crucial for determining the separation behavior during coal sorting. Related technologies typically rely on a single sensor to acquire motion information; however, in the dusty environment of underground mines, the data reliability of a single sensor is low, leading to inaccurate calculation of motion state parameters. This application provides a multi-sensor fusion method for acquiring motion state parameters, improving the measurement accuracy of motion state parameters through the collaborative work of a visual sensor and millimeter-wave radar. The method for acquiring motion state parameters provided in this application will be described in conjunction with the steps shown in this embodiment. Figure 3 As shown:

[0111] In step S301, the motion information of edge particles collected by the visual sensor and the motion information of edge particles collected synchronously by the millimeter-wave radar are fused to obtain multi-sensor fusion information.

[0112] Optionally, the vision sensor and millimeter-wave radar synchronously acquire data with the same timestamp, at a sampling frequency of 200Hz. The vision sensor uses an industrial camera, and the millimeter-wave radar uses a 77GHz band device.

[0113] Data fusion employs the Kalman filter algorithm, with the following specific steps:

[0114] The visual sensor provides particle location information (x v y v ) and size information; millimeter-wave radar provides particle velocity information (v r ) and distance information; construct the state vector X=[x, y, v x v y ]ᵀ;Visual measurement vector Z v =[x v y v ]ᵀ;Radar measurement vector Z r =[v r ·cosθ,v r ·sinθ]ᵀ (θ is the radar beam angle);

[0115] Calculate the Kalman gain K = P·H T ·(H·P·H) T +R) -1 Update the state estimate X̂ = X̂ + K·(ZH·X̂); During the fusion process, dynamically adjust the measurement noise covariance matrix R based on sensor reliability:

[0116] When the dust concentration is >15mg / m³, increase the R-value of the vision sensor (reduce its weight); when the particle size is <5mm, increase the R-value of the radar sensor (reduce its weight); under standard conditions, the weight ratio of vision to radar is 6:4. This step, through multi-sensor data fusion, effectively overcomes the limitations of a single sensor in the underground mining environment and improves the reliability of position and velocity measurements. The noise of the fused data is reduced by 40-60% compared to that of a single sensor.

[0117] In step S302, a particle identity prediction and tracking algorithm is used to analyze a single particle, or a grid statistical method is used to analyze particle units to obtain motion state characteristics.

[0118] First, determine whether the current scene is a single particle or a particle unit. The criteria are as follows: when the particle spacing is greater than 1.5 × the average particle diameter, it is determined to be a single particle scene; when the particle spacing is less than or equal to 1.5 × the average particle diameter and the number of particles is greater than or equal to 3, it is determined to be a particle unit scene.

[0119] In single-particle scenarios, an improved particle identity prediction and tracking algorithm is employed: A particle motion model is established: x K =F·x K-1 +w K Where F is the state transition matrix, w K To address process noise, the Hungarian algorithm is used to solve the data association problem; an adaptive threshold is used to handle occlusion: when the occlusion time is less than 3 frames, the motion model is used to predict the position; when the occlusion time is greater than or equal to 3 frames, the tracking is reinitialized.

[0120] In the particle unit scenario, a grid statistical method is used: the flow distribution region is divided into a 10mm × 10mm grid; the number and location distribution of particles within each grid are statistically analyzed; and the motion vector at the grid center is calculated: v grid =(1 / N)·Σv i +α·▽ρ; where N is the number of particles in the grid, and ▽ρ is the particle density gradient; the grid motion direction is determined by principal component analysis (PCA); this step uses a scene-adaptive tracking strategy to ensure accurate acquisition of motion state features under different particle density conditions. The algorithm's automatic switching mechanism has a delay of <1ms, which does not affect real-time performance.

[0121] In step S303, motion state parameters of edge particles are extracted from motion state features.

[0122] Optionally, relevant motion state parameters can be extracted based on the scene type.

[0123] In a single-particle scenario, the extracted parameters include:

[0124] Instantaneous velocity: v = √(v x ²+vy ²);

[0125] Direction of motion: θ = arctan2(v) y v x );

[0126] Acceleration: a = (v K -v K-1 ) / Δt;

[0127] Trajectory curvature: κ=|v x ·a y -v y ·a x | / v³;

[0128] In the particle unit scenario, the extracted parameters include:

[0129] Unit center velocity: v c =(1 / N)·Σv i ;

[0130] Unit diffusion rate: σv = √[(1 / N)·Σ(v)] i -v c )²];

[0131] Unit motion consistency: η = 1 - (σv / v) c );

[0132] Unit trajectory offset trend: d = (x c -x target );

[0133] After all parameters are calculated, validity verification is performed: velocity range verification: 0.5m / s≤v≤5.0m / s; acceleration range verification: -2.0m / s²≤a≤2.0m / s²; invalid parameters are marked as "NaN" and will not be included in subsequent processing; the motion state parameters output in this step are directly used for flow diversion behavior determination. The parameter extraction process takes into account the special characteristics of the underground mining environment to ensure the reliability and applicability of the output parameters.

[0134] In this embodiment, the complementary characteristics of vision and millimeter-wave radar are used to overcome the interference of dust environment on a single sensor; and the scene-adaptive tracking strategy takes into account the motion feature extraction requirements of single particles and particle units.

[0135] In one exemplary embodiment, a method for determining deviation in sorting behavior is provided. This method addresses the high false positive rate caused by traditional sorting methods that rely solely on position determination. It achieves a comprehensive determination by combining particle characteristics and motion state, such as... Figure 4 As shown:

[0136] In step S401, particle classification features and current motion state features are generated based on physical property measurements and motion state parameters.

[0137] Optionally, using physical property measurements and motion state parameters, the following characteristics can be calculated:

[0138] Particle classification features include:

[0139] Density characteristics: ρ = (GG) min ) / (G max -G min ), where G is the corrected grayscale value;

[0140] Dimensional characteristics: d = D / D max D is the corrected size;

[0141] Shape characteristics: s = 4πA / P², where A is the area and P is the perimeter.

[0142] Current motion state characteristics include:

[0143] Location feature: x = (x pos -x min ) / (x max -xmin), normalized position;

[0144] Velocity characteristic: v = |v| / v max Normalization speed magnitude;

[0145] Directional characteristics: θ = arctan2(v y v x ).

[0146] In step S402, the desired collection channel for edge particles is determined based on particle classification characteristics.

[0147] Optionally, a simple threshold classification method can be used to determine the desired collection channel: when the density feature ρ>0.65, it is determined to be clean coal particles and is expected to enter the clean coal channel; when the density feature ρ≤0.65, it is determined to be gangue particles and is expected to enter the gangue channel.

[0148] For larger particles (d>0.6), shape characteristics are also considered: if s>0.85, the density characteristic is still used for determination; if s≤0.85 (indicating that they may be agglomerated particles), the density threshold is reduced to 0.60; for example, the above particle density characteristic ρ=0.79>0.65 and size characteristic d=0.49<0.6, therefore the expected collection channel is determined to be a clean coal channel.

[0149] In step S403, a target boundary set corresponding to the desired collection channel is selected from a preset set of ideal space-motion state boundaries.

[0150] Optionally, two sets of boundary sets are preset in the ideal space-motion state boundary set, corresponding to the clean coal channel and the gangue channel respectively: the target boundary set of the clean coal channel is defined as:

[0151] Location boundary: x < 0.75 (i.e., x pos <+5mm);

[0152] Velocity direction boundary: |θ-θ target |<30°, where θ target This is the direction of the clean coal passage.

[0153] The target boundary set of the gangue channel is defined as follows:

[0154] Location boundary: x > 0.25 (i.e., x pos >-5mm);

[0155] Velocity direction boundary: |θ-θ target |<30°, where θ target This indicates the direction of the waste rock passage.

[0156] The boundary set is stored in the embedded system in the form of a lookup table. Based on the desired channel determined in step S402, the corresponding boundary set is selected. For example, for particles expected to enter the clean coal channel, the clean coal channel boundary set is selected for subsequent comparison.

[0157] In step S404, the current motion state features are compared with the target boundary set to obtain the comparison result.

[0158] Optionally, each boundary condition can be checked and the degree of deviation calculated:

[0159] Position deviation calculation: Δx=xx bound , where x bound These are boundary values ​​(0.75 for the clean coal channel and 0.25 for the gangue channel).

[0160] Directional deviation calculation: Δθ = |θ - θ target |, where θ target The target direction of the channel; the comparison results are represented in the form of a binary tuple: (position comparison result, direction comparison result).

[0161] Location comparison: If Δx≤0 (coal channel) or Δx≥0 (gangue channel), the location is correct; otherwise, it is not correct.

[0162] Directional comparison: If Δθ < 30°, then the directions match;

[0163] Otherwise, it does not meet the requirements.

[0164] For example, if the particle's positional characteristic x = 0.81 > 0.75 (clean coal channel boundary), the positional comparison result is inconsistent; however, if its direction of movement θ deviates from the direction of the clean coal channel by 25°, then the directional comparison result is consistent. The comparison result is (inconsistent, consistent).

[0165] In step S405, based on the comparison results, it is determined whether the divergence behavior of the edge particles deviates from the preset target.

[0166] Optionally, a weighted decision-making method can be used for the final determination:

[0167] Deviation calculation: deviation = w x ·|Δx|+w θ ·|Δθ| / 30°; where w x =0.7, w θ =0.3 is the weighting coefficient;

[0168] When deviation > 0.5, it is determined to deviate from the preset target; otherwise, it is determined not to deviate.

[0169] For comparison results of (disagree / agree), the persistence of the deviation must also be checked: if the comparison results of two consecutive frames are both (disagree / agree), then it is determined to be a deviation. For example, if the particle position deviation is Δx = 0.81 - 0.75 = 0.06 and the direction deviation is Δθ = 25°, then: deviation = 0.7 × 0.06 + 0.3 × (25 / 30) = 0.042 + 0.25 = 0.292 < 0.5. If there is only a deviation in this one frame, then it is determined to be no deviation; if there is a deviation in two consecutive frames, then it is determined to be a deviation.

[0170] In one exemplary embodiment, a method for dynamically adjusting the deviation threshold is provided. This method addresses the problem of poor adaptability of a fixed threshold under different operating conditions by adjusting the judgment threshold in real time, thereby further reducing the false judgment rate. Figure 5 As shown:

[0171] In step S501, the environmental fluid disturbance parameters of the diversion area and the response characteristics of the actuator are obtained.

[0172] Optionally, two sets of sensors can be installed to monitor the state of the ambient fluid:

[0173] Airflow monitoring: A miniature thermal anemometer is installed 5cm above the diversion area, with a sampling frequency of 100Hz, to measure the airflow velocity in the x, y, and z directions; measured data: airflow velocity <0.5m / s under normal operating conditions, and can reach 1.2m / s at the moment of equipment startup;

[0174] Dust concentration monitoring: A laser scattering dust sensor is used, with a measurement range of 0-1000μg / m³; Actual measured data: Dust concentration is 15-25mg / m³ during normal operation, and can reach 40mg / m³ after equipment maintenance.

[0175] The actuator response characteristics are obtained through factory testing and field calibration:

[0176] Response time: The time from the issuance of the command to the establishment of airflow was measured using a high-speed camera, and the actual measurement was 8.2 ± 0.5 ms;

[0177] Response curve: obtained through step response test, approximately first order: G(s) = 1 / (0.006s+1);

[0178] Specifically, environmental parameters are collected every 100ms, and historical data from the most recent 10 seconds is stored for trend analysis. Field tests revealed a close correlation between airflow disturbances and equipment vibration, so vibration sensors were added as supplementary monitoring.

[0179] In step S502, the deviation threshold for determining whether edge particles deviate from the desired trajectory is dynamically adjusted based on the perturbation parameters and response characteristics.

[0180] Optionally, design a threshold adjustment model:

[0181] base threshold threshold =0.5;

[0182] Disturbance factor: factor =0.3×(v air / 1.0)²+0.2×(dust / 30); where, v air is the airflow velocity (m / s), dust is the dust concentration (mg / m³).

[0183] Response impact factor: response factor =0.15×(responsetime / 8.0-1); where responsetime is the measured response time (ms);

[0184] dynamic threshold threshold :

[0185] dynamic threshold= base threshold ×(1+disturbance factor +response factor ).

[0186] In step S503, if the comparison result exceeds the deviation threshold, it is determined that the flow behavior of the edge particles deviates from the preset target.

[0187] Optionally, the deviation result can be calculated and compared with a dynamic threshold:

[0188] If deviation > threshold final

[0189] It was determined to have deviated from the preset target;

[0190] else

[0191] It was determined that there was no deviation.

[0192] end.

[0193] For example, if the particle deviation is 0.292, and the current dynamic threshold is... final =0.6837, then 0.292 < 0.6837, which is considered as no deviation. In addition, a persistence check is added: only when two consecutive frames deviation > threshold... final Only when this is done can a deviation be definitively identified, thus avoiding misjudgments caused by momentary disturbances.

[0194] In one exemplary embodiment, a method for generating microscopic correction instructions is provided. This method addresses the problem of inaccurate intervention caused by simple threshold determination in traditional sorting. By quantifying physical deviations and generating targeted instructions, the effectiveness of the intervention is improved. Figure 6 As shown:

[0195] In step S601, the physical deviation vector between the edge particles and the target boundary set is quantized based on the comparison results.

[0196] Optionally, based on the comparison results, a two-dimensional deviation vector can be constructed:

[0197] Position deviation calculation: Δx = x pos -x bound Where xpos is the current position of the particle (mm), x bound The boundary position is +5mm for the clean coal channel and -5mm for the gangue channel.

[0198] Directional deviation calculation: Δθ = θ - θ target Where θ is the angle (°) of the particle's direction of motion. target The target direction angle of the channel (0° for clean coal channels and 180° for gangue channels);

[0199] The physical deviation vector is defined as: deviationvector = (w x ·Δx, w θ·Δθ); where, w x =1.0mm -1 w θ =0.03° -1 These are the normalized weighting coefficients.

[0200] In step S602, micro-correction instructions for edge particles are generated based on the physical deviation vector.

[0201] Optionally, the deviation vector can be directly mapped to three key parameters of the micro-correction command:

[0202] Intervention direction: 90° + φ (perpendicular to the deviation direction, pointing towards the target channel); Point of application: current position of the particle (x pos y pos For particle elements, use the grid center position;

[0203] Force intensity: F = k × ||deviation vector| |; where k = 1.2N / unit (determined through on-site debugging).

[0204] It is understandable that in the coal sorting process, fixed mapping rules are difficult to adapt to the particle response characteristics under different working conditions, resulting in unstable intervention effects. This application provides a physical correction response mapping rule and an adaptive adjustment method, which continuously optimizes the intervention effect through a closed-loop feedback mechanism. First, based on the physical deviation vector, a preset physical correction response mapping rule is used to generate microscopic correction instructions for edge particles.

[0205] Optionally, based on the physical deviation vector δ, a piecewise linear mapping rule is used to generate the force intensity of the micro-correction command:

[0206] When |δ|≤1.0, F=k1×|δ|;

[0207] When 1.0 < |δ| ≤ 2.0, F = k2 × |δ| + b2;

[0208] When |δ|>2.0, F=k3×|δ|+b3;

[0209] Where, δ=(w x ·Δx, w y ·Δθ) is the physical deviation vector, |δ| is its magnitude, and the initial parameters are set as follows: k1=1.0N / unit, k2=0.8N / unit, b2=0.2N, k3=0.5N / unit, b3=0.8N.

[0210] The direction of intervention is calculated as: 90° + arctan2 (δ) x δ y );

[0211] The point of application is determined as: the current position of the particle (x) pos y pos For particle elements, this is the center of the grid.

[0212] For the particle unit scenario, the force intensity is further adjusted to: F adj =F×(1+0.15×(1-c);where, c is the consistency index of physical properties of particle units, calculated as: c=1-(σD / D avg +σG / G avg ) / 2; σD and σG are the standard deviations of particle size and gray level within the particle unit, respectively.

[0213] This step quantifies physical deviations into specific intervention parameters, providing clear operational instructions to the actuators. The piecewise linear mapping rule considers the nonlinear characteristics of the system, avoiding over-intervention for small deviations and under-intervention for large deviations.

[0214] Secondly, such as Figure 7 As shown, in step S701, the motion state of the edge particles after applying physical intervention is obtained.

[0215] Optionally, within 5-15 ms after the intervention, 3-5 frames of images are captured using a high-speed camera (500 fps) to recalculate the particle position, velocity, and orientation.

[0216] Data acquisition is divided into three time windows: early window: 5-8ms after intervention, used to detect the initial response; critical window: 10-12ms after intervention, when particles approach the critical line; and late window: 13-15ms after intervention, when particles cross the critical line.

[0217] The motion state parameters are calculated as follows:

[0218] x actual =(1 / N)×Σx i ;

[0219] θ actual =atan2((1 / N)×Σv yi (1 / N)×Σv xi ); where N is the number of valid frames within the window, x i For each frame position, v xi and v yi These are the velocity components for each frame.

[0220] This step uses high temporal resolution monitoring to accurately obtain the particle motion state after intervention, providing reliable data for subsequent error calculation.

[0221] In step S702, the motion state of the edge particles after physical intervention is compared with the expected correction effect corresponding to the micro-correction command to determine the correction error.

[0222] Optionally, the desired correction effect can be calculated based on a particle dynamics model:

[0223] Expected location: x exp =x ini -0.8×(F×t²) / (2×m;

[0224] Desired direction: θ exp =θ ini -0.6×(F×t×r) / (I); where t=10ms is the typical action time, m is the particle mass, r is the lever arm length, I is the moment of inertia, and 0.8 and 0.6 are empirical coefficients.

[0225] The correction error is calculated as follows:

[0226] e p =|x actual -x exp |;e d =|θ actual -θ exp |;

[0227] e t =w p ×e p +w d ×e d Among them, w p =0.7, w d =0.3 is the weighting coefficient.

[0228] In step S703, the physical correction response mapping rule is adaptively adjusted based on the correction error.

[0229] Optionally, the mapping rule parameters can be updated using a proportional adjustment algorithm:

[0230] For each parameter p (e.g., k1, k2, etc.):

[0231] p new =p old ×(1-α×e t ); where α=0.08 is the learning rate.

[0232] It is understandable that in the coal sorting process, when multiple edge particles require intervention simultaneously, a simple, independent intervention strategy can lead to overload of the actuator or interference between particles. This application provides a multi-objective optimization method for generating microscopic correction instructions, achieving globally optimal intervention by comprehensively considering system constraints and inter-particle interactions. Referring to the method flow of this embodiment, the method includes the following steps: Figure 8 As shown:

[0233] In step S801, the energy consumption limit of the actuator, the overall processing capacity of the sorting equipment, and the interaction effects between multiple edge particles are obtained.

[0234] Optional, monitor the following key parameters in real time: Energy consumption limit E lim =180W (maximum power of actuator 200W minus safety margin 20W);

[0235] Overall processing power C tot = System maximum processing capacity 500 particles / second minus the number of currently tracked particles;

[0236] Effect of I ij =exp(-d ij / 15)×(1-c) ij ); where d ij Let c be the distance (mm) between particles i and j. ij These parameters are indicators of particle unit consistency. They are updated every 10ms to ensure they reflect the current operating conditions.

[0237] In step S802, based on energy consumption constraints, overall processing capacity, interaction effects, and physical deviation vectors, a multi-objective optimization algorithm is used to generate micro-correction instructions for edge particles.

[0238] Optional, modeling optimization problem: Objective: Minimize overall intervention error, system energy consumption, and inter-particle interference;

[0239] The constraints include:

[0240] (1) Total energy consumption not exceeding 180W;

[0241] (2) The number of intervention particles does not exceed the system's processing capacity;

[0242] (3) The intervention force is in the range of 0.5N-5.0N;

[0243] (4) The difference in interference force between adjacent particles does not exceed 3.0 N × I ij .

[0244] To adapt to embedded systems, a simplified optimization algorithm is adopted: the particles requiring intervention are arranged in descending order of their physical deviation vector magnitude |δ|; starting with the particle with the largest deviation, the intervention force F is allocated. i =k·|δ i |; Check if the constraints are met; if not, reduce F proportionally. i Repeat the above steps until all particles are distributed or the energy consumption limit is reached.

[0245] Then, the optimized micro-correction instruction sequence is output, specifically: the optimization result is converted into a control signal recognizable by the actuator; intervention timing arrangement: sorted by intervention force magnitude, with an interval of at least 3ms between adjacent interventions; control signal: PWM signal duty cycle = (F i (5.0) × 90%.

[0246] It is understandable that during the coal sorting process, the performance of the actuator (micro-airflow nozzle) changes over time, resulting in different effects from the same intervention command. This embodiment provides a simple and effective method for performance monitoring and adjustment, such as... Figure 9 As shown:

[0247] In step S901, the operating performance data of the actuator is obtained.

[0248] Optionally, each time an intervention is performed, two key data points are automatically recorded: the actual airflow pressure value (measured by a pressure sensor); and the actual duration of action (the time from when the command is issued to when the airflow stops). For example, when the command "1.5N intervention force" is issued, the pressure sensor records the actual pressure as 45 kPa, and the timer records the duration of action as 14.2 ms.

[0249] In step S902, the running performance data is compared with the expected performance corresponding to the micro-correction command.

[0250] Optionally, a built-in actuator calibration data sheet allows users to find the expected airflow pressure and duration based on the magnitude of the intervention force. After each intervention, the percentage deviation between the actual and expected values ​​is automatically calculated. For example, if the instruction is 1.5N, the expected pressure should be 48kPa, but the actual pressure is only 45kPa, then the pressure deviation is 6.25%.

[0251] In step S903, when a performance deviation exists, the parameters of the micro-correction command are adaptively adjusted.

[0252] Optionally, if the deviation exceeds 5%, subsequent instructions will be automatically adjusted.

[0253] If the actual effect is weak (e.g., the pressure is too low), then increase the intervention force of subsequent instructions.

[0254] If the actual effect is too strong (e.g., the pressure is too high), then reduce the intervention force of subsequent instructions.

[0255] The adjustment range is determined based on the magnitude of the deviation, but a single adjustment shall not exceed 10%. For example, when the pressure deviation is 8%, the intervention force of subsequent commands will be increased by approximately 6%.

[0256] To prevent frequent adjustments, the system is set to adjust a maximum of once every 10 interventions, with an interval of at least 5 minutes between adjustments.

[0257] In step S904, the adjusted micro-correction command is sent to the control interface of the actuator.

[0258] Optionally, the adjusted instructions, including the adjusted intervention force parameters, are sent to the actuator controller via a standard industrial bus.

[0259] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0260] Based on the same inventive concept, this application also provides an intelligent control system for underground coal sorting based on a mining system, used to implement the aforementioned intelligent control method for underground coal sorting based on a mining system. The solution provided by this device is similar to the implementation scheme described in the above method. Therefore, the specific limitations of one or more embodiments of an intelligent control system for underground coal sorting based on a mining system provided below can be found in the limitations of the intelligent control method for underground coal sorting based on a mining system described above, and will not be repeated here.

[0261] In one exemplary embodiment, such as Figure 10 As shown, an intelligent control system for coal sorting in underground mines based on a mining system is provided. The system includes:

[0262] The parameter acquisition module 11 is used to acquire the physical property measurement values ​​and motion state parameters of edge particles in the sorting device; edge particles are particles whose spatial location is within the diversion area and whose current motion trajectory tends to or has entered the preset critical area between different collection channels;

[0263] The deviation judgment module 12 is used to determine whether the divergence behavior of edge particles deviates from the preset target based on the physical property measurement value and motion state parameters.

[0264] The instruction generation module 13 is used to generate micro-correction instructions for edge particles when it is determined that the particle deviates from the preset target; wherein, the micro-correction instructions include the physical intervention direction, the point of application, and the intensity of the force.

[0265] The intervention execution module 14 is used to drive the actuator to physically intervene on the edge particles based on micro-correction instructions.

[0266] This embodiment is implemented in the Huawei Mining Hong Industrial Internet Operating System environment, utilizing the distributed soft bus technology of the Mining Hong system to construct a sensor network and control command transmission channel. The parameter acquisition module 11, deviation judgment module 12, command generation module 13, and intervention execution module 14 are deployed as service components of the Mining Hong system, achieving standardized access to multi-source heterogeneous data and millisecond-level task scheduling through the Mining Hong API. The deterministic latency guarantee mechanism of the Mining Hong system ensures that the entire process control latency from edge particle detection to physical intervention is stably controlled within 15ms, providing a reliable system foundation for the micro-correction control of this invention.

[0267] The modules in the aforementioned intelligent control system for coal sorting in underground mines based on the Kuanghong system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of the computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0268] In one exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described intelligent control method for coal sorting in underground mines based on a mining system.

[0269] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the above-described intelligent control method for coal sorting in underground mines based on a mining system.

[0270] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the above-described intelligent control method for coal sorting in underground mines based on a mining system.

[0271] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0272] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0273] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for intelligent control of coal sorting in underground mines based on a mining system, characterized in that, The method includes: The physical property measurements and motion state parameters of edge particles in the sorting device are obtained; the edge particles are those whose spatial location is within the diversion area and whose current motion trajectory tends to or has entered the preset critical area between different collection channels. Based on the measured physical properties and the motion state parameters, particle classification features and current motion state features are generated. Based on the particle classification characteristics, the desired collection channel for the edge particles is determined; Select the target boundary set corresponding to the desired collection channel from the preset ideal space-motion state boundary set; The current motion state features are compared with the target boundary set to obtain the comparison result; Obtain the environmental fluid disturbance parameters and the response characteristics of the actuator in the diversion region; Based on the perturbation parameters and the response characteristics, the deviation threshold for determining whether edge particles deviate from the desired trajectory is dynamically adjusted; If the comparison result exceeds the deviation threshold, it is determined that the flow separation behavior of the edge particles deviates from the preset target; When it is determined that the edge particle deviates from the preset target, the physical deviation vector between the edge particle and the target boundary set is quantized according to the comparison result. Based on the physical deviation vector, a micro-correction instruction is generated for the edge particles; wherein, the micro-correction instruction includes the physical intervention direction, the point of application, and the force intensity; the force intensity is related to the physical deviation vector; Through the distributed soft bus of the mining system, the actuator is driven to physically intervene in the edge particles based on the micro-correction instructions.

2. The method according to claim 1, characterized in that, Obtain physical property measurements of edge particles, including: Obtain the original measurements of the intrinsic material properties of the edge particles; The surface state of the edge particles is determined using a vision system; When the edge particle is a single particle, the particle independence verification result of the edge particle is determined by the vision system; and when the edge particle is a particle unit, the consistency index of the particle unit is determined. Based on the particle surface state, the particle independence verification results, or the particle unit consistency index, the original measured values ​​of the intrinsic properties of the material are corrected to obtain the key physical property measured values ​​of the edge particles.

3. The method according to claim 1, characterized in that, Obtain the motion state parameters of the edge particles, including: The motion information of the edge particles collected by the visual sensor and the motion information of the edge particles collected synchronously by the millimeter-wave radar are fused to obtain multi-sensor fusion information. The motion state characteristics can be obtained by analyzing a single particle using a particle identity prediction and tracking algorithm, or by analyzing particle units using a grid statistical method. The motion state parameters of the edge particles are extracted from the motion state features.

4. The method according to claim 1, characterized in that, After the driving actuator physically intervenes in the edge particles, the method further includes: Obtain the motion state of the edge particles after the physical intervention is applied; The motion state of the edge particles is compared with the expected correction effect corresponding to the micro-correction command to determine the correction error; Based on the correction error, the physical correction response mapping rule is adaptively adjusted.

5. The method according to claim 3, characterized in that, Generate microscopic correction instructions for the edge particles, including: The energy consumption limits of the actuator, the overall processing capacity of the sorting equipment, and the interaction effects between multiple edge particles are obtained. Based on the energy consumption limit, the overall processing capacity, the interaction effects, and the physical deviation vector, a multi-objective optimization algorithm is used to generate microscopic correction instructions for the edge particles.

6. The method according to claim 1, characterized in that, Based on the microscopic correction command, the actuator is driven to physically intervene in the edge particles, including: Obtain operational performance data of the actuator; Compare the operational performance data with the expected performance corresponding to the micro-correction instructions to determine whether there is a performance deviation; When the performance deviation exists, the parameters of the micro-correction command are adaptively adjusted; The adjusted micro-correction command is sent to the control interface of the actuator to drive the actuator to physically intervene in the edge particles.

7. An intelligent control system for coal sorting in mines based on a mining system, characterized in that, The system includes: The parameter acquisition module is used to acquire the physical property measurement values ​​and motion state parameters of edge particles in the sorting device; the edge particles are particles whose spatial location is located in the diversion area and whose current motion trajectory tends to or has entered the preset critical area between different collection channels; The deviation judgment module is used to generate particle classification features and current motion state features based on the measured physical property values ​​and the motion state parameters; Based on the particle classification characteristics, the desired collection channel for the edge particles is determined; Select the target boundary set corresponding to the desired collection channel from the preset ideal space-motion state boundary set; The current motion state features are compared with the target boundary set to obtain the comparison result; Obtain the environmental fluid disturbance parameters and the response characteristics of the actuator in the diversion region; Based on the perturbation parameters and the response characteristics, the deviation threshold for determining whether edge particles deviate from the desired trajectory is dynamically adjusted; If the comparison result exceeds the deviation threshold, it is determined that the flow separation behavior of the edge particles deviates from the preset target; The instruction generation module is used to, when determined to be a deviation from the preset target, quantify the physical deviation vector between the edge particle and the target boundary set according to the comparison result; and generate a micro-correction instruction for the edge particle based on the physical deviation vector; wherein the micro-correction instruction includes a physical intervention direction, a point of application, and an intensity of force; the intensity of force is related to the physical deviation vector; The intervention execution module is used to drive the actuator to physically intervene in the edge particles based on the micro-correction instructions via the distributed soft bus of the mining system.

Images