Coal flow treatment method, medium, equipment and product

Abstract

The embodiment of the invention discloses a coal flow treatment method, medium, equipment and product, and is applied to a coal flow treatment system, and the coal flow treatment system comprises a coal briquette conveying belt for conveying coal flow, a sorting manipulator and a coal briquette crusher; acquiring multi-dimensional information corresponding to the coal flow; dividing the coal flow into sundries and coal briquettes according to the multi-dimensional information, and obtaining pose information corresponding to the sundries and granularity information corresponding to the coal briquettes; according to the pose information corresponding to the sundries, a sorting manipulator is controlled to grab the sundries so as to remove the sundries conveyed on the coal briquette conveying belt; and according to the granularity information corresponding to the coal briquettes, crushing parameters corresponding to the coal briquette crusher are adjusted, and the coal briquette crusher is controlled to crush the coal briquettes conveyed on the coal briquette conveying belt based on the adjusted crushing parameters. According to the embodiment of the invention, on-line impurity removal and on-line coal particle size regulation and control are integrated in the same conveying process, intelligent processing of coal flow is realized, and the data accuracy of quick coal quality detection and the overall production efficiency and stability are improved.

Description

Technical Field

[0001] This application belongs to the field of coal mine testing, specifically relating to a coal flow processing method, storage medium, electronic equipment, and computer program product. Background Technology

[0002] Currently, rapid coal quality testing devices lack the ability to perform real-time and precise sorting of the coal stream during transportation, especially when crushing and testing coal lumps. Impurities mixed in the coal stream cannot be effectively identified and removed in advance. These impurities will enter the subsequent precision sorting and crushing devices along with the coal stream, easily causing equipment damage. This not only affects the continuous and stable operation of the equipment but also leads to serious distortion of the final coal quality test results due to sample contamination. Furthermore, current monitoring of coal lump particle size distribution in the coal stream is usually only used for post-event statistics and cannot provide real-time feedback to the coal crusher. This results in uneven particle size of the coal lumps entering the subsequent precision sorting and crushing devices, which affects the accuracy of rapid coal quality testing data and the stability of production.

[0003] Therefore, knowing how to handle the coal flow on the conveyor belt during transportation is essential for rapid coal quality testing. Summary of the Invention

[0004] The purpose of this application is to provide a coal flow processing method, medium, equipment, and product that can solve the problem of insufficient coal flow processing capability during transportation, which in turn affects the results of rapid coal quality testing.

[0005] In a first aspect, embodiments of this application provide a coal flow processing method applied to a coal flow processing system, the coal flow processing system including a coal block conveyor belt for conveying the coal flow, a sorting robot, and a coal block crusher, the method comprising: Obtain multi-dimensional information corresponding to the coal flow; Based on the multi-dimensional information, the coal flow is divided into debris and coal blocks, and the pose information corresponding to the debris and the particle size information corresponding to the coal blocks are obtained. Based on the position information corresponding to the debris, the sorting robot is controlled to grab the debris in order to remove the debris being transported on the coal conveyor belt; Based on the particle size information of the coal block, the crushing parameters of the coal block crusher are adjusted, and the coal block crusher is controlled to crush the coal block conveyed on the coal block conveyor belt based on the adjusted crushing parameters.

[0006] Optionally, the coal flow processing system includes a camera, an infrared sensor, and a lidar; acquiring multi-dimensional information corresponding to the coal flow includes: The visible light image information captured by the camera is acquired; the visible light image information is used to reflect the visual morphological characteristics of the coal flow. The infrared thermal imaging information collected by the infrared sensor is acquired; the infrared thermal imaging information is used to reflect the temperature distribution of the coal flow. The three-dimensional point cloud information collected by the lidar is acquired; the three-dimensional point cloud information is used to reflect the three-dimensional dimensions of the coal flow. The visible light image information, the infrared thermal imaging information, and the three-dimensional point cloud information are determined as the multi-dimensional information corresponding to the coal flow.

[0007] Optionally, the step of dividing the coal flow into debris and coal blocks based on the multi-dimensional information, and obtaining the pose information corresponding to the debris and the particle size information corresponding to the coal blocks, includes: Based on the morphological differences between debris and coal lumps, and by analyzing the coal flow using the visible light image information, a first classification result is obtained; The second classification result is obtained by analyzing the coal flow based on the temperature difference between the debris and the coal lumps, as well as the infrared thermal imaging information. Based on the first classification result and the second classification result, the coal flow is divided into debris and coal blocks; The position and orientation of the debris on the coal conveyor belt are determined based on the three-dimensional point cloud information, and the position and orientation are determined as the pose information corresponding to the debris. Based on the visible light image information and the three-dimensional point cloud information, the particle size information corresponding to the coal block is determined.

[0008] Optionally, determining the particle size information corresponding to the coal block based on the visible light image information and the three-dimensional point cloud information includes: The visible light image information is segmented at the pixel level to extract the pixel area corresponding to the coal block; The first particle size estimate of the coal block is determined based on the preset calibration coefficient and the pixel area corresponding to the coal block. The three-dimensional point cloud information is segmented and geometrically analyzed to obtain the second particle size estimate corresponding to the coal block. The particle size information corresponding to the coal block is determined based on the first particle size estimate and the second particle size estimate corresponding to the coal block.

[0009] Optionally, controlling the sorting robot to grab the debris based on the pose information corresponding to the debris, so as to remove the debris conveyed on the coal conveyor belt, includes: The grasping time when the debris reaches the preset grasping position and the grasping posture of the sorting robot are determined based on the pose information corresponding to the debris. The sorting robot plans its grasping path based on the grasping time and the preset grasping position; Based on the grasping posture and grasping path of the sorting robot, the sorting robot is controlled to grasp the debris in order to remove the debris conveyed on the coal conveyor belt.

[0010] Optionally, adjusting the crushing parameters of the coal crusher according to the particle size information of the coal block, and controlling the coal crusher to crush the coal blocks conveyed on the coal block conveyor belt based on the adjusted crushing parameters, includes: When the particle size information corresponding to the coal block does not meet the preset particle size requirement, the crushing intensity and the number of crushing times in the crushing parameters are adjusted to obtain the adjusted crushing intensity and the number of crushing times. Based on the adjusted crushing intensity and number of crushing cycles, the coal crusher is controlled to crush the coal blocks conveyed on the coal block conveyor belt.

[0011] Optionally, after adjusting the crushing parameters of the coal block crusher according to the particle size information of the coal block, and controlling the coal block crusher to crush the coal blocks conveyed on the coal block conveyor belt based on the adjusted crushing parameters, the method further includes: After a preset time period, determine whether the crushed particle size information of the coal block after being crushed by the coal block crusher meets the preset particle size requirements; If the crushing particle size information does not meet the preset particle size requirement after N consecutive judgments, an abnormal signal is generated to characterize the abnormality of the coal block crusher; where N is a preset positive integer.

[0012] Secondly, embodiments of this application provide a storage medium that stores computer instructions, which, when executed by a computer, are used to perform the steps of the coal flow processing method as described in the first aspect.

[0013] Thirdly, embodiments of this application provide an electronic device including at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the steps of the coal flow processing method as described in the first aspect.

[0014] Fourthly, embodiments of this application provide a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the coal flow processing method as described in the first aspect.

[0015] In this embodiment, by acquiring multi-dimensional sensing information corresponding to the coal flow and accurately classifying debris and coal lumps accordingly, high-precision and robust online identification of foreign objects in the coal flow can be achieved. By dynamically controlling the sorting robot to grasp and remove debris through real-time positional information of the debris, automatic and precise online removal of debris before it enters the precision reduction and crushing device (coal lump crusher) can be realized. By acquiring the particle size information of coal lumps in real time and dynamically adjusting the crushing parameters accordingly, online and adaptive precision crushing of large coal lumps in the coal flow can be realized. Furthermore, by integrating online removal of debris and online control of coal lump particle size into the same conveying process for continuous execution, uninterrupted intelligent processing of the entire coal flow process can be achieved, thereby improving the data accuracy, overall production efficiency, and stability of rapid coal quality testing. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of a coal flow treatment system provided in an embodiment of this application; Figure 2 This is a flowchart of the steps of a coal flow treatment method provided in an embodiment of this application; Figure 3 This is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. 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.

[0018] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0019] The coal flow treatment method provided in this application will be described in detail below with reference to the accompanying drawings, through specific embodiments and application scenarios.

[0020] Current rapid coal quality testing devices generally lack the ability to perform real-time and accurate sorting of coal streams during transportation, when crushing, reducing, and testing coal blocks. Non-coal impurities such as wood, ironware, and gangue mixed in the coal stream cannot be effectively identified and removed in advance. These impurities will enter the subsequent precision reduction and crushing devices along with the coal stream, easily causing equipment blockage, tool wear, and even structural damage. This not only affects the continuous and stable operation of the equipment but also leads to serious distortion of the final coal quality test results due to sample contamination.

[0021] Furthermore, the particle size distribution of coal lumps on the conveyor belt can usually only be obtained through offline sampling or post-processing analysis. Statistics show that the average lag time for information feedback throughout the entire process is as long as 4 hours. This delayed and discontinuous monitoring method cannot provide real-time and accurate particle size feedback to the upstream crushing process. Therefore, crushers often operate with fixed or empirical parameters and cannot adaptively handle oversized coal lumps. This results in uneven particle size of coal samples entering the sample reduction stage, directly affecting the representativeness of the reduced samples and becoming a major bottleneck for the accuracy of rapid coal quality testing data. Simultaneously, uncrushed large coal lumps can also directly cause blockages in downstream equipment, affecting production continuity.

[0022] The coal flow treatment method provided in this application is applied to a coal flow treatment system, which can be understood as an improved treatment system of an existing rapid coal quality testing device.

[0023] Reference Figure 1 This illustration shows a schematic diagram of a coal flow processing system provided in an embodiment of this application. The coal flow processing system specifically includes a coal conveyor belt 1 for transporting the coal flow, a sorting robot 2, and a coal crusher 3. The coal flow refers to a continuous flow of mixed material consisting of coal and potentially mixed foreign matter (such as gangue, wood blocks, metal fragments, etc.) moving on the coal conveyor belt. The coal conveyor belt 1 refers to a belt conveyor device used to carry and continuously transport the coal flow. The sorting robot 2 refers to an actuator installed on the side or above the coal conveyor belt, capable of multi-degree-of-freedom movement, and equipped with specialized grippers (such as pneumatic grippers, electromagnetic chucks, or adaptive multi-finger grippers) for grasping and removing identified foreign matter. The coal crusher 3 refers to a crushing device installed at key points on the coal conveyor belt (such as above or in front of the drop point), capable of receiving relevant instructions and dynamically adjusting crushing parameters, for online crushing of excessively large coal lumps.

[0024] like Figure 1 As shown, the coal flow processing system also includes a data acquisition module 4, a data processing and identification module 5, a data transmission and storage module 6, a control and feedback module 7, and a human-machine interaction module 8.

[0025] Reference Figure 2This is a flowchart of a coal flow treatment method provided in an embodiment of this application, which specifically includes the following steps: Step 201: Obtain multi-dimensional information corresponding to the coal flow; In this embodiment, a data acquisition module is needed to collect raw data of the coal flow in different physical dimensions, and a data processing and recognition module is needed to preprocess the raw data, such as alignment, denoising, and feature extraction, to obtain multi-dimensional information corresponding to the coal flow. The multi-dimensional information can characterize the attributes of various targets in the coal flow from multiple angles and in all aspects, such as spatial dimension (e.g., three-dimensional geometry), spectral dimension (e.g., visible light texture and infrared thermal radiation), and temporal dimension (motion information of continuous frames).

[0026] Step 202: Divide the coal flow into debris and coal blocks according to the multi-dimensional information, and obtain the pose information corresponding to the debris and the particle size information corresponding to the coal blocks; In this embodiment, the data processing and identification module needs to classify each target in the coal flow into debris or coal lumps based on the fused multi-dimensional information. After classification, it is also necessary to determine the corresponding pose information for the identified debris and the corresponding particle size information for the identified coal lumps. Here, debris refers to any non-coal component that may damage subsequent equipment or detection results; coal lumps refer to coal materials that require particle size assessment.

[0027] It should be noted that in this embodiment of the application, the pose information and particle size information are obtained separately after the target classification (debris / coal block) is completed. This is to avoid performing complex geometric calculations on all targets, thereby improving the processing efficiency of the system and achieving optimized allocation of computing resources.

[0028] In order to achieve intelligent online identification and processing of impurities and particle size in coal flow, and to overcome the limitations of traditional manual and simple machine vision methods, in this embodiment of the application, a dedicated CNN (Convolutional Neural Network) model is constructed and trained based on a deep learning framework (such as TensorFlow, PyTorch, etc.), specifically including impurity identification models and particle size identification models for different tasks.

[0029] These specially designed and trained CNN models can quickly and efficiently process and analyze the multi-dimensional information acquired in real time by the data acquisition module. Among them, the debris recognition model focuses on accurately locating and classifying various debris (such as stones, wood, and metal) from complex backgrounds; the particle size recognition model focuses on accurately segmenting and calculating the particle size distribution of the crushed coal flow image to determine whether it meets the acceptable standards. By deploying the fully trained models in the data processing and recognition modules, the system can achieve real-time identification of debris types during belt conveying and continuous and accurate assessment of coal particle size, providing a highly reliable decision-making basis for subsequent automatic removal and adaptive crushing control.

[0030] Step 203: Based on the position information corresponding to the debris, control the sorting robot to grab the debris to remove the debris being transported on the coal conveyor belt; In this embodiment, after the data processing and identification module determines the pose information corresponding to the debris, it transmits this information to the control and feedback module in real time with low latency through the data transmission and storage module. The control and feedback module generates a grasping control signal for the sorting robot based on this pose information and sends it to the robot's drive controller. The sorting robot then moves precisely and collaboratively based on this instruction, grasping the debris conveyed on the coal conveyor belt at the appropriate time and position, and moving it to the waste material channel, thereby achieving online automatic removal of debris.

[0031] Step 204: Based on the particle size information of the coal block, adjust the crushing parameters of the coal block crusher, and control the coal block crusher to crush the coal blocks conveyed on the coal block conveyor belt based on the adjusted crushing parameters.

[0032] In this embodiment, after the data processing and identification module determines the particle size information corresponding to the coal block, it transmits this information to the control and feedback module in real time through the data transmission and storage module. The control and feedback module generates an adaptive control signal for the coal block crusher based on the particle size information and sends it to the coal block crusher. The coal block crusher adjusts its corresponding crushing parameters in real time based on this signal. At this time, the coal block crusher performs crushing actions with the adjusted crushing parameters that match its particle size, thereby achieving precise and efficient crushing of large pieces of coal conveyed on the coal block conveyor belt and avoiding over-crushing or under-crushing.

[0033] It should be noted that the particle size information of the coal block obtained in step 202 refers to the particle size distribution information of the crushed coal block obtained after crushing by the coal block crusher based on the adjusted crushing parameters, and not the particle size statistics of the original large coal block before crushing by the coal block crusher.

[0034] It is understood that the coal flow processing method provided in this application involves identifying impurities and uncrushed coal pieces in the upstream conveying section before the coal crusher, where the raw coal flow containing impurities and uncrushed coal pieces is transported by a coal conveyor belt. These impurities are then pre-selected and removed by a sorting robot. Subsequently, the impurity-free coal flow is transported to the coal crusher for crushing. Crucially, the system senses and analyzes the particle size information of the crushed coal pieces in real time at a monitoring point downstream of the crusher. Based on this real-time downstream particle size feedback, the system dynamically adjusts the crushing parameters of the upstream coal crusher, thereby achieving adaptive and precise crushing of subsequent coal pieces entering the crusher.

[0035] This application embodiment acquires multi-dimensional sensing information corresponding to the coal flow and accurately distinguishes debris from coal lumps accordingly. This enables high-precision and robust online identification of foreign objects in the coal flow. By dynamically controlling the sorting robot to grasp and remove debris through real-time positional information, it achieves automatic and precise online removal of debris before it enters the coal lump crusher. By acquiring the particle size information of coal lumps in real time and dynamically adjusting the crushing parameters accordingly, it enables online, adaptive, and precise crushing of large coal lumps in the coal flow. Furthermore, by integrating online debris removal and online coal particle size control into the same conveying process for continuous execution, it achieves full-process, uninterrupted intelligent processing of the coal flow, thereby improving the data accuracy, overall production efficiency, and stability of rapid coal quality testing.

[0036] In one embodiment of this application, such as Figure 1 As shown, a sampler and a near-infrared and X-ray fluorescence fusion spectroscopy module can also be used to perform representative sampling and rapid multi-element analysis on the final coal stream (i.e., the coal stream on the coal stream detection belt) after impurity removal and particle size identification, so as to realize the full-process automation and closed-loop quality control of coal quality rapid testing. Specifically, the sampler automatically collects representative physical samples from the purified coal stream and prepares and transports them to the near-infrared and X-ray fluorescence fusion spectroscopy module. This module combines the rapid analysis of organic components by near-infrared spectroscopy with the high-precision detection capability of inorganic elements by X-ray fluorescence spectroscopy to simultaneously and in real-time online determine the key quality indicators of the coal sample (such as ash content, sulfur content, volatile matter, calorific value, and the content of multiple elements).

[0037] In one embodiment of this application, the coal flow processing system includes a camera, an infrared sensor, and a lidar; acquiring multi-dimensional information corresponding to the coal flow includes: The visible light image information captured by the camera is acquired; the visible light image information is used to reflect the visual morphological characteristics of the coal flow. The infrared thermal imaging information collected by the infrared sensor is acquired; the infrared thermal imaging information is used to reflect the temperature distribution of the coal flow. The three-dimensional point cloud information collected by the lidar is acquired; the three-dimensional point cloud information is used to reflect the three-dimensional dimensions of the coal flow. The visible light image information, the infrared thermal imaging information, and the three-dimensional point cloud information are determined as the multi-dimensional information corresponding to the coal flow.

[0038] In this embodiment, the data acquisition module of the coal flow processing system includes a camera (resolution ≥ 2MP, frame rate ≥ 30fps), an infrared sensor, and a lidar. The camera (machine learning camera) includes a first camera and a second camera. The first camera is installed above the coal conveyor belt before the coal crusher to monitor the raw coal flow entering the crusher. The second camera is installed above the coal conveyor belt after the coal crusher and before the coal quality rapid testing device to monitor the homogenization effect of the crushed coal flow. The infrared sensor is co-located with the first camera to synchronously acquire temperature field information of the target area. The lidar includes a first lidar and a second lidar. The first lidar is installed in conjunction with the first camera to acquire three-dimensional point cloud information of the raw coal flow; the second lidar is installed in conjunction with the second camera to acquire three-dimensional point cloud information of the crushed coal flow.

[0039] Specifically, the first camera can acquire first visible light image information of the coal flow before crushing, reflecting the visual morphological features such as surface texture, color, and shape of the original coal blocks and impurities. The second camera can acquire second visible light image information of the coal flow after crushing, reflecting the visual morphological features such as particle size distribution and contour characteristics of the crushed coal blocks. The first lidar can simultaneously acquire high-precision three-dimensional point clouds (first three-dimensional point cloud information) of each target (coal block, impurities) in the original coal flow, used to calculate its true physical size and spatial pose. The second lidar can acquire three-dimensional point clouds (second three-dimensional point cloud information) of the crushed coal flow, used to accurately and objectively evaluate the particle size distribution of the crushed coal particles, providing direct feedback for adjusting the parameters of the crusher.

[0040] In one embodiment of this application, to overcome the severe impact of high dust and low illumination industrial environments such as underground coal mines or transfer stations on imaging quality, a ring-shaped or array-arranged diffuse reflection LED (Light Emitting Diode) supplementary light source can be integrated onto the camera's mounting bracket. The light emitted by the diffuse reflection LED supplementary light source is scattered by a specially designed diffuser, forming a uniform and soft surface light source, effectively eliminating specular reflections and heavy shadows on the surface of coal and debris. When the camera works in conjunction with this professional supplementary light source, it can significantly suppress the scattering interference of suspended dust on light and improve the signal-to-noise ratio under low illumination. This allows for stable and clear real-time capture of high-quality visible light image information with rich details and moderate contrast corresponding to coal and debris, even under harsh working conditions.

[0041] After acquiring visible light image information, infrared thermal imaging information, and 3D point cloud information, the data acquisition module transmits them to the data processing and recognition module for further data processing. Specifically, the data processing and recognition module uses time synchronization and spatial registration technology to perform multimodal data fusion of visible light image information, infrared thermal imaging information, and 3D point cloud information from the first camera and the first lidar to generate first multi-dimensional information for debris identification and initial coal particle size determination. At the same time, it fuses visible light image information from the second camera and 3D point cloud information from the second lidar to generate second multi-dimensional information for evaluating the crushing effect and serving as control feedback.

[0042] This application embodiment deploys multiple sets of sensors at key workstations (before and after the crusher) and implements collaborative data acquisition and fusion processing. The fusion information from the first set of sensors (first camera + infrared sensor + first lidar) ensures the accuracy of identification and preliminary judgment, while the fusion information from the second set of sensors (second camera + second lidar) provides an objective quantitative assessment of the crushing effect. Based on this, the upstream crusher parameters are dynamically adjusted, enabling the system to not only accurately remove impurities and initially crush large coal pieces, but also continuously optimize the crushing effect through downstream feedback. Ultimately, this ensures that the coal flow entering the coal quality rapid testing device reaches its optimal state in terms of purity and particle size uniformity, fundamentally improving the accuracy and reliability of rapid testing data.

[0043] In one embodiment of this application, the step of dividing the coal flow into impurities and coal blocks based on the multi-dimensional information, and obtaining the pose information corresponding to the impurities and the particle size information corresponding to the coal blocks, includes: Based on the morphological differences between debris and coal lumps, and by analyzing the coal flow using the visible light image information, a first classification result is obtained; The second classification result is obtained by analyzing the coal flow based on the temperature difference between the debris and the coal lumps, as well as the infrared thermal imaging information. Based on the first classification result and the second classification result, the coal flow is divided into debris and coal blocks; The position and orientation of the debris on the coal conveyor belt are determined based on the three-dimensional point cloud information, and the position and orientation are determined as the pose information corresponding to the debris. Based on the visible light image information and the three-dimensional point cloud information, the particle size information corresponding to the coal block is determined.

[0044] In practical applications, there are significant differences in visual morphology, such as surface texture, color, and geometric shape, between debris (e.g., stones, wood, metal) and coal. Visible light images captured by cameras can clearly capture these spatial and appearance features. Simultaneously, due to differences in specific heat capacity, thermal conductivity, and surface emissivity, the surface temperature distribution of debris and coal differs under the same environment. Infrared thermal imaging information acquired by infrared sensors can non-contactly detect and present these temperature field characteristics. Therefore, in this embodiment, by deploying a lightweight YOLOv5s target detection model (debris recognition model) to analyze visible light image information and by analyzing infrared thermal imaging information using a thermal image classification algorithm, each target in the coal flow can be independently perceived and initially judged from both visual and thermophysical dimensions. This results in a first classification result based on morphology to distinguish between debris and coal, and a second classification result based on thermal features to distinguish between debris and coal.

[0045] To achieve high-precision and robust real-time identification of debris in the complex and ever-changing real-world coal mine conveyor belt transportation scenario, it is necessary to train and optimize the CNN model based on a specific dataset to obtain the debris recognition model in this embodiment. Specifically, firstly, a dedicated pixel-level labeled dataset needs to be constructed. The data comes from continuously collected conveyor belt transportation images under real working conditions, covering various common debris types such as stones, wood, metal, and plastic. All images are pixel-level labeled to accurately outline the location and category of debris, forming the initial training set. To improve the debris recognition model's adaptability to complex interferences such as changes in lighting, dust, moisture, and target blur, data augmentation processing is performed on the dataset before training. Data augmentation processing can include random rotation, brightness and contrast perturbation, Gaussian blur simulation, and synthetic dust occlusion. Secondly, a proprietary model architecture is selected and constructed, preferably using the lightweight YOLOv5s architecture as the foundation. Its backbone network adopts CSPDarknet53, which can effectively perform feature extraction and gradient flow. In the feature fusion part, SPPF (Spatial Pyramid System) is introduced. Pooling-Fast (fast spatial pyramid pooling) multi-scale feature pooling module enhances the clutter recognition model's ability to perceive and detect clutter (especially small targets) at different scales in images. Finally, to address the class imbalance problem caused by the small proportion of clutter targets in coal mine images (negative samples far outnumber positive samples), Focal Loss is introduced into the loss function, making the training process more focused on samples that are difficult to classify. At the same time, a transfer learning method is adopted, loading model weights pre-trained on a large general dataset (such as COCO) as initialization, rather than random initialization, thereby significantly improving the model's convergence speed and final recognition accuracy.

[0046] Specifically, the data processing and recognition module inputs the visible light image information (first visible light image information) captured by the first camera into the visual classification model of the pre-trained clutter recognition model, and the model outputs the category label to which the target belongs. Confidence level The pixel-level bounding box is used as the first classification result; simultaneously, the infrared thermal imaging information collected by the infrared sensor is input into the thermal feature classification model in the pre-trained clutter recognition model, which will also output the category label of the target. and confidence level This is used as the second classification result. Subsequently, the system employs a confidence-weighted decision-level fusion method to comprehensively evaluate the first and second classification results, and then fuses the confidence scores. The corresponding category is determined as the final category of the target. This allows for the precise and robust separation of debris and coal chunks in the coal flow.

[0047] Specifically, in the process of comprehensively evaluating the first and second classification results using a confidence-weighted approach, if the category labels in the first and second classification results are the same (i.e., ... At this point, this category can be directly used as the final category. ( or Either approach is acceptable, and the confidence scores from the first and second classification results are fused to obtain the fused confidence score. The fusion confidence score at this point is calculated using the following formula:

[0048] If the category labels in the first classification result and the second classification result are different (i.e.) If the first classification result is obtained, a weighted comparison is used. Thermal image weights corresponding to the second classification results Based on visual weights, thermal imaging weights, and the confidence scores from the first and second classification results, the fusion confidence score can be obtained. Specifically, the weighted confidence score (visual confidence score) corresponding to the first classification result and the weighted confidence score (thermal image confidence score) corresponding to the second classification result are calculated using the following formula:

[0049]

[0050] in, For visual confidence, This represents the confidence level of the thermal image.

[0051] Subsequently, based on The category corresponding to the maximum value between the two weighted confidence scores can be selected as the final category. And this maximum value is used as the fusion confidence level. .

[0052] The final debris recognition model will output the final category corresponding to the coal flow. and their corresponding fusion confidence. This is used for subsequent control of the sorting robot and alarm triggering.

[0053] In some embodiments, it also includes: If the fusion confidence of the target is lower than the threshold θ, it is marked as "uncertain" and the security policy is triggered: The categories described in the target are classified as "miscellaneous" to prevent omissions. Simultaneously upload the target image to the human-computer interaction module and prompt for manual review; The target sample is automatically added to the "hard sample pool" for subsequent incremental training of the model.

[0054] After classifying the targets, it is also necessary to obtain the precise grasping coordinates (pose information) of the debris and the particle size information of the coal. Specifically, in the debris localization stage, the target point cloud clusters identified as debris are extracted using the three-dimensional point cloud information (first three-dimensional point cloud information) collected by the first lidar and synchronized with the first visible light image information. Using the precise three-dimensional coordinates of the point cloud, combined with the camera-lidar joint calibration parameters, the image bounding box is mapped to three-dimensional space, and the three-dimensional spatial coordinates (X, Y, Z) and main orientation (attitude) of the debris in the belt coordinate system are calculated. In the coal particle size analysis stage, the particle size of the targets identified as coal is calculated using the visible light image information (second visible light image information) collected by the second camera and the three-dimensional point cloud information (second three-dimensional point cloud information) collected by the second lidar.

[0055] In one embodiment of this application, the pixel-level bounding box output by the debris recognition model accurately identifies the position and range of the debris in the two-dimensional image. During the process of projecting this two-dimensional boundary into three-dimensional space using camera-LiDAR joint calibration parameters, a corresponding three-dimensional search area can be quickly defined in the LiDAR point cloud based on this pixel-level bounding box. This significantly reduces the data processing scope, allowing the system to accurately locate and extract the point cloud cluster belonging to the specific debris without complex general segmentation of massive panoramic point clouds. Based on this high-precision three-dimensional point cloud cluster, its geometric center can be calculated to obtain precise (X, Y, Z) coordinates and orientation. In other words, the pixel-level bounding box is essentially an efficient spatial index and filter, accurately associating the image recognition result with the three-dimensional physical world. This is a key prerequisite for achieving millimeter-level precise grasping by the robotic arm, enabling the sorting robot to plan a collision-free, optimized grasping trajectory and end-effector orientation in real time based on this high-precision pose information, thereby reliably completing the dynamic grasping and removal of moving debris.

[0056] This application embodiment, by fusing visual detection models (visible light image information) with thermal feature analysis (infrared thermal imaging information), can significantly improve the recognition accuracy under complex lighting conditions, dust interference, and camouflaged debris (such as metal covered by coal dust).

[0057] In one embodiment of this application, determining the particle size information corresponding to the coal block based on the visible light image information and the three-dimensional point cloud information includes: The visible light image information is segmented at the pixel level to extract the pixel area corresponding to the coal block; The first particle size estimate of the coal block is determined based on the preset calibration coefficient and the pixel area corresponding to the coal block. The three-dimensional point cloud information is segmented and geometrically analyzed to obtain the second particle size estimate corresponding to the coal block. The particle size information corresponding to the coal block is determined based on the first particle size estimate and the second particle size estimate corresponding to the coal block.

[0058] In this embodiment of the application, in order to evaluate whether the crushed coal sample meets the rapid detection particle size requirements, in the data processing and recognition module, the visible light image information (second visible light image information) acquired by the second camera needs to be semantically segmented (pixel-level segmentation) based on the U-Net network based on the particle size recognition model to accurately extract the contour of each particle in the crushed coal flow; then, based on the contour mask of each particle, the pixel area of ​​each contour is calculated, and converted into the equivalent diameter according to the pixel / millimeter coefficient (preset calibration coefficient) pre-calibrated on the belt plane, so as to obtain the first particle size estimate based on the two-dimensional image, and a particle size distribution histogram is generated statistically.

[0059] To accurately and automatically assess whether the particle size of the crushed coal flow meets the preset particle size requirements, it is necessary to train and optimize the CNN model based on a professional dataset to obtain the particle size recognition model in this embodiment. Specifically, firstly, according to the requirements of industry-standard rapid inspection equipment, particle size recognition needs to be transformed into a binary classification problem, that is, the coal particle size is determined into two categories: "qualified" and "unqualified". The quantification standard is the preset particle size requirement, which is directly related to production quality and is the ultimate optimization goal of model training. Secondly, in order to accurately segment each coal particle in the image, the U-Net semantic segmentation network architecture can be used to achieve pixel-level accurate segmentation with limited training data. During training, a labeled coal flow image dataset is required, in which the contour of each coal particle is accurately labeled at the pixel level. The trained U-Net model can process the input coal flow image in real time.

[0060] Specifically, in order to obtain more accurate granular data that is unaffected by perspective, the data processing and recognition module also needs to perform point cloud clustering and three-dimensional geometric analysis on the three-dimensional point cloud information (second three-dimensional point cloud information) collected by the second lidar through a granularity recognition model, so as to directly calculate the minimum circumscribed cuboid size or equivalent sphere volume of each coal block point cloud cluster, and obtain an objective second granularity estimate based on three-dimensional measurement.

[0061] By performing independent and parallel granularity calculations on 2D images (U-Net segmentation) and 3D point clouds (point cloud segmentation and geometric analysis), two sets of granularity distribution data can be obtained. After obtaining the first and second granularity estimates, they need to be compared and fused based on preset granularity requirements to obtain the final granularity information.

[0062] As an example, when the sieve particle size fraction greater than 6mm calculated from both the first and second particle size estimates is less than 5%, the particle size is deemed acceptable. If discrepancies exist, the 3D point cloud result is prioritized, or manual verification is triggered. It should be noted that the particle size information corresponding to the coal block refers to the preset particle size requirement of "6mm sieve particle size fraction ≤ 5%" and is merely an example. Specific standards can be set based on the specific technical requirements of the coal quality rapid testing device, relevant industry standards, or production process indicators. This application's embodiments do not impose any limitations on this.

[0063] This application's embodiments, by combining high-precision U-Net image segmentation with three-dimensional point cloud geometric measurement and introducing a dual-source data verification mechanism, can fundamentally overcome the particle size analysis errors caused by particle stacking, perspective distortion, and fixed calibration in traditional pure image methods, ensuring that the judgment of whether the particle size of the crushed coal sample is qualified is objective, accurate, and reliable.

[0064] In one embodiment of this application, controlling the sorting robot to grab the debris based on the pose information corresponding to the debris, so as to remove the debris conveyed on the coal conveyor belt, includes: The grasping time when the debris reaches the preset grasping position and the grasping posture of the sorting robot are determined based on the pose information corresponding to the debris. The sorting robot plans its grasping path based on the grasping time and the preset grasping position; Based on the grasping posture and grasping path of the sorting robot, the sorting robot is controlled to grasp the debris in order to remove the debris conveyed on the coal conveyor belt.

[0065] In this embodiment of the application, after the data processing and recognition module detects the presence of debris on the conveyor belt, it needs to determine the precise time when the debris arrives at the preset gripping station based on the real-time pose information (coordinates, attitude) of the debris and the real-time running speed of the coal conveyor belt. The module then transmits the task instruction containing the predicted time, target pose, and debris type information to the control and feedback module to trigger the generation of the gripping control signal of the sorting robot.

[0066] Specifically, in the control and feedback module, the embedded trajectory planner calculates a collision-free, high-efficiency grasping path in real time based on the received task instructions and the kinematic model and dynamic constraints of the sorting robot. It then matches the optimal grasping posture determined by the type and orientation of the debris (e.g., electromagnetic adsorption for metal blocks and gripper gripping for wood blocks). Subsequently, based on the grasping posture and path, a grasping control signal containing joint angle sequences, speed, and torque commands is generated and transmitted via a high-speed bus to the sorting robot's servo drive system. This enables the sorting robot to automatically and accurately move to the grasping point and execute the grasping action at a predetermined time in the planned posture, completing the removal of the debris.

[0067] In one embodiment of this application, such as Figure 1 As shown, in addition to controlling the sorting robot to grab debris, the system can also control pneumatic push rods to perform rapid rejection actions to handle specific types of debris or working conditions. Specifically, for lightweight, regularly shaped debris (such as large pieces of plastic or fabric) or debris located at the edge of the robot's grasping range, the system can switch to pneumatic push rod rejection mode. Based on the debris's positional information, the control and feedback module drives the corresponding pneumatic push rod to extend rapidly at a precise moment, pushing the debris laterally away from the conveyor belt and causing it to fall into the waste collection device on the side. This mode has the advantages of extremely fast response speed, simple structure, and convenient maintenance. It complements the sorting robot, enriching the system's debris rejection methods and improving the overall robustness and efficiency of the processing.

[0068] In one embodiment of this application, staff can also view the debris information (such as category, size, and location image) reported in real time by the data processing and recognition module through the human-machine interface, and issue intervention commands through the human-machine interaction module, such as pausing automatic rejection, manually marking new debris types for model learning, or remotely controlling the sorting robot to perform cleaning tasks at specific locations, thereby achieving flexible monitoring and high-level intervention of the system.

[0069] This application embodiment enables dynamic collaborative operation between sorting robots and high-speed conveyor belts through real-time prediction, intelligent planning, and precise control. It achieves fully automated and high-success-rate removal of debris without stopping the machine or affecting the main coal flow, greatly improving the continuity and intelligence of production.

[0070] In one embodiment of this application, the system also incorporates robust security recovery logic to ensure safe and automatic recovery after abnormal situations or manual intervention. Specifically, this is achieved through the following three levels: I. Alarm Triggering and Equipment Suspension: When the fusion confidence score of the debris recognition model output is greater than 0.8 and its pixel area in the image is greater than 500 pixels, the system immediately triggers an audible and visual alarm and automatically sends a control command to the PLC (Programmable Logic Controller) to suspend the operation of the coal block conveyor belt and the downstream reduction equipment. At the same time, the image, recognition result, timestamp and other information of the debris are automatically stored as a "pending event" log.

[0071] II. Alarm Stop and Task Clearing Conditions: When the system triggers a debris alarm, if the staff manually confirms "cleared" through the human-machine interface, the sorting robot confirms "removal completed" through feedback signal, or no target debris is detected in the image recognition results of 5 consecutive frames, then the current grasping task will be cleared and the alarm will be stopped.

[0072] III. Conditions for restoring automatic operation: After the alarm stops, the system must simultaneously meet all the following conditions: the camera monitoring screen indicates that the original debris target has disappeared; the coal conveyor belt has been running unloaded at the preset grabbing station for 3 seconds; the status feedback of the associated coal crusher and other downstream equipment is "normal"; and the system self-check program confirms that there are no faults in each module (data acquisition, processing, and control) before it can automatically restore the normal monitoring and processing process.

[0073] This application embodiment, through the aforementioned security recovery logic, can automatically control the belt and equipment to resume operation, and record the complete timestamp and event details of this interruption and recovery in the log, thereby greatly improving the reliability, security and event traceability of the system's long-term operation while ensuring processing efficiency.

[0074] In one embodiment of this application, adjusting the crushing parameters of the coal block crusher according to the particle size information of the coal block, and controlling the coal block crusher to crush the coal blocks conveyed on the coal block conveyor belt based on the adjusted crushing parameters, includes: When the particle size information corresponding to the coal block does not meet the preset particle size requirement, the crushing intensity and the number of crushing times in the crushing parameters are adjusted to obtain the adjusted crushing intensity and the number of crushing times. Based on the adjusted crushing intensity and number of crushing cycles, the coal crusher is controlled to crush the coal blocks conveyed on the coal block conveyor belt.

[0075] In this embodiment, the system uses the particle size distribution of the crushed coal stream (particle size information corresponding to coal blocks) as the core feedback indicator. As an example, the preset particle size requirement can be set as follows: the cumulative mass percentage of oversize material with a particle size greater than 6mm in the crushed coal stream does not exceed 5% (i.e., qualified). This embodiment does not limit the preset particle size requirement.

[0076] The data processing and recognition module analyzes data collected by the second camera and the second lidar to calculate the particle size distribution of the current coal flow in real time. If a situation of "excessively large particles" is identified (e.g., the mass fraction on a 6mm screen > 5%), the control and feedback module will automatically trigger a parameter adjustment strategy. On the one hand, it sends instructions to the coal crusher to reduce the set values ​​of its key components (such as the jaw plate spacing, roller gap, or hammer-screen gap) to enhance the intensity of a single crushing operation; on the other hand, it simultaneously increases the operating speed of the crusher's main shaft or rotor to increase the number of impacts or compressions on the material per unit time.

[0077] After adjusting the crushing intensity and number of crushing cycles, the coal crusher will perform enhanced crushing on the coal subsequently transported to its working area based on the adjusted crushing intensity and number of crushing cycles to obtain crushed products with finer particle size and more uniform distribution. Simultaneously, the system will continuously monitor and analyze the particle size information of the subsequent coal flow. If the particle size returns to acceptable levels, the current parameters will be maintained; if it still does not meet the standards, the crushing parameters will be iteratively fine-tuned based on the new particle size deviation until the preset acceptable standards are reached and stabilized.

[0078] This application embodiment ensures that the particle size of coal samples entering subsequent rapid coal quality testing remains stable within the qualified standard by implementing real-time closed-loop control of downstream particle size monitoring results and upstream crushing equipment parameters, thereby guaranteeing the accuracy of rapid testing data.

[0079] In one embodiment of this application, after adjusting the crushing parameters of the coal block crusher according to the particle size information corresponding to the coal block, and controlling the coal block crusher to crush the coal blocks conveyed on the coal block conveyor belt based on the adjusted crushing parameters, the method further includes: After a preset time period, determine whether the crushed particle size information of the coal block after being crushed by the coal block crusher meets the preset particle size requirements; If the crushing particle size information does not meet the preset particle size requirement after N consecutive judgments, an abnormal signal is generated to characterize the abnormality of the coal block crusher; where N is a preset positive integer.

[0080] In this embodiment, to ensure the stability of parameter adjustments and avoid system oscillations, the system automatically enters a preset observation period after each adjustment of crushing parameters (such as gap and rotation speed). During this period, the data processing and identification module continuously analyzes the particle size of the downstream coal flow (the crushed particle size information of the coal blocks after being crushed by the coal block crusher). If the particle size is consistently unqualified within N consecutive observation periods, the control and feedback module will determine that the automatic adjustment has failed and immediately generate an alarm (abnormal signal) indicating equipment malfunction. This alarm not only provides audible and visual prompts through the human-machine interface but also automatically associates and retrieves particle size data, equipment operating parameters, and historical records for the abnormal time period, providing clear maintenance guidance for staff and indicating that it may be necessary to inspect or replace vulnerable parts such as crusher hammers and screen plates.

[0081] As an example, the preset time period can be set to 30 seconds and N can be set to 3. It should be noted that the specific values ​​of the preset time period and N are only examples. Those skilled in the art can set them according to the actual process flow, equipment response characteristics and the trade-off between control stability and sensitivity. The embodiments of this application do not limit this.

[0082] This application embodiment determines whether the coal crusher is abnormal by continuously judging within a preset time period and a preset number of times, making the automatic adjustment behavior of the system more reliable, effectively avoiding misadjustment caused by instantaneous fluctuations, ensuring the stability of the particle size control process, and turning passive post-event maintenance into proactive pre-event warning, thereby significantly improving the maintainability, operating efficiency and long-term reliability of the entire system.

[0083] In one embodiment of this application, the human-computer interaction module can provide an intuitive operating interface, making it convenient for staff to view information such as system operating status, recognition results, and equipment parameters in real time. At the same time, it supports staff to manually intervene in and adjust settings of the system, such as viewing historical data.

[0084] In one embodiment of this application, a machine learning-based quality assessment module can be added to the subsequent stages of the coal rapid testing device. Based on the detected coal particle size, appearance characteristics, and other relevant data, a pre-trained machine learning model is used to conduct a more comprehensive and in-depth assessment of the coal quality, predict the performance of the coal in different scenarios such as combustion and chemical utilization, and provide a more scientific basis for the rational utilization and value enhancement of coal.

[0085] This application embodiment deploys machine learning cameras, infrared sensors, and lidar on the coal conveyor belt before crushing and reducing the coal mass. This enables real-time identification and removal of debris, preventing hard foreign objects such as metal and wood from entering the coal crushing equipment (coal crusher). This significantly reduces mechanical impact and blockage risks, lowers equipment damage rates, extends equipment lifespan, and reduces maintenance frequency and downtime. Furthermore, by deploying machine learning cameras and lidar on the coal conveyor belt after crushing and reducing the coal mass, the system calculates online whether the crushed coal meets preset particle size requirements. Closed-loop control of the coal crusher's crushing parameters ensures that the coal samples entering the rapid testing device have uniform and qualified particle sizes, eliminating detection errors caused by particle size deviations and improving the accuracy, reproducibility, and traceability of coal quality data. Additionally, the CNN model completes debris identification and particle size prediction at the edge level in milliseconds. The system automatically executes the entire sequence of "identification—alarm—pause—processing—resumption" without manual intervention, significantly reducing labor intensity and costs, and achieving unmanned operation of the rapid coal quality testing process. Finally, all data and identification results are entered into the database in real time. The background system statistically analyzes the types of debris, their frequency of occurrence, and particle size distribution trends, providing quantitative basis for process parameter optimization, equipment status diagnosis, and production scheduling decisions. This continuously releases the value of data and helps coal preparation plants improve quality and efficiency and upgrade their intelligence.

[0086] It should be noted that, for the sake of simplicity, the method embodiments are all described as a series of actions. However, those skilled in the art should understand that the embodiments of this application are not limited to the described order of actions, because according to the embodiments of this application, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also understand that the embodiments described in the specification are all preferred embodiments, and the actions involved are not necessarily required by the embodiments of this application.

[0087] This application also provides a storage medium that stores computer instructions. When the computer executes the computer instructions, it is used to perform the various processes of the above-described coal flow processing method embodiments and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0088] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0089] This application also provides an electronic device, including a processor 3010, a memory 309, and a program or instructions stored in the memory 309 and executable on the processor 3010. When the program or instructions are executed by the processor 3010, they implement the various processes of the above-described coal flow processing method embodiments and achieve the same technical effects. To avoid repetition, they will not be described again here.

[0090] It should be noted that the electronic devices in the embodiments of this application include the mobile electronic devices and non-mobile electronic devices described above.

[0091] Figure 3 A schematic diagram of the hardware structure of an electronic device to implement an embodiment of this application. The electronic device 300 includes, but is not limited to, components such as: radio frequency unit 301, network module 302, audio output unit 303, input unit 304, sensor 305, display unit 306, user input unit 307, interface unit 308, memory 309, and processor 3010.

[0092] Those skilled in the art will understand that the electronic device 300 may also include a power supply (such as a battery) for supplying power to various components. The power supply may be logically connected to the processor 3010 through a power management system, thereby enabling functions such as managing charging, discharging, and power consumption through the power management system. Figure 3 The electronic device structure shown does not constitute a limitation on the electronic device. The electronic device may include more or fewer components than shown, or combine certain components, or have different component arrangements, which will not be elaborated here. This application also provides a computer program product, including a computer program / instruction. When the computer program / instruction is executed by a processor, it implements the various processes of the above-described coal flow processing method embodiments and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0093] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0094] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0095] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

Claims

1. A method for treating coal streams, characterized in that, Applied to a coal flow processing system, the coal flow processing system including a coal block conveyor belt for conveying the coal flow, a sorting robot, and a coal block crusher, the method includes: Obtain multi-dimensional information corresponding to the coal flow; Based on the multi-dimensional information, the coal flow is divided into debris and coal blocks, and the pose information corresponding to the debris and the particle size information corresponding to the coal blocks are obtained. Based on the position information corresponding to the debris, the sorting robot is controlled to grab the debris in order to remove the debris being transported on the coal conveyor belt; Based on the particle size information of the coal block, the crushing parameters of the coal block crusher are adjusted, and the coal block crusher is controlled to crush the coal block conveyed on the coal block conveyor belt based on the adjusted crushing parameters.

2. The method according to claim 1, characterized in that, The coal flow processing system includes a camera, an infrared sensor, and a lidar. The acquisition of multi-dimensional information corresponding to the coal flow includes: The visible light image information captured by the camera is acquired; the visible light image information is used to reflect the visual morphological characteristics of the coal flow. The infrared thermal imaging information collected by the infrared sensor is acquired; the infrared thermal imaging information is used to reflect the temperature distribution of the coal flow. The three-dimensional point cloud information collected by the lidar is acquired; the three-dimensional point cloud information is used to reflect the three-dimensional dimensions of the coal flow. The visible light image information, the infrared thermal imaging information, and the three-dimensional point cloud information are determined as the multi-dimensional information corresponding to the coal flow.

3. The method according to claim 2, characterized in that, The step of dividing the coal flow into debris and coal blocks based on the multi-dimensional information, and obtaining the pose information corresponding to the debris and the particle size information corresponding to the coal blocks, includes: Based on the morphological differences between debris and coal lumps, and by analyzing the coal flow using the visible light image information, a first classification result is obtained; The second classification result is obtained by analyzing the coal flow based on the temperature difference between the debris and the coal lumps, as well as the infrared thermal imaging information. Based on the first classification result and the second classification result, the coal flow is divided into debris and coal blocks; The position and orientation of the debris on the coal conveyor belt are determined based on the three-dimensional point cloud information, and the position and orientation are determined as the pose information corresponding to the debris. Based on the visible light image information and the three-dimensional point cloud information, the particle size information corresponding to the coal block is determined.

4. The method according to claim 3, characterized in that, The step of determining the particle size information corresponding to the coal block based on the visible light image information and the three-dimensional point cloud information includes: The visible light image information is segmented at the pixel level to extract the pixel area corresponding to the coal block; The first particle size estimate of the coal block is determined based on the preset calibration coefficient and the pixel area corresponding to the coal block. The three-dimensional point cloud information is segmented and geometrically analyzed to obtain the second particle size estimate corresponding to the coal block. The particle size information corresponding to the coal block is determined based on the first particle size estimate and the second particle size estimate corresponding to the coal block.

5. The method according to claim 1, characterized in that, The step of controlling the sorting robot to grab the debris based on the pose information corresponding to the debris, so as to remove the debris conveyed on the coal conveyor belt, includes: The grasping time when the debris reaches the preset grasping position and the grasping posture of the sorting robot are determined based on the pose information corresponding to the debris. The sorting robot plans its grasping path based on the grasping time and the preset grasping position; Based on the grasping posture and grasping path of the sorting robot, the sorting robot is controlled to grasp the debris in order to remove the debris conveyed on the coal conveyor belt.

6. The method according to claim 1, characterized in that, The step of adjusting the crushing parameters of the coal block crusher according to the particle size information of the coal block, and controlling the coal block crusher to crush the coal blocks conveyed on the coal block conveyor belt based on the adjusted crushing parameters, includes: When the particle size information corresponding to the coal block does not meet the preset particle size requirement, the crushing intensity and the number of crushing times in the crushing parameters are adjusted to obtain the adjusted crushing intensity and the number of crushing times. Based on the adjusted crushing intensity and number of crushing cycles, the coal crusher is controlled to crush the coal blocks conveyed on the coal block conveyor belt.

7. The method according to claim 6, characterized in that, After adjusting the crushing parameters of the coal block crusher according to the particle size information of the coal block, and controlling the coal block crusher to crush the coal blocks conveyed on the coal block conveyor belt based on the adjusted crushing parameters, the method further includes: After a preset time period, determine whether the crushed particle size information of the coal block after being crushed by the coal block crusher meets the preset particle size requirements; If the crushing particle size information does not meet the preset particle size requirement after N consecutive judgments, an abnormal signal is generated to characterize the abnormality of the coal block crusher; where N is a preset positive integer.

8. A storage medium, characterized in that, The storage medium stores computer instructions, which, when executed by the computer, are used to perform a coal flow processing method as described in any one of claims 1-7.

9. An electronic device, characterized in that, Includes at least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform a coal flow processing method as described in any one of claims 1-7.

10. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement a coal flow processing method as described in any one of claims 1-7.

Images