Intelligent perception system for explosion-proof unmanned transport vehicle in coal mine

Abstract

The application discloses a kind of coal mine explosion-proof unmanned transport vehicle intelligent sensing system, comprising: multi-modal mixed sensing module, gas weather self-adaptive monitoring module, explosion-proof edge computing module;Multi-modal mixed sensing module is used to collect the multi-modal environment sensing information of target coal mine by multidimensional environment sensing detection unit in real time;Gas weather self-adaptive monitoring module is used to collect the gas-weather parameter information of the area where explosion-proof unmanned transport vehicle is located in target coal mine in real time, wherein, gas parameter information includes at least one of methane concentration, carbon monoxide concentration, dust concentration, and weather parameter information includes at least one of air temperature, air humidity;Explosion-proof edge computing module is used to determine the fusion weight of multi-modal environment sensing information based on gas-weather parameter information, and fusion processing is carried out on multi-modal environment sensing information based on fusion weight, and based on fusion processing result, explosion-proof unmanned transport vehicle is autonomously navigated.

Description

Technical Field

[0001] This invention relates to the field of intelligent mining technology, and in particular to an intelligent sensing system for explosion-proof unmanned transport vehicles in coal mines. Background Technology

[0002] With the continuous advancement of intelligent coal mine construction, replacing traditional manual or semi-automatic transportation methods with unmanned transport vehicles has become an important development direction for improving underground logistics efficiency and reducing safety risks. Therefore, the safe and stable operation of unmanned transport vehicles has become particularly crucial.

[0003] Currently, intelligent driving of unmanned transport vehicles is typically achieved by following preset tracks or using magnetic nails for navigation. However, this method cannot cope with temporary obstacles or path changes, has poor flexibility, and can lead to vehicle damage and affect safe production in coal mines. Furthermore, magnetic nail navigation has high initial deployment and maintenance costs. Summary of the Invention

[0004] This invention provides an intelligent sensing system for explosion-proof unmanned transport vehicles in coal mines, which mainly improves the flexibility of unmanned transport vehicles, ensures stable and safe operation of vehicles, saves vehicle operating costs, and ensures safe and stable production in coal mines.

[0005] According to a first aspect of the present invention, an intelligent sensing system for explosion-proof unmanned transport vehicles in coal mines is provided, comprising: Multimodal hybrid sensing module, gas and meteorological adaptive monitoring module, explosion-proof edge computing module; The multimodal hybrid sensing module is used to collect multimodal environmental sensing information of the target coal mine in real time through a multidimensional environmental sensing and detection unit. The multimodal environmental sensing information includes at least static environmental sensing information and dynamic obstacle sensing information. The gas meteorological adaptive monitoring module is used to collect gas-meteorological parameter information in the area where the explosion-proof unmanned transport vehicle is located in the target coal mine in real time. The gas parameter information includes at least one of methane concentration, carbon monoxide concentration, and dust concentration, and the meteorological parameter information includes at least one of air temperature and air humidity. The explosion-proof edge computing module is used to determine the fusion weight of the multimodal environmental perception information based on the gas-meteorological parameter information, and to perform fusion processing on the multimodal environmental perception information collected by the multidimensional environmental perception detection unit based on the fusion weight, and to perform autonomous navigation of the explosion-proof unmanned transport vehicle based on the fusion processing result.

[0006] Optionally, the multi-dimensional environmental perception and detection unit includes a solid-state lidar, a millimeter-wave radar, an infrared thermal imaging device, and a visible light binocular stereo vision array. The solid-state lidar is used to generate a high-precision three-dimensional point cloud map in real time to identify the static environmental perception information of the target coal mine, wherein the static environmental perception information includes at least the roadway outline, track, support structure, and static obstacles. The millimeter-wave radar is used to penetrate a preset medium to detect dynamic obstacle perception information of the target coal mine in real time, and to measure the relative speed information between the dynamic obstacle and the explosion-proof unmanned transport vehicle. The preset medium includes at least one of dust and water mist. The dynamic obstacle perception information includes at least moving vehicles, moving personnel, and moving mining equipment. The infrared thermal imaging device is used to detect the heat source of the target coal mine in real time under a preset environment in order to identify the dynamic obstacle perception information of the target coal mine, wherein the preset environment includes at least one of low illumination and dust environment. The visible light binocular stereo vision array is used to collect color image information of the area surrounding the explosion-proof unmanned transport vehicle in real time, and to identify the static environmental perception information of the target coal mine based on the color image information. The static environmental perception information includes at least traffic signals, signs, ground markings, and personnel postures.

[0007] Optionally, the explosion-proof edge computing module is used to determine the fusion weights of the multimodal environmental perception information based on the gas-meteorological parameter information by performing the following steps: If the dust concentration is greater than a preset dust concentration threshold, the fusion weight of the multimodal environmental perception information collected by the millimeter-wave radar and the infrared thermal imaging device is increased, while the fusion weight of the multimodal environmental perception information collected by the solid-state lidar and the visible light binocular stereo vision array is decreased.

[0008] Optionally, The explosion-proof edge computing module is also used to determine in real time whether the methane concentration is greater than the preset methane concentration threshold. If so, it determines the speed reduction of the explosion-proof unmanned transport vehicle and the safety zone of the target coal mine. Based on the speed reduction, it controls the explosion-proof unmanned transport vehicle to decelerate and travel to the safety zone, and issues a warning of excessive methane concentration through preset warning information.

[0009] Optionally, it may also include a vehicle-road cooperative communication module; The vehicle-road cooperative communication module is used to realize the first information interaction between multiple explosion-proof unmanned transport vehicles, the second information interaction between the explosion-proof unmanned transport vehicles and the underground infrastructure of the target coal mine, and the third information interaction between the explosion-proof unmanned transport vehicles and the ground dispatch center corresponding to the target coal mine using UWB (Ultra-Wideband) communication. The first and second information include at least one of location information and intent information, and the third information includes at least one of regional video information of the area where the explosion-proof unmanned transport vehicles are located, vehicle status information, and instruction information from the ground dispatch center.

[0010] Optionally, The vehicle-road cooperative communication module is used to determine the distance between the explosion-proof unmanned transport vehicle and the roadway access control facility of the underground infrastructure in real time. When the distance is less than a preset distance threshold, it generates a passage status switching command for the roadway access control facility and sends the passage status switching command to the roadway access control facility using 5G communication technology to realize the passage status switching of the roadway access control facility. The passage status switching command is used to control the roadway access control facility to switch from a first working state to a second working state. The first working state indicates that passage is prohibited or restricted, and the second working state indicates that passage is permitted.

[0011] Optionally, the explosion-proof edge computing module has a built-in preset motion trajectory prediction model. The explosion-proof edge computing module is used to perform the following steps to achieve autonomous navigation of the explosion-proof unmanned transport vehicle based on the fusion processing results. The preset motion trajectory prediction model is used to identify and classify at least one target object in the target coal mine based on the fusion result, wherein the target object includes at least one of dynamic organisms, unstructured debris, and underground pipeline facilities; The preset motion trajectory prediction model is also used to obtain the historical motion state of the target object within a preset time period, and predict the motion trajectory and state change trend of the target object within a preset time period based on the historical motion state. The explosion-proof edge computing module is used to generate vehicle control commands based on the classification of the target object, the motion trajectory, and the state change trend to adjust the driving state of the explosion-proof unmanned transport vehicle.

[0012] Optionally, the explosion-proof edge computing module has a built-in model training unit; The model training unit is used to train and construct the preset motion trajectory prediction model; The model training unit is used to train and construct the preset motion trajectory prediction model by performing the following steps: Construct a pre-defined initial motion trajectory prediction model; Obtain a sample dataset, wherein the sample dataset includes multimodal environmental perception information of sample coal mine scenarios with object category labels and object motion trajectory labels, and the historical motion state of sample objects; The sample dataset is divided into a training set and a test set. The training set is used to train the preset initial motion trajectory prediction model, and the test set is used to test the trained preset initial motion trajectory prediction model. Finally, the trained preset initial motion trajectory prediction model that meets the test conditions is taken as the preset motion trajectory prediction model.

[0013] Optionally, The preset motion trajectory prediction model is used to predict the motion trajectory of the dynamic organism within a preset time period based on the historical motion state by performing the following steps: Obtain the biological attribute information of the dynamic organism, predict the spatial location distribution of the dynamic organism within a future preset time based on the biological attribute information, the historical movement state, and the fusion result, and use the spatial location distribution as the movement trajectory of the dynamic organism. The biological attribute information includes at least one of the following: current movement speed, current movement direction, and movement behavior intention. The preset motion trajectory prediction model is used to predict the state change trend of the unstructured debris within a preset time period based on the historical motion state by performing the following steps: Obtain the geometric structural feature information of the unstructured debris, determine the stability of the unstructured debris based on the geometric structural feature information, and predict the collapse probability and collapse impact area of ​​the unstructured debris within a preset time period based on the stability. Use the collapse probability and the collapse impact area as the state change trend of the unstructured debris. The preset motion trajectory prediction model is used to predict the motion trajectory of the downhole pipeline facility within a preset time period based on the historical motion state by performing the following steps: The airflow disturbance parameters caused by the explosion-proof unmanned transport vehicle during its operation and the physical suspension characteristics of the underground pipeline facility are obtained. Based on the airflow disturbance parameters and the physical suspension characteristics, the swing amplitude and deformation range of the underground pipeline facility are predicted, and the swing amplitude and the deformation range are used as the motion trajectory of the underground pipeline facility.

[0014] Optionally, the gas meteorological adaptive monitoring module integrates a methane concentration sensor, a carbon monoxide concentration sensor, a dust concentration sensor, a temperature sensor, and a humidity sensor. The methane concentration sensor is used to collect the methane concentration in the area where the explosion-proof unmanned transport vehicle is located in the target coal mine in real time. The carbon monoxide concentration sensor is used to collect the carbon monoxide concentration in the area where the explosion-proof unmanned transport vehicle is located in the target coal mine in real time. The dust concentration sensor is used to collect the dust concentration in the area where the explosion-proof unmanned transport vehicle is located in the target coal mine in real time. The temperature sensor is used to collect the air temperature in the area where the explosion-proof unmanned transport vehicle is located in the target coal mine in real time. The humidity sensor is used to collect the air humidity in the area where the explosion-proof unmanned transport vehicle is located in the target coal mine in real time.

[0015] The intelligent perception system for explosion-proof unmanned transport vehicles in coal mines provided by this invention, compared with the current method of intelligent driving of unmanned transport vehicles by using navigation along preset tracks or magnetic nails, utilizes a multi-dimensional environmental perception and detection unit deployed in a multi-modal hybrid perception module. This allows for the perception of the target coal mine environment from multiple dimensions and in multiple ways, acquiring rich and comprehensive environmental information. This helps to more accurately understand the complex environmental conditions within the coal mine, providing sufficient data support for subsequent intelligent decision-making and autonomous navigation. Navigating the vehicle using real-time acquired environmental perception information improves the flexibility of vehicle navigation, avoids collisions, and ensures the safe and stable operation of the vehicle. By determining the fusion weights of multi-modal environmental perception information through real-time collected gas-meteorological parameter information, the system can more rationally allocate attention to different perception information based on the current environmental conditions, improving the accuracy and effectiveness of information fusion. Multi-modal environmental perception information comes from different sensors and perception methods, each with its own advantages and limitations. Through fusion processing, these multi-source information can be comprehensively utilized, giving full play to the advantages of different sensors, compensating for the shortcomings of a single sensor, improving the comprehensiveness of environmental perception, and thus improving the accuracy of subsequent vehicle navigation. Attached Figure Description

[0016] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings: Figure 1 This diagram illustrates the structure of an intelligent sensing system for explosion-proof unmanned transport vehicles in coal mines, as provided in an embodiment of the present invention. Figure 2This invention provides a structural diagram of another intelligent sensing system for explosion-proof unmanned transport vehicles in coal mines, according to an embodiment of the present invention. Detailed Implementation

[0017] The present invention will be described in detail below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in the present application can be combined with each other.

[0018] Currently, the method of achieving intelligent driving of unmanned transport vehicles by using pre-set tracks or magnetic nail navigation cannot cope with temporary obstacles or path changes, has poor flexibility, can lead to vehicle damage, and has high initial deployment and maintenance costs.

[0019] To address the aforementioned problems, embodiments of the present invention provide an intelligent sensing system for explosion-proof unmanned transport vehicles in coal mines, such as... Figure 1 As shown, it includes: a multimodal hybrid sensing module, a gas and meteorological adaptive monitoring module, and an explosion-proof edge computing module; The multimodal hybrid sensing module is used to collect multimodal environmental sensing information of the target coal mine in real time through a multidimensional environmental sensing and detection unit. The multimodal environmental sensing information includes at least static environmental sensing information and dynamic obstacle sensing information. The gas meteorological adaptive monitoring module is used to collect gas-meteorological parameter information in the area where the explosion-proof unmanned transport vehicle is located in the target coal mine in real time. The gas parameter information includes at least one of methane concentration, carbon monoxide concentration, and dust concentration, and the meteorological parameter information includes at least one of air temperature and air humidity. The explosion-proof edge computing module is used to determine the fusion weight of the multimodal environmental perception information based on the gas-meteorological parameter information, and to perform fusion processing on the multimodal environmental perception information collected by the multidimensional environmental perception detection unit based on the fusion weight, and to perform autonomous navigation of the explosion-proof unmanned transport vehicle based on the fusion processing result.

[0020] This invention provides a highly integrated, adaptable, and deep learning-capable intelligent sensing system for explosion-proof unmanned transport vehicles in coal mines. The system is mounted on an explosion-proof lithium battery-powered unmanned transport flatbed vehicle, which is primarily responsible for transporting equipment, materials, and support supplies between the main transport roadway and various coal mining faces in underground coal mines.

[0021] This invention addresses the limitations of unmanned explosion-proof transport vehicles in coal mines by proposing an intelligent perception system for these vehicles. This system integrates multiple modules, including a multimodal hybrid perception module, a gas and meteorological adaptive monitoring module, and an explosion-proof edge computing module, to achieve flexible navigation of the unmanned transport vehicle and ensure its stable and safe operation. The multimodal hybrid sensing module includes a multi-dimensional environmental sensing and detection unit for acquiring high-precision 3D point cloud data. It primarily identifies static environmental sensing information such as the geometric contours of roadways and fixed support structures, unaffected by changes in lighting. It is also used to detect heat sources in complete darkness or smoky environments, focusing on identifying dynamic obstacles with temperature characteristics, such as workers and operating equipment. Furthermore, it collects texture information in well-lit areas to assist in identifying traffic signs, road markings, and obstacle color features. This allows for multi-dimensional and multi-method perception of the target coal mine environment, acquiring rich and comprehensive environmental information, which helps to more accurately understand the complex environmental conditions within the coal mine. The gas and meteorological adaptive monitoring module is deployed around the vehicle's key air intakes and sensors, incorporating a high-precision explosion-proof sensor array to collect real-time gas and meteorological parameter information of the vehicle's microenvironment. The gas and meteorological adaptive monitoring module receives real-time gas and meteorological parameter information and, based on the gas- The dynamic calculation of meteorological parameter information to determine the fusion weights of each environmental sensing and detection unit allows for a more rational allocation of attention to different sensing information based on current environmental conditions, improving the accuracy and effectiveness of information fusion. Based on the calculated fusion weights, the static environmental information and dynamic obstacle information collected by each detection unit are spatiotemporally aligned and fused at the feature level. Through a weighted algorithm, a high-confidence comprehensive semantic map of the underground environment is generated. In this map, even if a single sensor fails or its performance degrades, the fusion result can still accurately reflect the location, type, and movement trend of obstacles. Finally, based on the fusion processing results, the system plans the optimal driving path in real time, generates speed, steering, and braking commands, and controls the explosion-proof unmanned transport vehicle to achieve autonomous navigation. Through fusion processing, multi-source information can be comprehensively utilized, fully leveraging the advantages of different sensors, compensating for the shortcomings of a single sensor, improving the comprehensiveness of environmental perception, and thus enhancing the safety and stable operation of subsequent vehicles.

[0022] In this embodiment of the invention, optionally, the multi-dimensional environmental perception and detection unit includes a solid-state lidar, a millimeter-wave radar, an infrared thermal imaging device, and a visible light binocular stereo vision array; the solid-state lidar is used to generate a high-precision three-dimensional point cloud map in real time to identify the static environmental perception information of the target coal mine, wherein the static environmental perception information includes at least the roadway outline, track, support structure, and static obstacles; the millimeter-wave radar is used to penetrate a preset medium to detect the dynamic obstacle perception information of the target coal mine in real time, and to measure the relative speed information between the dynamic obstacles and the explosion-proof unmanned transport vehicle, wherein the preset medium includes dust, water mist, and other particles. One less; the dynamic obstacle perception information includes at least moving vehicles, moving personnel, and moving mining equipment; the infrared thermal imaging device is used to detect the heat source of the target coal mine in real time under a preset environment to identify the dynamic obstacle perception information of the target coal mine, wherein the preset environment includes at least one of low illumination and dust environment; the visible light binocular stereo vision array is used to collect color image information of the area around the explosion-proof unmanned transport vehicle in real time, and identify the static environment perception information of the target coal mine based on the color image information, wherein the static environment perception information includes at least traffic signals, signs, ground markings, and personnel posture.

[0023] Specifically, the multi-dimensional environmental perception and detection unit is responsible for acquiring information about the surrounding environment of the explosion-proof unmanned transport vehicle in all directions and under all weather conditions. To overcome the limitations of single sensors in coal mines with high dust levels, low illumination, water mist, and complex and variable road surfaces, this embodiment of the invention adopts a hybrid perception scheme. First, four sets of fusion systems combining solid-state lidar and millimeter-wave radar are symmetrically installed in the front, rear, left, and right directions of the vehicle. The solid-state lidar is responsible for high-precision 3D modeling of the tunnel environment, generating point cloud maps in real time, and accurately identifying tunnel contours, tracks, support structures, and static obstacles. The millimeter-wave radar, installed alongside the lidar, has wavelength characteristics that allow it to penetrate dust and water mist, specifically for detecting distant dynamic targets, such as moving mine cars, personnel, or large equipment, and can accurately measure the relative speed of the target, providing predictive information for the vehicle's decision-making system. Second, a wide dynamic range infrared thermal imaging device and a visible light binocular stereo vision array are installed at the front and rear of the vehicle. The visible light binocular stereo vision array (visible light binocular stereo vision camera), by simulating human vision, can not only acquire high-definition color image information for recognizing static information such as traffic signals, signs, and ground markings, but also accurately identify personnel postures through deep learning algorithms, such as determining whether a worker is walking, crouching down to work, or has accidentally fallen. Its wide dynamic range ensures that images are not overexposed or underexposed at the boundary between light and dark areas in the tunnel, such as when entering a dark area from an illuminated area. The infrared thermal imaging device is unaffected by light. By detecting heat sources in the environment, it can detect personnel and heat-generating equipment in complete darkness or dense dust, providing a key guarantee for safety redundancy. For example, when a worker is at a corner where visibility is obstructed, the thermal imaging device can detect the infrared radiation generated by their body temperature in advance. Finally, twelve ultrasonic sensors are evenly distributed around the vehicle chassis as a supplement to near-range obstacle detection. They are mainly used for accurate distance measurement when the vehicle starts, stops, or passes through narrow areas to prevent low-speed collisions with tunnel walls or other equipment.

[0024] In this embodiment of the invention, optionally, the gas meteorological adaptive monitoring module integrates a methane concentration sensor, a carbon monoxide concentration sensor, a dust concentration sensor, a temperature sensor, and a humidity sensor; the methane concentration sensor is used to collect the methane concentration in the area where the explosion-proof unmanned transport vehicle is located in the target coal mine in real time; the carbon monoxide concentration sensor is used to collect the carbon monoxide concentration in the area where the explosion-proof unmanned transport vehicle is located in the target coal mine in real time; the dust concentration sensor is used to collect the dust concentration in the area where the explosion-proof unmanned transport vehicle is located in the target coal mine in real time; the temperature sensor is used to collect the air temperature in the area where the explosion-proof unmanned transport vehicle is located in the target coal mine in real time; and the humidity sensor is used to collect the air humidity in the area where the explosion-proof unmanned transport vehicle is located in the target coal mine in real time.

[0025] The gas and meteorological adaptive monitoring module is integrated into the vehicle itself or deployed at key roadway nodes to construct the micro-meteorological field of the vehicle's operating area in real time. Specifically, the gas and meteorological adaptive monitoring module highly integrates five types of high-precision sensor arrays to form a multi-dimensional environmental perception unit: a methane concentration sensor, which uses catalytic combustion or laser absorption spectroscopy technology to capture the volume fraction of methane (CH4) in the area in real time; a carbon monoxide concentration sensor, which uses electrochemical principles to sensitively detect the concentration of carbon monoxide (CO) produced by spontaneous combustion or incomplete combustion of coal; a dust concentration sensor, which uses light scattering principles to quantify the mass concentration of suspended coal dust and rock dust in the air in real time; a temperature sensor, which accurately measures the thermodynamic temperature of the ambient air; and a humidity sensor, which monitors the relative humidity of the ambient air in real time.

[0026] In an embodiment of the present invention, optionally, the explosion-proof edge computing module is used to determine the fusion weight of the multimodal environmental perception information based on the gas-meteorological parameter information by performing the following steps: determining whether the dust concentration is greater than a preset dust concentration threshold; if so, increasing the fusion weight of the multimodal environmental perception information collected by the millimeter-wave radar and the infrared thermal imaging device, and decreasing the fusion weight of the multimodal environmental perception information collected by the solid-state lidar and the visible light binocular stereo vision array.

[0027] Specifically, the explosion-proof edge computing module is a high-performance edge computing platform encapsulated in an explosion-proof shell. Internally, it houses a multi-core heterogeneous processor, including a CPU, GPU, and a dedicated AI acceleration chip. The core task of this module is to process massive amounts of data from the multimodal hybrid sensing module in real time. During operation, the explosion-proof edge computing module first performs temporal and spatial synchronization alignment on the multimodal environmental sensing information collected by the multi-dimensional environmental sensing detection unit and the gas-meteorological parameter information collected by the gas-meteorological adaptive monitoring module. Then, it runs a multi-sensor deep fusion algorithm based on an attention mechanism. This algorithm dynamically adjusts the weights of data collected by different environmental sensing detection units according to the current gas-meteorological parameter information. For example, when the dust concentration in the gas-meteorological parameter information is extremely high, the algorithm automatically reduces the weight of data collected by LiDAR and visible light cameras, relying more heavily on data collected by millimeter-wave radar and infrared thermal imaging devices for decision-making. This embodiment of the invention dynamically adjusts weights based on gas-meteorological parameters, integrating the advantages of different sensors to obtain more comprehensive and accurate environmental information in various environments, thereby improving the overall quality of environmental sensing.

[0028] In an embodiment of the present invention, optionally, the explosion-proof edge calculation module is further configured to determine in real time whether the methane concentration is greater than a preset methane concentration threshold. If so, the module determines the speed reduction of the explosion-proof unmanned transport vehicle and the safety zone of the target coal mine, controls the explosion-proof unmanned transport vehicle to decelerate and travel to the safety zone based on the speed reduction, and issues a warning of excessive methane concentration through preset warning information.

[0029] Among them, the system can use in-vehicle voice or flashing lights to issue warnings of excessive methane concentrations. The preset warning information covers information on excessive methane concentrations and avoidance measures, and can be set according to actual needs.

[0030] Specifically, the explosion-proof edge computing module receives methane concentration data from the gas and meteorological adaptive monitoring module in real time and compares it with a preset methane concentration safety threshold (set according to actual needs). If the current methane concentration is less than or equal to the threshold, the vehicle maintains normal autonomous navigation operation; if the current methane concentration is greater than the preset threshold, the module immediately triggers the "gas over-limit emergency avoidance mode". Once the emergency avoidance mode is entered, the module simultaneously performs the following two core calculations: Speed ​​reduction calculation: Based on the current methane concentration exceeding the standard and the vehicle's current speed, the speed reduction is calculated using a preset nonlinear decay algorithm. The greater the exceedance, the greater the required deceleration, ensuring that the vehicle decelerates smoothly without generating mechanical sparks. Avoidance area locking: Combining the vehicle's current location, underground electronic map, and real-time ventilation network data, the system quickly retrieves and determines the nearest avoidance area that meets safety standards, such as: refuge chambers, fresh air flow tunnels, or designated temporary parking points. The system then plans an optimal safe path from the current location to the avoidance area, prioritizing the avoidance of high methane accumulation areas and complex obstacle areas. The final module, based on the calculated speed reduction, sends a deceleration command to the vehicle chassis control system, controlling the explosion-proof unmanned transport vehicle to perform a smooth deceleration operation and guiding it to automatically travel along the planned path to the locked refuge area. Once the vehicle reaches the refuge area, it automatically executes a parking power-off or low-power standby command. During vehicle deceleration and travel, the module simultaneously controls the vehicle's audible and visual alarm to emit a high-frequency sound and a red flashing light, alerting nearby personnel and vehicles to take evasive action. This embodiment of the invention ensures that in the event of an abnormal gas outburst, the unmanned transport vehicle can automatically reduce speed, evacuate the danger zone, and issue an alarm immediately, minimizing the risk of explosion.

[0031] In an embodiment of the invention, optionally, the explosion-proof edge computing module incorporates a preset motion trajectory prediction model. The explosion-proof edge computing module is used to perform the following steps to achieve autonomous navigation of the explosion-proof unmanned transport vehicle based on the fusion processing results: the preset motion trajectory prediction model is used to identify and classify at least one target object in the target coal mine based on the fusion results, wherein the target object includes at least one of dynamic organisms, unstructured debris, and underground pipeline facilities; the preset motion trajectory prediction model is also used to obtain the historical motion state of the target object within a preset time period, and predict the motion trajectory and state change trend of the target object within a preset time period based on the historical motion state; the explosion-proof edge computing module is used to generate vehicle control commands to adjust the driving state of the explosion-proof unmanned transport vehicle based on the classification of the target object, the motion trajectory, and the state change trend.

[0032] Among them, dynamic organisms include, but are not limited to, underground workers, inspection robots or other mobile vehicles; unstructured debris includes, but is not limited to, fallen rocks on the roadway floor, piled coal gangue, fallen water pipes or temporarily piled support materials; underground pipeline facilities include, but are not limited to, suspended ventilation ducts, laid cable trays, drainage pipes and track facilities; historical motion status includes, but is not limited to, the target object's position coordinate sequence, instantaneous velocity vector, acceleration changes and heading angle fluctuations at historical moments.

[0033] Specifically, the explosion-proof edge computing module first receives multi-source fusion processing results from multiple sensors, including LiDAR, millimeter-wave radar, and visible / infrared cameras. A preset motion trajectory prediction model, based on deep learning algorithms, performs real-time inference on the fused data, identifying and classifying at least one type of target object from the complex underground background. For each identified target object, the model constructs its spatiotemporal database, extracting historical motion state data within a preset time window. Based on this historical motion state data, the model infers the object's probabilistic travel path or movement range envelope within a preset future time window. For example, it predicts whether workers are about to cross the tunnel or whether debris is on a vehicle's likely path. Simultaneously, the model predicts the state change trend of the target object. For example, it predicts whether the ventilation duct will sway and intrude into the vehicle clearance due to airflow impact, or whether the vehicle ahead is likely to brake suddenly or change lanes. Then, the explosion-proof edge computing module integrates the target object's classification, predicted motion trajectory, and state change trend to conduct a collision risk assessment. Based on the collision risk assessment results, it generates vehicle control commands to adjust the driving state of the explosion-proof unmanned transport vehicle. For example, if the predicted trajectory indicates that a dynamic organism will enter within the vehicle's safe braking distance, the module assigns the highest avoidance weight to that object; if unstructured debris is predicted to be stationary, a detour path is planned; if pipeline facilities are predicted to sway, a larger safe lateral clearance is reserved in advance. Control commands include longitudinal control commands, lateral control commands, and audio-visual interaction commands. Based on the executed vehicle driving state, commands such as smooth deceleration, emergency braking, or maintaining a constant speed are executed to match the expected speed of the dynamic target ahead. Based on the lateral control commands, commands such as fine-tuning steering, changing lanes, or stopping to wait are executed to avoid predicted danger zones. Based on the audio-visual interaction commands, commands such as triggering audio-visual warnings in advance if it is predicted that personnel may not notice the vehicle. This embodiment of the invention, by predicting the trajectory and state change trends of target objects, enables vehicles to have "predictive" driving capabilities, effectively responding to emergencies in unstructured underground environments, and significantly improving the safety and traffic efficiency of unmanned transportation systems.

[0034] In this embodiment of the invention, optionally, the explosion-proof edge computing module has a built-in model training unit; the model training unit is used to train and construct the preset motion trajectory prediction model; the model training unit is used to train and construct the preset motion trajectory prediction model by performing the following steps: constructing a preset initial motion trajectory prediction model; obtaining a sample dataset, wherein the sample dataset includes multimodal environmental perception information of a sample coal mine scene with object category labels and object motion trajectory labels, and the historical motion state of sample objects; dividing the sample dataset into a training set and a test set, using the training set to train the preset initial motion trajectory prediction model, and using the test set to test the trained preset initial motion trajectory prediction model, and finally using the trained preset initial motion trajectory prediction model that meets the test conditions as the preset motion trajectory prediction model.

[0035] Specifically, during model training, a pre-defined initial motion trajectory prediction model is first constructed, followed by the acquisition of a sample dataset. The dataset ensures it contains all necessary files, including historical motion states of various target objects in the sample coal mine scenario, such as multimodal environmental perception information, dynamic organisms, unstructured debris, and underground pipeline facilities, including historical acceleration, velocity, and position data, as well as the actual future motion trajectories of these target objects relative to historical time. The data is then converted to a format understandable by the pre-defined initial motion trajectory prediction model. Finally, the model is trained and tested. Specifically, the dataset can be divided first: using random or specific strategies (such as stratified sampling), the sample dataset is divided into training and testing sets. The training set is then used to train the model, and the testing set is used to test the trained model and evaluate its performance on unseen data. Precision, recall, and other metrics on the test set are calculated and recorded. If the model performance does not meet requirements, it can return to the training phase for further iterations or adjustments. This process yields a pre-defined motion trajectory prediction model that meets the requirements.

[0036] In another embodiment of the present invention, the model structure of the preset initial motion trajectory prediction model is the same as that of the preset motion trajectory prediction model. The preset motion trajectory prediction model includes an input layer, a feature extraction layer, and a motion trajectory and state change trend recognition layer. When using the preset motion trajectory prediction model to identify the motion trajectory and state change trend of an object, the historical motion state of the target object is first input into the preset motion trajectory prediction model. The input is then fed into the feature extraction layer for feature extraction. The motion trajectory and state change trend recognition layer identifies the features output by the feature extraction layer to obtain the motion trajectory and state change trend of the target object.

[0037] In an embodiment of the present invention, optionally, the preset motion trajectory prediction model is used to predict the motion trajectory of the dynamic organism within a preset time period based on the historical motion state by performing the following steps: obtaining the organism attribute information of the dynamic organism, predicting the spatial location distribution of the dynamic organism within a preset time period based on the organism attribute information, the historical motion state, and the fusion result, and using the spatial location distribution as the motion trajectory of the dynamic organism, wherein the organism attribute information includes at least one of current movement speed, current movement direction, and movement behavior intention; the preset motion trajectory prediction model is used to predict the state change trend of the unstructured debris within a preset time period based on the historical motion state by performing the following steps: obtaining the geometric structure of the unstructured debris. The system uses geometric structural features to determine the stability of the unstructured debris and predicts its collapse probability and impact area within a preset timeframe based on the stability. The collapse probability and impact area are used as the trend of the unstructured debris's state change. The preset motion trajectory prediction model is used to predict the motion trajectory of the underground pipeline facility within a preset timeframe based on the historical motion state by performing the following steps: acquiring airflow disturbance parameters caused by the explosion-proof unmanned transport vehicle during operation and the physical suspension characteristics of the underground pipeline facility; predicting the swing amplitude and deformation range of the underground pipeline facility based on the airflow disturbance parameters and physical suspension characteristics; and using the swing amplitude and deformation range as the motion trajectory of the underground pipeline facility.

[0038] Specifically, regarding the prediction process of the movement trajectory of a dynamic organism: the model first extracts the organism's biological attribute information from the fused perception results, including the current instantaneous movement velocity vector, heading angle, and movement behavior intentions derived from posture recognition, such as normal walking, running urgently, crouching down to work, crossing alleys, or stopping to observe. Combining the above attribute information with the historical movement state within a preset time period, such as acceleration change patterns and gait periodicity, the model simulates the organism's decision-making process and outputs the spatial position distribution of the dynamic organism within a preset time window in the future. This position distribution not only includes the most likely driving path but also covers the potential deviation range caused by the uncertainty of intention.

[0039] The prediction process for the state change trend of unstructured debris involves the model analyzing the geometric structural characteristics of the debris, including accumulation slope, center of gravity location, surface roughness, and contact area with the tunnel floor. Based on these geometric features, the current stability coefficient of the debris is calculated. Simultaneously, ground micro-vibration data caused by approaching vehicles are used to dynamically correct the stability assessment. If the stability is lower than a preset threshold set according to actual needs, the model further simulates the instability process under gravity, predicting the probability of collapse within a preset timeframe and the impact area (i.e., the coverage area of ​​the debris rolling or sliding) should a collapse occur. Based on these predictions, the collapse probability and impact area are output as the state change trend. If the predicted probability is high and the impact area covers the vehicle's predetermined path, the system will trigger a detour or stop command in advance.

[0040] The prediction process for underground pipeline facilities involves real-time calculation of airflow disturbance parameters such as wind speed increment, wind pressure distribution, and turbulence intensity generated by explosion-proof unmanned transport vehicles traveling at high speeds. Physical suspension characteristics of the pipeline, such as material elastic modulus, suspension spacing, initial sag, and fixed-end constraints, are extracted. The model establishes a coupled vibration equation between airflow and structure, simulating the excitation force generated on the pipeline by the airflow impact at the moment the vehicle passes. It predicts the pipeline's sway amplitude (e.g., maximum lateral displacement) and deformation range (e.g., longitudinal tension or torsion) within a preset timeframe. This prediction result is transformed into the pipeline's dynamic motion trajectory envelope, ensuring that vehicles not only avoid the pipeline's static position but also allow sufficient dynamic safety margin to prevent scraping or collision accidents.

[0041] The embodiments of the present invention enable explosion-proof unmanned transport vehicles to determine the intentions of personnel, the risk of falling rocks, and the swaying of ventilation ducts, thereby achieving true proactive safety defense in complex underground unstructured environments.

[0042] In another embodiment of the present invention, the high-performance explosion-proof edge computing unit also integrates a real-time localization and mapping (SLAM) algorithm, which can achieve continuous and accurate vehicle positioning even in underground wells where GPS signals are completely absent by matching real-time point cloud data with pre-stored high-precision maps.

[0043] In embodiments of the present invention, optionally, such as Figure 2As shown, the intelligent perception system for explosion-proof unmanned transport vehicles in coal mines also includes a vehicle-road cooperative communication module. This module uses UWB (Ultra-Wideband) communication to achieve: first information interaction between multiple explosion-proof unmanned transport vehicles; second information interaction between the explosion-proof unmanned transport vehicles and the underground infrastructure of the target coal mine; and third information interaction between the explosion-proof unmanned transport vehicles and the ground dispatch center corresponding to the target coal mine. The first and second information include at least one of location information and intent information, and the third information includes at least one of regional video information of the area where the explosion-proof unmanned transport vehicle is located, vehicle status information, and instruction information from the ground dispatch center.

[0044] Among them, intent information refers to the actions or movement trends that the vehicle plans to perform within a preset time window in the future, such as lane change intent, turning intent, acceleration and deceleration intent, etc.; vehicle status information includes information such as speed, remaining battery power, load information, and operation mode; instruction information refers to the control commands, task assignments, or strategy adjustment signals issued by the ground dispatch center to the explosion-proof unmanned transport vehicle through the vehicle-road cooperative communication module.

[0045] Specifically, the vehicles are equipped with vehicle-to-infrastructure (V2I) communication modules based on UWB (Ultra-Wideband) and 5G technologies. Through UWB communication, vehicles can exchange high-precision, low-latency location and driving intention information with pre-deployed positioning base stations and other vehicles equipped with the same system within the tunnel, forming a "perception network" among vehicles. At intersections or in blind spots, even if a vehicle's own sensors haven't detected another vehicle, it can learn the other's location and intention in advance through communication, thereby collaboratively planning the passage order and completely eliminating the risk of collisions. Through the underground 5G network, vehicles can transmit real-time high-definition video, report vehicle status, and receive remote commands with the ground dispatch center. The dispatch center can take over the vehicles at any time for remote control driving and handle extreme emergencies. Furthermore, vehicles can also interact with intelligent devices underground. This embodiment of the invention, through the interaction of "intention information" and "vehicle status information," allows the ground dispatch center to achieve refined management of the underground vehicle fleet, while multiple vehicles underground can autonomously and collaboratively avoid traffic lights based on each other's intentions, significantly improving the efficiency and safety of underground coal mine transportation.

[0046] In an embodiment of the present invention, optionally, the vehicle-road cooperative communication module is used to determine in real time the distance between the explosion-proof unmanned transport vehicle and the roadway access control facility of the underground infrastructure. When the distance is less than a preset distance threshold, it generates a passage status switching command for the roadway access control facility and sends the passage status switching command to the roadway access control facility using 5G communication technology to realize the passage status switching of the roadway access control facility. The passage status switching command is used to control the roadway access control facility to switch from a first working state to a second working state. The first working state indicates that passage is prohibited or restricted, and the second working state indicates that passage is permitted.

[0047] The preset distance threshold is set according to actual needs. Specifically, the vehicle-road cooperative communication module uses UWB high-precision positioning technology to calculate in real time the straight-line distance or the distance along the road path between the current position of the explosion-proof unmanned transport vehicle and the target roadway access control facility ahead. When the distance is greater than the preset distance threshold, the vehicle continues to drive normally, and the facility maintains its current state; when the distance is less than or equal to the preset distance threshold, the module immediately triggers the passage request logic. Once the trigger logic is activated, the vehicle-road cooperative communication module automatically generates a passage state switching instruction. This instruction includes the facility ID and vehicle identity authentication, and instructs the facility to switch from the first working state (prohibited / restricted passage, such as the damper closed and the barrier lowered) to the second working state (allowed passage, such as the damper open and the barrier raised). Considering the complex electromagnetic environment underground and the high requirements for low latency, the module uses 5G communication technology to encapsulate the generated access status switching command into a data packet, which is then sent in real time to the receiving terminal of the target roadway access control facility via an underground 5G base station. If the 5G signal is interrupted, the module can automatically downgrade to Wi-Fi 6 or leaky cable communication as a backup link to ensure the command can be delivered. After receiving the command, the roadway access control facility verifies the vehicle's permissions locally and immediately drives the actuator (motor or hydraulic rod) to smoothly switch the facility from the first working state to the second working state, opening the passage for the vehicle. During the switching process, the audible and visual alarms around the facility are activated simultaneously to warn nearby personnel to take evasive action. After the vehicle has completely passed through the facility, the facility automatically or by sending a "close command" from the vehicle restores it from the second working state to the first working state to maintain the stability of the underground ventilation system or prevent the spread of potential risks. This embodiment of the invention realizes active interaction between vehicles and infrastructure, eliminating the safety hazards and efficiency bottlenecks of traditional manual door opening or fixed delayed door opening, and ensuring the continuous and efficient passage of explosion-proof unmanned transport vehicles in the complex underground road network.

[0048] In another embodiment of the present invention, the intelligent sensing system for explosion-proof unmanned transport vehicles in coal mines further includes an integrated explosion-proof power supply and control module for supplying power and intelligently controlling the explosion-proof unmanned transport vehicles in coal mines.

[0049] The following is a demonstration of the workflow of the intelligent sensing system for explosion-proof unmanned transport vehicles in coal mines: An unmanned transport vehicle equipped with this system receives an instruction from the dispatch center: to transport a batch of hydraulic supports to the No. 3 coal face. Startup and Self-Check: After starting the vehicle, the system first performs a comprehensive self-check to confirm that all sensors, computing units, and communication modules are functioning normally. Path Planning: Based on the mission objective and the built-in high-precision map, the edge computing unit plans an optimal path, which comprehensively considers distance, gradient, tunnel width, and real-time traffic information. Driving and Perception: The vehicle starts and travels along the predetermined path. The real-time 3D point cloud constructed by the lidar is matched with the map to achieve precise navigation. The millimeter-wave radar continuously scans for distant dynamic targets ahead. The binocular camera identifies speed limit signs and turn indicators on the tunnel walls. Dynamic Obstacle Avoidance and Collaboration: During travel, the millimeter-wave radar detects a coal transport vehicle traveling in the same direction at a relatively slow speed 200 meters ahead. The system calculates the collision time and smoothly reduces the vehicle's speed to maintain a safe following distance. Meanwhile, an infrared thermal imaging camera in an equipment maintenance chamber on one side of the tunnel detected a worker operating there. Although the light was dim, the worker's clear thermal signal caused the vehicle to slow down in advance, and a voice prompt was issued through the vehicle's loudspeaker: "Attention, unmanned vehicle is approaching, please maintain a safe distance." Environmental Adaptation: When the vehicle entered a dusty tunneling area, the environmental monitoring module detected that the dust concentration exceeded the preset value. The fusion algorithm of the edge computing unit automatically increased the data weight of the millimeter-wave radar and infrared thermal imaging, and the vehicle speed adaptively decreased by 20% to ensure sufficient reaction time even with reduced visibility. Intelligent Linkage: When approaching a windproof door, the vehicle-road cooperative communication module automatically established a connection with the door controller, and the door opened smoothly. After the vehicle passed smoothly, the door closed automatically. Mission Completion: The vehicle arrived at the designated unloading point of the No. 3 coal mining face and accurately stopped at the designated location using ultrasonic sensors and visual positioning. After completing the mission, the vehicle packaged and uploaded complete data of this transportation, including the driving trajectory, obstacles encountered, and changes in environmental parameters, to the cloud server for subsequent algorithm optimization and big data analysis.

[0050] Through the coordinated operation of the above modules, the intelligent perception system in this embodiment not only enables the unmanned transport vehicle to operate safely, efficiently, and autonomously in complex underground environments, but also, through environmental adaptation and vehicle-road collaboration, elevates the vehicle from an isolated automated unit into an intelligent node integrated into the overall intelligent production system of the mine, greatly improving the safety and efficiency of coal mine transportation.

[0051] 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 specification.

[0052] 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. An intelligent sensing system for explosion-proof unmanned transport vehicles in coal mines, characterized in that, include: Multimodal hybrid sensing module, gas and meteorological adaptive monitoring module, explosion-proof edge computing module; The multimodal hybrid sensing module is used to collect multimodal environmental sensing information of the target coal mine in real time through a multidimensional environmental sensing and detection unit. The multimodal environmental sensing information includes at least static environmental sensing information and dynamic obstacle sensing information. The gas meteorological adaptive monitoring module is used to collect gas-meteorological parameter information in the area where the explosion-proof unmanned transport vehicle is located in the target coal mine in real time. The gas parameter information includes at least one of methane concentration, carbon monoxide concentration, and dust concentration, and the meteorological parameter information includes at least one of air temperature and air humidity. The explosion-proof edge computing module is used to determine the fusion weight of the multimodal environmental perception information based on the gas-meteorological parameter information, and to perform fusion processing on the multimodal environmental perception information collected by the multidimensional environmental perception detection unit based on the fusion weight, and to perform autonomous navigation of the explosion-proof unmanned transport vehicle based on the fusion processing result.

2. The intelligent sensing system for explosion-proof unmanned transport vehicles in coal mines according to claim 1, characterized in that, The multi-dimensional environmental perception and detection unit includes a solid-state lidar, a millimeter-wave radar, an infrared thermal imaging device, and a visible light binocular stereo vision array. The solid-state lidar is used to generate a high-precision three-dimensional point cloud map in real time to identify the static environmental perception information of the target coal mine, wherein the static environmental perception information includes at least the roadway outline, track, support structure, and static obstacles. The millimeter-wave radar is used to penetrate a preset medium to detect dynamic obstacle perception information of the target coal mine in real time, and to measure the relative speed information between the dynamic obstacle and the explosion-proof unmanned transport vehicle. The preset medium includes at least one of dust and water mist. The dynamic obstacle perception information includes at least moving vehicles, moving personnel, and moving mining equipment. The infrared thermal imaging device is used to detect the heat source of the target coal mine in real time under a preset environment in order to identify the dynamic obstacle perception information of the target coal mine, wherein the preset environment includes at least one of low illumination and dust environment. The visible light binocular stereo vision array is used to collect color image information of the area surrounding the explosion-proof unmanned transport vehicle in real time, and to identify the static environmental perception information of the target coal mine based on the color image information. The static environmental perception information includes at least traffic signals, signs, ground markings, and personnel postures.

3. The intelligent sensing system for explosion-proof unmanned transport vehicles in coal mines according to claim 2, characterized in that, The explosion-proof edge computing module is used to determine the fusion weights of the multimodal environmental perception information based on the gas-meteorological parameter information by performing the following steps: If the dust concentration is greater than a preset dust concentration threshold, the fusion weight of the multimodal environmental perception information collected by the millimeter-wave radar and the infrared thermal imaging device is increased, while the fusion weight of the multimodal environmental perception information collected by the solid-state lidar and the visible light binocular stereo vision array is decreased.

4. The intelligent sensing system for explosion-proof unmanned transport vehicles in coal mines according to claim 1, characterized in that, The explosion-proof edge computing module is also used to determine in real time whether the methane concentration is greater than the preset methane concentration threshold. If so, it determines the speed reduction of the explosion-proof unmanned transport vehicle and the safety zone of the target coal mine. Based on the speed reduction, it controls the explosion-proof unmanned transport vehicle to decelerate and travel to the safety zone, and issues a warning of excessive methane concentration through preset warning information.

5. The intelligent sensing system for explosion-proof unmanned transport vehicles in coal mines according to claim 1, characterized in that, It also includes a vehicle-road cooperative communication module; The vehicle-road cooperative communication module is used to realize the first information interaction between multiple explosion-proof unmanned transport vehicles, the second information interaction between the explosion-proof unmanned transport vehicles and the underground infrastructure of the target coal mine, and the third information interaction between the explosion-proof unmanned transport vehicles and the ground dispatch center corresponding to the target coal mine using UWB (Ultra-Wideband) communication. The first and second information include at least one of location information and intent information, and the third information includes at least one of regional video information of the area where the explosion-proof unmanned transport vehicles are located, vehicle status information, and instruction information from the ground dispatch center.

6. The intelligent sensing system for explosion-proof unmanned transport vehicles in coal mines according to claim 5, characterized in that, The vehicle-road cooperative communication module is used to determine the distance between the explosion-proof unmanned transport vehicle and the roadway access control facility of the underground infrastructure in real time. When the distance is less than a preset distance threshold, it generates a passage status switching command for the roadway access control facility and sends the passage status switching command to the roadway access control facility using 5G communication technology to realize the passage status switching of the roadway access control facility. The passage status switching command is used to control the roadway access control facility to switch from a first working state to a second working state. The first working state indicates that passage is prohibited or restricted, and the second working state indicates that passage is permitted.

7. The intelligent sensing system for explosion-proof unmanned transport vehicles in coal mines according to claim 1, characterized in that, The explosion-proof edge computing module has a built-in preset motion trajectory prediction model. The explosion-proof edge computing module is used to achieve autonomous navigation of the explosion-proof unmanned transport vehicle based on the fusion processing results by performing the following steps. The preset motion trajectory prediction model is used to identify and classify at least one target object in the target coal mine based on the fusion result, wherein the target object includes at least one of dynamic organisms, unstructured debris, and underground pipeline facilities; The preset motion trajectory prediction model is also used to obtain the historical motion state of the target object within a preset time period, and predict the motion trajectory and state change trend of the target object within a preset time period based on the historical motion state. The explosion-proof edge computing module is used to generate vehicle control commands based on the classification of the target object, the motion trajectory, and the state change trend to adjust the driving state of the explosion-proof unmanned transport vehicle.

8. The intelligent sensing system for explosion-proof unmanned transport vehicles in coal mines according to claim 7, characterized in that, The explosion-proof edge computing module has a built-in model training unit; The model training unit is used to train and construct the preset motion trajectory prediction model; The model training unit is used to train and construct the preset motion trajectory prediction model by performing the following steps: Construct a pre-defined initial motion trajectory prediction model; Obtain a sample dataset, wherein the sample dataset includes multimodal environmental perception information of sample coal mine scenarios with object category labels and object motion trajectory labels, and the historical motion state of sample objects; The sample dataset is divided into a training set and a test set. The training set is used to train the preset initial motion trajectory prediction model, and the test set is used to test the trained preset initial motion trajectory prediction model. Finally, the trained preset initial motion trajectory prediction model that meets the test conditions is taken as the preset motion trajectory prediction model.

9. The intelligent sensing system for explosion-proof unmanned transport vehicles in coal mines according to claim 7, characterized in that, The preset motion trajectory prediction model is used to predict the motion trajectory of the dynamic organism within a preset time period based on the historical motion state by performing the following steps: Obtain the biological attribute information of the dynamic organism, predict the spatial location distribution of the dynamic organism within a future preset time based on the biological attribute information, the historical movement state, and the fusion result, and use the spatial location distribution as the movement trajectory of the dynamic organism. The biological attribute information includes at least one of the following: current movement speed, current movement direction, and movement behavior intention. The preset motion trajectory prediction model is used to predict the state change trend of the unstructured debris within a preset time period based on the historical motion state by performing the following steps: Obtain the geometric structural feature information of the unstructured debris, determine the stability of the unstructured debris based on the geometric structural feature information, and predict the collapse probability and collapse impact area of ​​the unstructured debris within a preset time period based on the stability. Use the collapse probability and the collapse impact area as the state change trend of the unstructured debris. The preset motion trajectory prediction model is used to predict the motion trajectory of the downhole pipeline facility within a preset time period based on the historical motion state by performing the following steps: The airflow disturbance parameters caused by the explosion-proof unmanned transport vehicle during its operation and the physical suspension characteristics of the underground pipeline facility are obtained. Based on the airflow disturbance parameters and the physical suspension characteristics, the swing amplitude and deformation range of the underground pipeline facility are predicted, and the swing amplitude and the deformation range are used as the motion trajectory of the underground pipeline facility.

10. The intelligent sensing system for explosion-proof unmanned transport vehicles in coal mines according to claim 1, characterized in that, The gas meteorological adaptive monitoring module integrates a methane concentration sensor, a carbon monoxide concentration sensor, a dust concentration sensor, a temperature sensor, and a humidity sensor. The methane concentration sensor is used to collect the methane concentration in the area where the explosion-proof unmanned transport vehicle is located in the target coal mine in real time. The carbon monoxide concentration sensor is used to collect the carbon monoxide concentration in the area where the explosion-proof unmanned transport vehicle is located in the target coal mine in real time. The dust concentration sensor is used to collect the dust concentration in the area where the explosion-proof unmanned transport vehicle is located in the target coal mine in real time. The temperature sensor is used to collect the air temperature in the area where the explosion-proof unmanned transport vehicle is located in the target coal mine in real time. The humidity sensor is used to collect the air humidity in the area where the explosion-proof unmanned transport vehicle is located in the target coal mine in real time.

Images