A mining underground pipe installation robot and its control system

Abstract

This invention relates to the field of underground coal mine robots, and discloses an underground mine safety management robot and its control system, including a robot body, a support plate, and a shell. The support plate integrates a central control module, an emergency execution module, and an explosion-proof lithium battery power module, while a sensing module is located on the top of the shell. The sensing module acquires raw sensing data and transmits it to the central control module. The central control module performs noise reduction processing and spatial coordinate transformation on the data, outputting a fused data stream. Its internal mining artificial intelligence algorithm unit extracts features from the fused data stream and outputs a safety status result; the emergency linkage logic unit determines the alarm level based on the safety status result and generates an emergency control command. The emergency execution module receives the command and outputs a drive electrical signal. This invention improves the accuracy of safety status identification through multimodal sensing data joint determination and, combined with a braking circuit, achieves rapid braking of the equipment in abnormal conditions.

Description

Technical Field

[0001] This invention relates to the field of underground robots in coal mines, specifically to an underground mine pipe installation robot and its control system. Background Technology

[0002] Coal mine underground inspection and safety management robots are often used to assist in environmental detection and hazard identification. However, existing mining robots still have some shortcomings in practical applications. Due to the interference of dust and moisture underground, existing sensing systems typically rely on single sensors for simple threshold determination, lacking sufficient fusion of multimodal data, leading to false alarms or missed alarms under complex working conditions. Furthermore, when processing large-scale sensing data such as laser point clouds, existing equipment usually relies on a single main control chip to handle all computational tasks, increasing the processor load, easily causing data processing delays, and reducing the real-time response of the system.

[0003] Furthermore, existing mining robots lack reliable hardware-level emergency protection mechanisms when faced with sudden dangers or malfunctions in their own control systems. The equipment typically relies solely on software commands to stop the motors, failing to effectively handle the back electromotive force in the motion drive circuit while simultaneously cutting off power input. This situation makes the robot prone to uncontrolled sliding after a power outage, and the energy accumulated in the motor windings cannot be dissipated in time, potentially posing safety risks in the demanding explosion-proof underground environment. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a mine underground safety management robot and its control system, which solves the problems of existing mine robots being prone to false alarms and missed alarms in complex working conditions due to insufficient fusion of sensing data, the delay caused by a single processor processing large-scale sensing data, and the lack of hardware-level emergency protection mechanisms, which makes the equipment prone to uncontrolled sliding and causing safety risks in abnormal states.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] The first aspect of this invention provides a mine underground pipe installation robot, comprising: Robot body, support plate, and shell; The support plate is fixed to the upper end of the robot body; The housing is fitted over the upper end of the support plate and forms a sealed cavity inside; The upper end of the support plate is integrated with a central control module, an emergency execution module, and an explosion-proof lithium battery power module. A sensing module is provided on the top of the housing; The sensing module is communicatively connected to the central control module and is used to collect external environmental information and send it to the central control module. The central control module is electrically connected to the emergency execution module and is used to perform calculations and judgments on the external environment information, and issue control commands to the emergency execution module based on the judgment results.

[0007] Preferably, the bottom of the robot body is provided with a tracked chassis structure; The robot body is equipped with a mobile drive mechanism inside. The mobile drive mechanism includes a DC motor, a motor drive module, and an incremental motor encoder. The motor drive module receives digital control signals from the central control module and outputs drive voltage to the DC motor; The output shaft of the DC motor is connected to the drive wheel of the tracked chassis structure via a transmission mechanism. The incremental motor encoder is coaxially mounted on the output end of the DC motor, collects the rotation information of the DC motor, and feeds back the speed pulse signal to the central control module.

[0008] Preferably, the upper edge of the support plate is provided with an annular sealing groove; A high-elastic silicone O-ring is installed inside the annular sealing groove; The bottom edge of the housing is attached to the high-elastic silicone O-ring; The support plate and the housing are mechanically fastened together by explosion-proof bolts. The outer surface of the housing is coated with a polytetrafluoroethylene anti-stick coating; The sensing module enters the sealed cavity through the housing via a cable, and an explosion-proof gland connector is fitted at the cable hole.

[0009] Preferably, the sensing module is mounted on a rotatable sensing tower. The bottom of the sensing tower is equipped with a tower base, a slewing support bearing, an explosion-proof stepper motor, a gear transmission assembly, an explosion-proof conductive slip ring, and an absolute encoder. The tower base is fixed to the top panel of the housing; The explosion-proof stepper motor is installed inside the sealed cavity; The output shaft of the explosion-proof stepper motor is connected to the rotary support bearing through the gear transmission assembly. The absolute encoder is coaxially mounted on the output end of the explosion-proof stepper motor; The explosion-proof conductive slip ring is coaxially mounted at the rotation axis of the sensing tower.

[0010] Preferably, a multi-modal explosion-proof sensor array is installed on the sensing tower; The multimodal explosion-proof sensor array includes an infrared thermal imager, a hazardous gas detector, and a laser scanning radar; The infrared thermal imager extracts the maximum temperature value of the current field of view and converts it into a basic probability allocation function for infrared temperature. The harmful gas detector outputs a carbon monoxide concentration value and converts it into a basic probability distribution function for carbon monoxide concentration; The accident hazard identification model inside the central control module integrates the basic probability allocation function of infrared temperature and the basic probability allocation function of carbon monoxide concentration to output a safety status result.

[0011] Preferably, the explosion-proof lithium battery power module is internally equipped with a lithium iron phosphate battery pack; The output terminal of the lithium iron phosphate battery pack is connected in series with an explosion-proof current limiting circuit and a dual overcurrent protector. The explosion-proof current limiting circuit and the dual overcurrent protector are integrally cast into an epoxy resin insulating curing layer. The explosion-proof lithium battery power module is connected to a power management and distribution circuit. The power management distribution circuit includes a voltage regulation branch and a voltage boost branch, which output operating voltages to the central control module and the sensing module, respectively.

[0012] Preferably, the central control module is internally equipped with a microcontroller, a field-programmable gate array (FPGA), and a microprocessor; The microcontroller reads the carbon monoxide concentration value and packages it into a status data frame; The field-programmable gate array receives laser point cloud data collected by the laser scanning radar, executes a voxel filtering algorithm to remove noise points, and moves the noise-reduced point cloud data to the system memory address space of the microprocessor. The microprocessor performs feature extraction operations on the multimodal data in the system memory address space and outputs the safety status result and motion planning instructions. The microprocessor sends the motion planning instructions to the microcontroller.

[0013] Preferably, the emergency execution module includes a hardware watchdog circuit, redundant safety relays, and an energy consumption braking discharge circuit; The hardware watchdog circuit monitors the operating status of the microprocessor and outputs a low-level signal to trigger the redundant safety relay to operate. The redundant safety relay disconnects the main power input circuit; The energy consumption braking discharge circuit is connected in parallel to both ends of the busbar of the motor of the moving drive mechanism. The energy-consuming braking discharge circuit is activated the instant the main power input circuit is disconnected, converting the back electromotive force of the winding into heat energy dissipation.

[0014] Preferably, the circuit board inside the central control module is equipped with a 5G communication module and a UWB positioning module; The UWB positioning module communicates with the positioning base station to obtain distance parameters; The microprocessor acquires the distance parameter and calculates the spatial coordinate information, and embeds the spatial coordinate information into the header of the data frame transmitted by the 5G communication module; The 5G communication module extracts data frames containing spatial coordinate information, performs digital modulation and power amplification via the radio frequency front end, and sends them to the underground micro base station.

[0015] A second aspect of the present invention provides a control system for a mine underground pipe-laying robot, comprising: A sensing module is used to acquire raw sensing data from underground and transmit the raw sensing data to a central control module. The central control module is used to receive the raw sensing data, perform noise reduction processing and spatial coordinate system transformation, and output a fused data stream. The central control module includes a mining artificial intelligence algorithm unit and an emergency response logic unit; The mining artificial intelligence algorithm unit is used to acquire the fused data stream, perform feature extraction operations, and output safety status results; The emergency linkage logic unit is used to determine the alarm level based on the mine safety status results and generate corresponding emergency control commands. An emergency execution module is used to receive the emergency control command and output the corresponding drive electrical signal.

[0016] This invention provides a mine underground pipe installation robot and its control system. It has the following beneficial effects: 1. This invention utilizes a multimodal sensor array, including an infrared thermal imager and a hazardous gas detector, installed in a sensing tower. A central control module employs evidence theory to quantitatively and jointly determine the basic probability allocation functions for infrared temperature and carbon monoxide concentration. This structure integrates environmental parameters from different dimensions, reducing the impact of interference factors such as dust and water vapor in the well on individual sensors and improving the accuracy of the robot's identification of the safety status in the well.

[0017] 2. The emergency execution module of this invention integrates a hardware watchdog circuit, redundant safety relays, and an energy-consuming braking discharge circuit. When the system malfunctions, the watchdog circuit triggers the redundant safety relays to disconnect the main power input circuit, while the energy-consuming braking discharge circuit is activated, converting the back electromotive force of the motor windings of the moving drive mechanism into heat dissipation. This design can quickly cut off power and achieve braking when the control system crashes or a high alarm level is detected, preventing the robot from moving uncontrollably.

[0018] 3. This invention employs an architecture in which a field-programmable gate array (FPGA) and a microprocessor collaboratively process data in the central control module. The FPGA is specifically responsible for receiving laser point cloud data and executing voxel filtering algorithms, directly transferring the denoised data to the microprocessor's system memory address space. This hardware division of labor reduces the computational load on the microprocessor, improves the system's processing speed for large-scale spatial data, and ensures the real-time performance of robot motion planning and emergency control command generation. Attached Figure Description

[0019] Figure 1 This is a perspective view of the present invention; Figure 2 This is a diagram showing the support plate and upper component of the present invention; Figure 3 This is a diagram showing the housing and upper component of the present invention; Figure 4 This is a system architecture diagram of the present invention; Figure 5 This is a line graph showing the point cloud efficiency comparison in high-dust environments according to the present invention. Figure 6 This is a bar chart comparing the disaster hazard identification and anti-interference performance of the present invention. Figure 7 The following is a comparative line graph of the transient response waveform of emergency braking according to the present invention. Subgraph (a) is a line graph of bus current attenuation, and subgraph (b) is a line graph of slope displacement evolution.

[0020] The components include: 1. Robot body; 2. Support plate; 21. Central control module; 211. Mining artificial intelligence algorithm unit; 212. Accident hazard identification model; 213. Emergency linkage logic unit; 3. Shell; 31. Sensing module; 4. Emergency execution module; and 5. Explosion-proof lithium battery power module. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] See attached document Figures 1 to 4 The present invention provides a mine underground pipe installation robot, including a robot body 1, a support plate 2 and a shell 3.

[0023] The support plate 2 is fixed to the upper end of the robot body 1 by explosion-proof bolts. The housing 3 covers the upper end of the support plate 2 and forms a sealed cavity inside. The upper end of the support plate 2 integrates a central control module 21, an emergency execution module 4, and an explosion-proof lithium battery power module 5. The above modules are connected by explosion-proof terminals, and the wiring is placed in explosion-proof cable trays.

[0024] A sensing module 31 is installed on the top exterior of the housing 3. Various environmental sensors inside the sensing module 31 are integrated and mounted on a sensing tower, which adopts a rotatable structure design to expand the acquisition range.

[0025] The explosion-proof lithium battery power module 5 provides the necessary operating voltage to all modules in the system. During operation, the robot's overall system architecture executes action logic according to the set signal flow direction, with the specific steps as follows: The drive mechanism of the robot body 1 receives the driving command and drives the robot to move in the underground tunnel area.

[0026] The sensing module 31, mounted on the top of the housing 3, begins operation. As the sensing tower rotates, the sensing module 31 performs multi-dimensional scanning and sampling of the surrounding environment, acquiring raw sensing data including the three-dimensional structure of the environment, gas concentration, thermal field distribution, and on-site visuals. The acquired data is transmitted in real-time via cable to the central control module 21 within the internal sealed cavity.

[0027] The central control module 21 receives the raw sensing data transmitted by the sensing module 31. The preprocessing circuit of the central control module 21 first performs filtering calculations on various data streams to remove noise signals caused by environmental interference.

[0028] The hardware synchronization chip in the central control module 21 unifies the time base and transforms the spatial coordinate system of the multiple heterogeneous data streams after noise reduction, and outputs a fused data stream with spatiotemporal information alignment.

[0029] The mining artificial intelligence algorithm unit 211 inside the central control module 21 acquires the fused data stream and performs feature extraction calculations. The extracted environmental features are input into the accident hazard identification model 212 for calculation and matching, and the safety status result of the current working condition is output.

[0030] The emergency linkage logic unit 213 within the central control module 21 determines the alarm level based on the safety status results output by the accident hazard identification model 212. When the determination result exceeds the set safety threshold, the emergency linkage logic unit 213 generates a corresponding emergency control command.

[0031] Emergency execution module 4 receives the emergency control command and executes hardware-level actions. The control port of emergency execution module 4 outputs a drive electrical signal to trigger external audible and visual alarm devices for on-site warning and drives the explosion-proof power-off device to cut off the power supply to the equipment at the corresponding work site. Simultaneously, the central control module 21 reports the current data log and alarm information to the ground dispatch center system via the communication module.

[0032] The robot body 1 is equipped with a tracked chassis structure at its bottom, providing ground mobility and supporting the upper equipment. The track components of the tracked chassis structure are made of wear-resistant rubber, and the outer surface of the track is molded with anti-slip texture. The ground clearance of the robot body 1 is set to be greater than 15 cm, and the rotation center height of the idler wheel at the front of the tracked chassis structure is higher than the ground clearance. This structural configuration enables the robot body 1 to maintain traction when encountering geological surfaces such as gravel, mud, and slopes, and meets the operational requirements of climbing angles not exceeding 30 degrees and obstacle-crossing heights not less than 15 cm.

[0033] The robot body 1 houses a motion drive mechanism, which includes a 550-type DC motor with rated parameters of 24V / 3000r / min and an output torque of 0.8N·m, an L298N motor drive module, and an E6B2-CWZ6C incremental motor encoder. The L298N motor drive module is electrically connected between the central control module 21 and the 550-type DC motor, receiving digital control signals and outputting a 24V drive voltage to the motor coils. The output shaft of the 550-type DC motor is connected to the drive wheels of the tracked chassis via a transmission mechanism. The E6B2-CWZ6C incremental motor encoder (with a resolution of 1000P / R) is coaxially mounted at the motor's output end, collecting the motor's rotation information and feeding back speed pulse signals to the central control module 21.

[0034] Assume the output torque of the 550 DC motor is The mechanical reduction ratio of the transmission mechanism is The energy transfer efficiency of the transmission system is The effective radius of the chassis drive wheels is The total driving force output by the tracked chassis structure in motion. Calculate using the following formula: ; in: This refers to the total driving force output by the tracked chassis structure during movement. This refers to the output torque of the 550-type DC motor. This refers to the mechanical reduction ratio of the transmission mechanism; The energy transfer efficiency of the transmission system; This is the effective radius of the chassis drive wheels.

[0035] Let the total mass of the robot be... The gravitational acceleration constant is The robot is tilted at an angle of... When driving on a slope, the component of gravity that creates resistance downhill along the slope. The calculation formula is as follows: ; in: This refers to the total mass of the robot. It is the gravitational acceleration constant; For the robot at an angle of inclination The ramp surface; This represents the downward resistance component caused by gravity along the slope.

[0036] To achieve the maximum climbing angle of the tracked chassis structure The technical requirement is ≤30°, assuming the frictional resistance between the track and the ground is... By selecting the output torque of the 550-type DC motor and the mechanical reduction ratio of the transmission mechanism, the total driving force is made to satisfy the following inequality condition: ; in: This refers to the frictional resistance between the tracks and the ground.

[0037] The closed-loop logic for the movement control of robot body 1 consists of the following steps: The central control module 21 outputs the target running speed command based on the current navigation data.

[0038] The central control module 21 converts the target running speed command into a corresponding pulse width modulation duty cycle digital signal and transmits the digital signal to the control pin of the L298N motor drive module.

[0039] The L298N motor drive module adjusts the on / off time ratio of the power bridge based on the input pulse width modulation duty cycle digital signal, thereby changing the average operating voltage output to the 550 DC motor and adjusting the actual rotation speed.

[0040] The E6B2-CWZ6C incremental motor encoder continuously acquires displacement data as the motor spindle rotates, generates orthogonal phase electrical pulse signals, and sends them back to the input interface of the central control module 21.

[0041] The central control module 21 calculates the number of pulses per unit time to obtain the actual feedback speed, compares the actual feedback speed with the target running speed command, and outputs a corrected pulse width modulation duty cycle digital signal to maintain the stability of the chassis movement speed.

[0042] For the meshing structure of the reduction gear set and the mechanical assembly method of the track tensioning device inside the above-mentioned transmission mechanism, those skilled in the art can select and match the components according to the basic mechanical design specifications. The internal structure assembly is a well-known technology in the field and will not be described in detail here.

[0043] The housing 3 is mounted on the upper end of the support plate 2, forming a sealed cavity between them. This design is suitable for underground environments with relative humidity not exceeding 95% and dust concentration not exceeding 1000 mg / m³. 3 To withstand harsh environments, the overall protection level of the sealed cavity is designed to reach IP68 level, in order to prevent external high humidity water vapor and explosive coal dust from entering the internal electronic circuits.

[0044] The IP68 protection rating is achieved through the following sealing mechanical structure. An annular sealing groove is formed on the upper edge of the support plate 2, within which a high-elasticity silicone O-ring is installed with an interference fit. The bottom edge of the housing 3 is fitted onto this high-elasticity silicone O-ring. The support plate 2 and housing 3 are mechanically secured together by multiple explosion-proof bolts. The tightening of these bolts compresses the high-elasticity silicone O-ring, causing it to elastically deform and fill the gap between their contact surfaces. When the external sensing module enters the sealed cavity via a cable passing through the housing 3, a stainless steel explosion-proof gland connector is fitted at the cable insertion hole to physically lock and seal the cable gap.

[0045] Establish a calculation model for the clamping force of the sealing structure, and set the effective clamping force per unit length required for the high-elasticity silicone O-ring to achieve the IP68 water immersion standard as: The total perimeter of the mating surface between the shell 3 and the support plate 2 is The number of evenly distributed explosion-proof bolts is The axial preload required for a single explosion-proof bolt. The calculation formula is as follows: ; in: The axial preload required for a single explosion-proof bolt; The effective clamping force per unit length required for a high-elasticity silicone O-ring to achieve the IP68 water immersion resistance standard; The number of explosion-proof bolts that are evenly distributed; It is the total perimeter of the mating surface between the shell 3 and the support plate 2.

[0046] The outer surface of the casing 3 is coated with a 50μm thick polytetrafluoroethylene (PTFE) anti-stick coating. This PTFE anti-stick coating has extremely low surface energy characteristics, which is used to resist particle adhesion and water droplet condensation in the high dust and high humidity environment of mines, and to prevent corrosion of the outer surface due to the accumulation of mud and dirt. The coating thickness is set at 50μm. If the thickness is less than this value, the coating is prone to local peeling under the friction of hard coal and rock particles underground, and cannot maintain a long-term anti-stick effect; if the thickness is greater than this value, it will lead to increased internal curing shrinkage stress of the coating, which is prone to microcracks and thus undermines the protective integrity of the overall structure.

[0047] The surface moisture-proof and dust-proof physical mechanism of the polytetrafluoroethylene (PTFE) anti-stick coating is described by Young's equation. Let the contact angle of the liquid droplet in the environment on the surface of the shell 3 be... The surface tension between the solid and the gas on the surface of shell 3 is The surface tension between the solid and the liquid on the surface of shell 3 is The surface tension between the liquid and the gas is The tension balance equations at the three-phase interface are as follows: ; in: The contact angle of the droplets in the environment on the surface of the housing 3; The surface tension between the solid and the gas on the surface of shell 3; The surface tension between the solid and the liquid on the surface of shell 3; The surface tension is the tension between a liquid and a gas.

[0048] Because the 50μm thick PTFE anti-stick coating reduces the surface tension between the solid and gas on the surface of shell 3, the value calculated by substituting into the formula is... The value is negative, thus representing the contact angle of the droplet on the surface of shell 3. The water vapor exhibits a hydrophobicity of over 90°. In high-humidity environments, water vapor cannot spread on the surface of the shell 3; instead, it forms spherical water droplets that slide off under gravity. During the sliding process, it carries away coal dust particles adhering to the surface, thus achieving self-cleaning.

[0049] The manufacturing and surface treatment of housing 3 shall be performed according to the following steps: The outer surface of the aluminum alloy shell substrate 3 is sandblasted and degreased to remove rust, thereby increasing the surface roughness of the substrate.

[0050] The polytetrafluoroethylene powder is uniformly coated onto the outer surface of the housing 3 using an electrostatic spraying process, while controlling the spraying stroke and speed per stroke.

[0051] The coated shell 3 is placed in a high-temperature curing oven for baking, so that the powder melts, flows, and cross-links and cures. After cooling, a 50μm thick polytetrafluoroethylene non-stick coating is formed.

[0052] Regarding the rubber compression sealing structure inside the aforementioned explosion-proof gland and the specific operating parameters of the electrostatic spraying equipment, those skilled in the art can operate them according to the assembly specifications of conventional explosion-proof electrical equipment. The operating methods are well-known in the field and will not be elaborated here.

[0053] A sensing module 31 is located at the upper end of the housing 3. Various sensors within the sensing module 31 are integrated onto a 360° rotatable sensing tower. This 360° rotatable sensing tower is used to expand the data acquisition range of the downhole environment.

[0054] The mechanical and electrical transmission structure at the bottom of the 360° rotatable sensing tower specifically includes a tower base, a slewing bearing, an explosion-proof stepper motor, a gear transmission assembly, an explosion-proof conductive slip ring, and an absolute encoder. The tower base is fixed to the top panel of housing 3. The slewing bearing is embedded inside the tower base, bearing the weight of the upper sensor array and providing rotational freedom. The explosion-proof stepper motor is installed in a sealed cavity, and its output shaft is connected to the inner ring of the slewing bearing via a gear transmission assembly. The absolute encoder is coaxially mounted at the output end of the explosion-proof stepper motor and is used to acquire the real-time mechanical angle of the spindle.

[0055] Because the sensing tower needs to perform continuous 360-degree rotation, an explosion-proof conductive slip ring is coaxially installed at the center of the rotating shaft to prevent cable kinking and breakage during the rotation of the sensor array. The stator end cable of the explosion-proof conductive slip ring passes downward into the sealed cavity and is electrically connected to the central control module 21 and the explosion-proof lithium battery power module 5, while the rotor end cable passes upward and is distributed to the various sensors in the sensing module 31.

[0056] Let the output angular velocity of the explosion-proof stepper motor be... The reduction ratio of the gear transmission set is The rotational angular velocity of the sensing tower. Calculate using the following formula: ; in: The output angular velocity of the explosion-proof stepper motor; This refers to the reduction ratio of the gear transmission assembly. To sense the angular velocity of the tower's rotation.

[0057] In this embodiment, the rotational angular velocity of the sensing tower is maintained at 10° / s by configuring the reduction ratio of the gear transmission group and the motor drive pulse frequency. The reason for setting this value is as follows: If the rotational angular velocity is higher than this value, the point cloud data of a single-circle scan by devices such as lidar will be too sparse in spatial distribution, forming a perception blind spot. If the rotational angular velocity is lower than this value, the cycle of a single environmental scan is prolonged, which cannot meet the real-time response requirements for sudden emergencies downhole, such as abnormal gas outbursts.

[0058] Let the initial scanning angle of the control tower be... The system uptime is The current scanning angle of the control tower at any given time. The calculation formula is as follows: ; in: This is the initial scanning angle of the control tower; This refers to system uptime. This represents the current scanning angle of the control tower at any given time. To sense the angular velocity of the tower's rotation.

[0059] The rotation control and data transmission process of the sensing tower is executed according to the following steps: The central control module 21 sends pulse control signals of a specified frequency to the driver of the explosion-proof stepper motor.

[0060] The explosion-proof stepper motor operates, and the power is reduced and increased in torque through the gear transmission group, which drives the slewing support bearing and the sensor array fixed on it to rotate continuously in the horizontal plane.

[0061] During the rotational motion, the precious metal brush inside the explosion-proof conductive slip ring maintains physical sliding contact with the conductive ring channel, supplies working voltage to the sensor array, and synchronously transmits the collected analog and digital signals back to the central control module 21 inside the sealed cavity.

[0062] The absolute encoder reads the real-time physical position of the mechanical shaft and feeds the position signal back to the central control module 21. The central control module 21 uses this position signal to calculate the corresponding current scanning angle, providing an orientation reference for subsequent data space coordinate transformation.

[0063] For the raceway clearance fit of the aforementioned slewing bearing and the internal stator winding structure of the explosion-proof stepper motor, those skilled in the art can select them according to conventional mechatronics equipment design specifications. Their mechanical and electrical structures are well-known technologies in this field and will not be elaborated here.

[0064] The sensing tower within sensing module 31 is equipped with a multimodal explosion-proof sensor array. This array includes a laser scanning radar, an infrared thermal imager, a visible light camera, a hazardous gas detector, and a ventilation parameter sensor, specifically comprising five major hazard sensors: a gas detector (KJ101), a dust sensor (GCG1000), a water level sensor (SW100), a temperature sensor (WRNK-101), and a stress sensor (GYM-2). It is also equipped with a DHT11 temperature and humidity sensor (5V power supply, accuracy ±2℃ / ±5%) and an E18-D80NK infrared obstacle avoidance sensor (detection distance 3-80cm, meeting Ex ia I Mb explosion-proof rating). All core components meet Ex ia I Mb or Ex d I Mb explosion-proof rating requirements and comply with the "Coal Mine Safety Regulations" (GB3836.1-2021).

[0065] The laser scanning radar specifically adopts the YDLIDAR X4M-9 explosion-proof radar, with a ranging range of 0.1-10m, used to generate point cloud data characterizing the three-dimensional physical structure of the tunnel. An infrared thermal imager and a visible light camera are installed side-by-side. The infrared thermal imager has a resolution of 640×512 pixels, a temperature measurement range of -20℃ to 500℃, and a detection distance of at least 15 meters for abnormal thermal fields indicating impending disasters. The visible light camera uses 1080P resolution to acquire high-definition visible light image video stream data of the underground environment. (The infrared thermal imager and visible light camera are installed side-by-side. The infrared thermal imager has a resolution of 640×512 pixels, a temperature measurement range of -20℃ to 500℃, and a detection distance of at least 15 meters for abnormal thermal fields indicating impending disasters. The visible light camera uses 1080P resolution.)

[0066] The hazardous gas detector uses the KJ101 model, and the ventilation parameter sensor uses the GFW15 model. The hazardous gas detector can simultaneously detect CH4, CO, and O2 gases, with a detection accuracy of no less than 0.01%VOL for methane gas and a data response time ≤2s. The above multi-source data is fused using the DS evidence theory algorithm and input into the accident hazard identification model 212. The misjudgment rate for safety hazards such as underground gas exceeding limits, spontaneous combustion of coal, roof collapse, water inrush, and coal dust accumulation is controlled to ≤1.2%.

[0067] Assume the surface emissivity of the target object is The Stefan-Boltzmann constant is The thermodynamic temperature of the target object is The background ambient temperature is The total radiant energy received by the detector of the infrared thermal imager. The calculation formula is as follows: ; in: This represents the total radiant energy received by the detector of the infrared thermal imager. The emissivity of the target object's surface; It is the Stefan-Boltzmann constant; The thermodynamic temperature of the target object; The background ambient temperature.

[0068] The central control module 21 obtains the thermodynamic temperature matrix of the target object by inversely solving the above equation based on the total radiation energy data received by the detector output by the infrared thermal imager, combined with the preset background ambient temperature and the surface emissivity of the target object.

[0069] The hazardous gas detector and ventilation parameter sensor are located inside the metal louvers on the side of the sensing tower. The hazardous gas detector integrates a methane detection unit based on the non-dispersive infrared principle and carbon monoxide and oxygen detection units based on the electrochemical principle.

[0070] The mechanism for measuring methane concentration using nondispersive infrared spectroscopy is based on Beer-Lambert's law. Let the intensity of the infrared light received by the detector be... The intensity of the incident infrared light is The physical length of the air chamber inside the detection unit is The absorption coefficient of methane gas for infrared light of a specified wavelength is: Concentration of methane gas The calculation formula is as follows: ; in: The intensity of the infrared light received by the detector; Intensity of incident infrared light; To measure the physical length of the air chamber inside the detection unit; is the absorption coefficient of methane gas for infrared light of a specified wavelength; This represents the concentration of methane gas.

[0071] Through logarithmic transformation, the concentration of methane gas is derived. The calculation process is expressed as follows: ; The ventilation parameter sensor uses an ultrasonic wind speed probe to calculate the real-time wind speed and direction in the tunnel by measuring the time difference of ultrasonic wave propagation with and against the wind.

[0072] The data acquisition and transmission of the multimodal explosion-proof sensor array are performed according to the following steps: The internal transmitter of the laser scanning radar emits pulsed lasers at a set frequency, and the receiver captures the echo signal, calculates the spatial distance and reflection intensity, and encapsulates it into a single frame of point cloud data.

[0073] Infrared thermal imagers and visible light cameras simultaneously encode infrared thermal radiation energy data and visible light pixel data within the same spatial field of view into digital video stream signals.

[0074] An ambient airflow carrying mine dust and gases passes through metal louvers and enters the gas chamber of the hazardous gas detector. The methane detection unit continuously calculates and outputs a digital methane gas concentration value based on the attenuation of the absorbed infrared light intensity.

[0075] All digital and analog signals generated by the multimodal explosion-proof sensor array are aggregated through a dedicated cable and transmitted downward through the casing to the central control module in the internal sealed cavity via the explosion-proof conductive slip ring inside the sensing tower.

[0076] The chemical reaction equations for measuring carbon monoxide concentration based on the above electrochemical principle, as well as the wide dynamic range image sensor circuit for visible light cameras, can be obtained by those skilled in the art by consulting relevant instrument design manuals. Their basic principles are well-known technologies in this field and will not be elaborated here.

[0077] The support plate 2 integrates an explosion-proof lithium battery power module 5. The explosion-proof lithium battery power module 5 is internally configured with a mining-grade intrinsically safe power supply and power management system circuit.

[0078] The explosion-proof lithium battery power module 5 uses a lithium iron phosphate battery pack with a rated voltage of 11.1V and a capacity of 10Ah as the energy storage unit. To meet the intrinsically safe explosion-proof requirements of underground coal mines and prevent the generation of electric sparks sufficient to ignite explosive gas mixtures in the event of a short circuit or component failure, an explosion-proof current-limiting circuit and a dual overcurrent protector are connected in series at the output of the lithium iron phosphate battery pack. Specifically, the explosion-proof current-limiting circuit consists of two high-power wire-wound resistors connected in series to form a dual redundant current-limiting structure, and this dual redundant current-limiting structure and the dual overcurrent protector are integrally cast within an epoxy resin insulating curing layer. This structure ensures that even if a single electronic component experiences a breakdown short circuit, the backup component can still maintain its current-limiting function, and the epoxy resin insulating curing layer isolates any potential internal electric sparks from physical contact with external gas.

[0079] Let the open-circuit voltage of the lithium iron phosphate battery pack be The equivalent current-limiting resistance of the explosion-proof current-limiting circuit is The equivalent internal resistance of the line when an external short circuit occurs is The maximum short-circuit current of the circuit under short-circuit conditions. Calculate using the following formula: ; in: This is the open-circuit voltage of the lithium iron phosphate battery pack; This is the equivalent current-limiting resistor for the explosion-proof current-limiting circuit. This is the equivalent internal resistance of the line when an external short circuit occurs. This represents the maximum short-circuit current of the circuit under short-circuit conditions.

[0080] Let the maximum permissible short-circuit current threshold specified in the intrinsically safe explosion-proof standard for underground coal mines be... In the system hardware design, by adjusting the equivalent current-limiting resistor of the explosion-proof current-limiting circuit, the calculated maximum short-circuit current of the circuit under short-circuit conditions is made to meet the following criteria: ; This condition ensures that, under extreme electrical fault conditions, the electrical spark energy released by the system is lower than the minimum ignition energy of the gas.

[0081] The explosion-proof lithium battery power module 5 adopts a 3S2P structure intrinsically safe lithium battery pack for mining, with a rated voltage of 11.1V, a capacity of 10Ah, and supports continuous operation for ≥12 hours. The battery pack is equipped with a BMS battery management system based on DW01 and 8205A chips, providing overcharge (12.6V), over-discharge (9.0V), overcurrent (5A), and short-circuit protection. Its output terminal is connected in series with a 3.9Ω, 50W rated power explosion-proof current-limiting resistor, limiting the ultimate short-circuit current to ≤3A.

[0082] Meanwhile, the BMS battery management system chip is connected to the STM32 microcontroller via I2C hardware interface pins configured as SDA / PB9 and SCL / PB10, which is used to provide real-time feedback of the remaining battery charge (SOC) data to the central control module. The power monitoring error is ≤3%, so as to ensure that the entire system can support power scheduling for continuous operation for ≥12 hours.

[0083] The power management distribution circuit includes a 5V regulated branch, a 3.3V regulated branch, and a 24V boost branch. The 5V regulated branch uses an explosion-proof LDO chip AMS1117-5.0, outputting 5V / 1A with ripple ≤10mV, powering the main controller, sensors, voice module WM8960, and intrinsically safe noise-canceling microphone array for mining applications. The 3.3V regulated branch uses an XC6206P3302 voltage regulator chip, outputting 3.3V / 300mA, powering the 5G and UWB modules. The 24V regulated branch uses an LM2596-24V boost module with a conversion efficiency ≥85%, providing 24V / 2A power to the chassis L298N motor drive, explosion-proof relay module, and speaker amplifier TPA3116D2 (parameters: 24V / 5W), and adapting to an explosion-proof speaker (parameters: 8Ω / 5W, meeting Ex d I Mb explosion-proof rating). Meanwhile, the system is also equipped with an AXP228 intrinsically safe power management chip for mining, which supports dynamic power consumption adjustment and can output stable power of 5V / 0.05A-3A to supply power to various functional modules.

[0084] The energy flow of an intrinsically safe power supply and power management system circuit for mining is performed according to the following steps: The lithium iron phosphate battery pack outputs a basic DC voltage, and the current flows through an explosion-proof current limiting circuit to limit the peak transient power output to subsequent circuits.

[0085] The restricted DC voltage parallel input power management distribution circuit is fed into the input terminals of the LM2596-24V boost module, AMS1117-5.0 and XC6206P3302 voltage regulator chips, respectively.

[0086] Each module and chip adjusts according to its internal feedback loop to output stable 24V, 5V and 3.3V voltages respectively.

[0087] The 5V main circuit supplies the external sensor array, and its bypass current flows into the AMS1117-3.3 low dropout linear regulator. After filtering out high-frequency ripple, a stable 3.3V voltage is generated and sent to the central control module 21.

[0088] For the selection of the peripheral energy storage inductor of the XL6009 boost chip and the design of the cell equalization charging management board inside the lithium iron phosphate battery pack, those skilled in the art can consult general power management chip datasheets. The circuit topology and component matching are well-known technologies in the field and will not be described in detail here.

[0089] A central control module 21 is fixed on the support plate 2. The circuit board inside the central control module 21 is configured with a three-level heterogeneous hardware topology structure, which includes STM32F103C8T6 preprocessing + FPGA such as Xilinx Kintex-7 series or equivalent chips with PCIe hard core synchronous + RK3568 AI judgment.

[0090] The STM32F103C8T6 microcontroller has 64KB of built-in Flash and 20KB of RAM, meeting the Ex ia I Mb explosion-proof standard. Combined with the AD7606 analog-to-digital converter chip (16-bit, 8-channel, 200kSPS sampling rate) as the front-end preprocessing core, it uses the 3σ criterion to remove invalid data and mark suspected anomalies, with a data processing latency of ≤50ms. The microcontroller is also equipped with an AM26LS31 opto-isolation chip with an isolation voltage ≥2500V to ensure safe signal transmission. An FPGA (EP4CE6E22C8) provides a hardware concurrent computing unit to achieve multi-source data timestamp alignment with a synchronization error ≤10ms, and binds spatial location with UWB positioning data, achieving a positioning error ≤10cm. The RK3568 microprocessor has a built-in neural network accelerator (NPU) to acquire fused datasets and run mining-specific artificial intelligence algorithms.

[0091] The mining AI algorithm unit 211 is optimized based on the DeepSeek open-source framework. Specifically designed for high-dust environments underground, the Kalman filter Q-matrix parameter is optimized from the conventional 0.01 to 0.05. This mining AI algorithm unit 211 employs a dust filtering algorithm based on an improved Kalman filter, dynamically adjusting the filter coefficients to handle laser scanning radar applications with dust concentrations ≤1000 mg / m³. 3 The point cloud data in the environment is filtered to remove interference noise. The mining artificial intelligence algorithm unit 211 is trained by importing 15,000 sets of underground working condition data (including high dust, low light, and multiple disaster precursor scenarios) to build a basic model. This improves the efficiency of the processed point cloud data from 75% in the existing technology to over 98%, with autonomous map construction accuracy ≤10cm and obstacle avoidance success rate of no less than 99.5%, effectively supporting unmanned autonomous inspection mode.

[0092] Assume the physical transfer rate of a single channel in a PCIe 3.0 high-speed serial bus configuration is as follows: The number of channels on the PCIe 3.0 high-speed serial bus is The data encoding efficiency of the link layer is The protocol layer payload overhead accounts for a certain percentage. Effective data transmission bandwidth The calculation formula is as follows: ; in: For effective data transmission bandwidth; Single-channel physical transfer rate configured for PCIe 3.0 high-speed serial bus; This refers to the number of channels on the PCIe 3.0 high-speed serial bus. For data encoding efficiency at the link layer; This represents the percentage of effective payload overhead at the protocol layer.

[0093] During system hardware configuration, the number of channels on the PCIe 3.0 high-speed serial bus is set to 2. In the high-dust environment of underground coal mines, the noise echo of the laser scanning radar increases, and combined with the infrared thermal imaging video stream, the peak data traffic can reach 5Gbps to 6Gbps. Using a dual-channel configuration results in a calculated effective data transmission bandwidth of approximately 15.7Gbps, which is greater than the sum of the aforementioned peak data traffic. This condition ensures that multimodal high-concurrency data transmission between chips will not cause channel congestion or data packet loss.

[0094] The physical process of processing multi-source sensing data and control logic in a three-level heterogeneous hardware topology is performed according to the following steps: The STM32 microcontroller continuously reads the feedback speed pulses of the underlying motor and the analog quantity of the ambient gas concentration through the quadrature encoder interface (TIM) of the timer, and packages them into a status data frame containing a timestamp.

[0095] The hardware timer of the field-programmable gate array outputs a global synchronization second pulse signal, which triggers an external laser scanning radar and an infrared thermal imager to perform synchronous exposure and sampling through physical connections.

[0096] The field-programmable gate array receives the input laser point cloud data, calls the internal logic circuit to execute the voxel filtering algorithm to remove stray noise points, and writes the noise-reduced point cloud data and infrared thermal imaging video stream into the internal buffer memory.

[0097] When the Direct Memory Access Controller inside the Field Programmable Gate Array (FPGA) starts working, it directly maps and moves data blocks in the buffer memory to the system memory address space of the RK3568 microprocessor via the PCIe 3.0 high-speed serial bus, without occupying the computing resources of the RK3568 microprocessor.

[0098] The RK3568 microprocessor calls its internal neural network processing unit to perform feature extraction and computation on multimodal data in system memory, and outputs safety status results and motion planning instructions.

[0099] The RK3568 microprocessor generates motion planning instructions and sends them to the STM32 microcontroller via the controller area network bus. The STM32 microcontroller then converts these instructions into electrical signals that drive the underlying hardware to perform the actions.

[0100] For the configuration of the direct memory access controller registers inside the aforementioned field-programmable gate array and the physical layer transceiver circuit connections of the controller area network bus, those skilled in the art can consult the hardware design reference manuals provided by the chip manufacturer. The underlying circuit construction is a well-known technology in this field and will not be described in detail here.

[0101] The circuit board of the central control module 21 is equipped with a mining-specific anti-interference and signal isolation circuit. Frequent start-ups and shutdowns of large electromechanical equipment in underground coal mines generate strong electromagnetic interference. To prevent electromagnetic interference from coupling to the central control module 21 through signal or power lines, causing system crashes, the mining-specific anti-interference and signal isolation circuit includes an optocoupler isolation module, an isolated DC power supply module, and a bus transient suppression module.

[0102] The optocoupler isolation module is connected in series between the STM32 microcontroller and the motor drive module of the moving drive mechanism. Specifically, the optocoupler isolation module uses a 6N137 high-speed optocoupler to convert the electrical control signal output by the STM32 microcontroller into an optical signal and then back into an electrical signal, thus cutting off the physical conductive connection between the control circuit and the drive circuit.

[0103] The isolated DC power supply module is used to independently power the output side of the optocoupler isolation module. Specifically, the isolated DC power supply module uses the B0505S-1W micropower isolation power chip, utilizing the magnetic coupling mechanism of the chip's internal miniature transformer to transfer electrical energy. This design achieves physical isolation between the digital ground inside the STM32 microcontroller and the power ground of external high-power devices, blocking ground loop noise interference.

[0104] The bus transient suppression module is installed at the physical interface of the Controller Area Network (CLAN) bus. The module employs a bidirectional transient voltage suppressor diode connected in parallel between the bus signal line and ground to absorb surge voltages induced by the external environment. Considering the standard operating level of the CLAN bus, the breakdown voltage of this bidirectional transient voltage suppressor diode is set slightly higher than the peak communication voltage, such as selecting an operating voltage of 5V or 24V, to ensure that signal transmission is not affected during normal communication, while instantaneously dissipating energy during surges.

[0105] To further enhance environmental adaptability, the system is equipped with the following anti-interference and calibration circuits at the sensor power supply and signal acquisition terminals: Power supply protection: Each disaster monitoring sensor is assigned an independent power supply branch, and a high-frequency ferrite bead (parameters 100Ω / 100MHz) and a 0.1μF decoupling capacitor are connected in series on the branch to avoid power line crosstalk when each high-frequency acquisition module is working. Signal protection: All analog signal transmission cables of the sensors are double-shielded twisted-pair cables, and the shielding layer is forced to be grounded in a single phase (grounding resistance ≤4Ω) to minimize spatial electromagnetic interference generated by equipment such as downhole frequency converters; Data calibration mechanism: A standard signal calibration interface is reserved on the central control module board to adapt to 4-20mA analog signal input, which is used for periodic connection to the mining standard calibrator. The AI ​​core supports automatic calibration of the measurement range of various sensors through this interface, thereby eliminating the sensor zero-point drift problem caused by long-term operation under harsh conditions, and ensuring the calibration accuracy ≤±0.5%FS throughout the entire life cycle.

[0106] A resistor-capacitor (RC) low-pass filter circuit is connected in series at the front end of the analog signal input channel of the hazardous gas detector to filter out high-frequency harmonic interference generated by the downhole frequency converter. Let the filter resistance of the RC low-pass filter circuit be... The filter capacitor of the RC low-pass filter circuit is The cutoff frequency of an RC low-pass filter circuit. The calculation formula is as follows: ; in: This is the cutoff frequency of the RC low-pass filter circuit; For RC low-pass filter circuit; This is the filter capacitor in a resistor-capacitor low-pass filter circuit.

[0107] During system hardware configuration, the values ​​of the filter resistor and filter capacitor in the RC low-pass filter circuit are adjusted to set the calculated cutoff frequency of the RC low-pass filter circuit to 50Hz. Setting it higher than this frequency will fail to filter out high-frequency noise generated by the downhole power frequency and motor frequency conversion; setting it lower will result in a severe lag in the response to gas concentration change signals. This frequency setting preserves low-frequency analog signals while significantly attenuating high-frequency electromagnetic noise.

[0108] The signal processing procedure of the mining anti-interference and signal isolation circuit is performed according to the following steps: The analog voltage signal output by the hazardous gas detector enters the RC low-pass filter circuit. After being smoothed by capacitor charging and discharging to remove high-frequency glitches, it is input to the analog-to-digital conversion pin of the STM32 microcontroller.

[0109] The digital pulse signal output from the STM32 microcontroller is connected to the input side of the optocoupler isolation module, driving the LED inside the optocoupler to emit light.

[0110] The photosensitive element on the output side of the optocoupler isolation module receives the optical signal and conducts it. The independent power supply provided by the isolated DC power supply module restores the electrical signal and outputs it to the external motor drive module.

[0111] When external devices communicate with the central control module 21 via the controller area network bus, if a transient surge voltage is generated by cable coupling, the bidirectional transient voltage suppression diode will conduct within a nanosecond time to clamp the bus voltage within the safe operating voltage threshold of the transceiver chip.

[0112] For the calculation of the current-limiting resistor on the input side of the aforementioned optocoupler isolation module and the selection of the junction capacitance of the bidirectional transient voltage suppression diode, those skilled in the art can configure them according to conventional electromagnetic compatibility design specifications. The specific parameter matching is a well-known technology in this field and will not be elaborated here.

[0113] The microprocessor inside the central control module 21 runs a lidar point cloud filtering algorithm specifically designed for high-dust conditions underground. In underground coal mine working faces and return airways, high concentrations of suspended coal dust particles in the air cause diffuse reflection and scattering of laser pulses. After receiving the reflected light from the dust, the lidar generates a large number of randomly distributed isolated noise points in three-dimensional space, blurring the features of the roadway walls and causing obstacle avoidance misjudgments.

[0114] Conventional point cloud filtering algorithms, if relying solely on reflection intensity, can easily misclassify black coal and rock entities as dust points; if relying solely on spatial distance calculations, the computational load becomes excessive when processing hundreds of thousands of point cloud data points. The lidar point cloud filtering algorithm designed for high-dust underground working conditions combines reflection intensity thresholds with spatial statistical characteristics. It leverages the weak laser reflection intensity from high-concentration dust and the discrete, disordered distribution of dust particles in space. Intensity pre-screening reduces the computational scale, and spatial statistics are then used to accurately remove noise points.

[0115] Let the set of single-frame point cloud data output by the lidar be... any one of them Spatial coordinates are .point Includes the reflection intensity value output from the hardware layer. Let the threshold for determining reflection intensity be... In underground coal mine environments, the reflection intensity values ​​output by mainstream lidar typically range from 0 to 255. The reflection intensity values ​​of coal and rock walls are mostly distributed between 10 and 30, while suspended dust, due to the scattering and transmission of the laser beam, usually has an echo reflection intensity value below 15. Therefore, the system is configured with a reflection intensity judgment threshold of 15. This value setting can initially eliminate dust points with extremely weak reflections and prevent the accidental deletion of coal and rock wall features.

[0116] Let the parameter for the number of nearest neighbors be... Select point The closest in space Let the nth nearest neighbor be the i-th node. The spatial coordinates of the nearest neighbors are: .point Average spatial distance The calculation formula is as follows: ; Calculate the distance distribution characteristics of all points in the extracted subset to be processed. Let the mean of the global average distance be... The standard deviation of the global average distance is The preset multiplier coefficients are Dynamic truncation distance threshold The calculation formula is as follows: ; in: This is a dynamic truncation distance threshold; This represents the mean of the global average distance. The standard deviation of the global average distance; These are the preset multiplier coefficients.

[0117] In the actual system configuration, the nearest neighbor number parameter is set to 30, balancing the statistical significance of local spatial features with the processor's computational load. The preset multiplier coefficient is set between 1.0 and 2.0. When the underground working face is in a high-dust-concentration operation such as drilling or coal cutting, the preset multiplier coefficient is set to a smaller value, such as 1.0, to enhance filtration intensity; under inspection conditions with lower dust concentrations, the preset multiplier coefficient is set to a larger value, such as 2.0, to retain more environmental details. By adjusting the specific value of the preset multiplier coefficient, the retention ratio of the environmental point cloud is adjusted, so that the calculated dynamic cutoff distance threshold adapts to dusty roadway conditions with different visibility levels.

[0118] The logical processing of the lidar point cloud filtering algorithm for high-dust underground working conditions is performed according to the following steps: The microprocessor receives a frame of raw point cloud data containing spatial coordinates and reflection intensity values, and iterates through each point in the set.

[0119] Compare the reflection intensity values ​​at each point With the threshold for determining reflection intensity .like If the point is determined to be a high-reflectivity solid structure point, it is directly retained; if The point was identified as a suspected dust noise point and added to the subset to be processed.

[0120] A spatial index is created using a KD-tree data structure on the original point cloud dataset. This index is then applied to any point within the subset to be processed. Search the spatial index for the nearest Substitute the nearest neighbor points into the formula to calculate the point. Average spatial distance .

[0121] The distance distribution characteristics of all points in the subset to be processed are statistically analyzed, and the mean of the global average distance is calculated. Standard deviation of distance from global mean Then, substitute the values ​​into the formula to calculate the dynamic cutoff distance threshold. .

[0122] Each suspected dust noise point Average spatial distance With dynamic cutoff distance threshold Perform a comparison. If... If the point is confirmed to be discrete suspended dust noise, it will be removed from memory; Points identified as low-reflectivity coal and rock wall edge points are retained to complete the denoising and cleaning of single-frame point cloud data.

[0123] For the spatial partitioning principle of the aforementioned KD-tree data structure and the numerical statistical algorithm for calculating the standard deviation of the global average distance, those skilled in the art can consult the design documents of general open-source 3D point cloud processing libraries. The underlying mathematical logic is a well-known technology in this field and will not be elaborated here.

[0124] The RK3568 microprocessor inside the central control module 21 is equipped with a multi-source disaster precursor fusion identification model based on DS evidence theory. Single sensors are prone to false alarms in complex downhole environments. For example, mechanical friction overheating caused by insufficient lubrication in conveyor belt idler roller bearings can lead to abnormally high local ambient temperatures. If relying solely on temperature data from an infrared thermal imager, the system could easily misjudge this as a potential coal spontaneous combustion hazard. Since coal spontaneous combustion inevitably involves the release of carbon monoxide gas along with the temperature rise, while simple metal mechanical friction overheating typically does not produce carbon monoxide, the system performs joint calculations using temperature data and gas concentration data to eliminate the ambiguity of a single information source.

[0125] The framework set for identifying system environment states is defined as follows: .in, Indicates normal operating status. This indicates a potential risk of spontaneous combustion of coal. This indicates the overheating state caused by mechanical friction in the equipment. The microprocessor, based on a preset membership function, converts the maximum temperature value extracted by the infrared thermal imager from the current field of view into a basic probability allocation function for infrared temperature of each state within the identification frame set. Simultaneously, it converts the carbon monoxide concentration value output by the hazardous gas detector into a basic probability allocation function for carbon monoxide concentration of each state. This preset membership function specifically adopts a semi-trapezoidal distribution function shape.

[0126] Let the basic probability assignment function for infrared temperature be: The basic probability assignment function for carbon monoxide concentration is: Let the first subset of the identification frame set be... The second subset of the identification frame set is Conflict of evidence coefficient The calculation formula is as follows: ; Let the target subset of the identification frame set be The fused basic probability allocation function The calculation formula is as follows: ; In the specific probability mapping logic, when the temperature is 30℃ and the carbon monoxide concentration is 0ppm, both the infrared temperature basic probability allocation function and the carbon monoxide concentration basic probability allocation function calculated based on the semi-trapezoidal distribution function allocate the higher probability value to... When the temperature is 90℃ and the carbon monoxide concentration is 0ppm, the infrared temperature fundamental probability assignment function assigns probabilities to... and However, the fundamental probability assignment function for carbon monoxide concentration does not support this. After fusion calculation, the fused basic probability allocation function is in The maximum value is achieved at 90℃ and a carbon monoxide concentration of 50ppm. Both reach their maximum values ​​at this temperature. The probability allocations on the above produce a positive superposition, and the fused basic probability allocation function is in The value on the graph approaches 1, thus accurately identifying the precursors to spontaneous combustion of coal.

[0127] The operation of the multi-source disaster precursor fusion identification model based on DS evidence theory is carried out in the following steps: The microprocessor synchronously reads the pre-processed temperature distribution matrix of the infrared thermal imager and the real-time carbon monoxide concentration value of the hazardous gas detector through the system's internal bus.

[0128] The microprocessor extracts the local highest temperature peak from the temperature distribution matrix, calls the nonlinear membership function curve pre-stored in the system memory, and calculates the basic probability assignment function of infrared temperature and the basic probability assignment function of carbon monoxide concentration at that moment.

[0129] The microprocessor substitutes the obtained function value into the formula to perform orthogonal summation and calculates the evidence conflict coefficient between the two sets of sensor data, which is used to measure the degree of contradiction between the states indicated by different physical quantities.

[0130] Using the DS fusion rule formula, the basic probability allocation function after fusion corresponding to all possible system environment states within the identification framework set is calculated.

[0131] The microprocessor compares the final probability values ​​of various environmental states. When the probability value of an abnormal state exceeds the set decision threshold (preset to 0.85), it outputs the corresponding disaster warning level code and linkage control command to the external control terminal.

[0132] For the specific fitting method of the above-mentioned nonlinear membership function curve and the extended derivation of DS evidence theory when there are more than two sensor data sources, those skilled in the art can consult the relevant data fusion algorithm manual. The mathematical basis calculation is a well-known technology in this field and will not be elaborated here.

[0133] The microprocessor inside the central control module 21 operates a two-level hazard threshold judgment logic model. During underground coal mine inspections, when environmental parameters change abnormally, the system needs to execute emergency actions to ensure the safety of equipment and personnel. The two-level hazard threshold judgment logic model quantifies environmental parameters into risk indices and maps them to corresponding hardware linkage control loops, enabling the chassis control and alarm modules to respond and output.

[0134] Define the real-time ambient temperature value extracted by the microprocessor. The upper limit of the safe temperature range is Real-time harmful gas concentration value The upper limit of the gas safety range is The temperature risk weighting coefficient is set as follows: The gas risk weighting coefficient is Comprehensive Environmental Risk Index The calculation formula is as follows: ; in: It is a comprehensive environmental risk index; Temperature risk weighting coefficient; This refers to the gas risk weighting coefficient; The real-time ambient temperature value extracted by the microprocessor; This is the upper limit of the safe temperature range; This is the upper limit of the safe range for gas; This represents the real-time concentration of harmful gases.

[0135] In the system configuration, the first-level warning threshold is set to... The second-level power outage threshold is In the configuration of the underground environment in coal mines, according to the "Coal Mine Safety Regulations," the upper limit of the safe temperature range is set at 40℃, and the upper limit of the safe gas range, taking carbon monoxide as an example, is set at 24ppm. Based on a balanced consideration of thermodynamic and toxic gas risks, both the temperature risk weighting coefficient and the gas risk weighting coefficient are set to 0.5. The first-level warning threshold is set at 0.6, and the second-level power outage threshold is set at 0.9. To avoid frequent relay operation caused by numerical fluctuations, the microprocessor introduces hysteresis comparator logic to determine the comprehensive environmental risk index, setting the hysteresis value to 0.05. That is, when the comprehensive environmental risk index exceeds the threshold and triggers an action, it must fall back to below the threshold minus 0.05 before the corresponding emergency action state is released.

[0136] When the calculated comprehensive environmental risk index is greater than or equal to the Level 1 warning threshold and less than the Level 2 power outage threshold, the system is determined to be in an initial abnormal state. At this time, the Level 1 emergency linkage logic is triggered. The microprocessor sends a speed limit command to the motor driver of the moving drive mechanism through the underlying bus, reducing the chassis running speed to 30% of the rated operating speed. At the same time, the low-frequency control pin of the audible and visual alarm is activated, causing the external explosion-proof warning light to flash at a frequency of 1Hz.

[0137] The emergency linkage logic unit 213 is equipped with preset two-level hazard thresholds. When a single-point parameter exceeds the standard, such as a gas concentration ≥ 0.8% VOL, the first-level warning logic is triggered. The system controls the audible and visual alarm to issue a local yellow audible and visual alarm with a sound pressure level of not less than 100dB, and uploads the warning information to the ground via the 5G network with a transmission delay of ≤ 100ms.

[0138] When multiple parameters exceed the standard, such as gas concentration ≥1.2%VOL and wind speed <0.2m / s, a secondary early warning logic is triggered. This early warning trigger control circuit is equipped with 5 relays corresponding to 5 types of disaster early warnings. The trigger level is output from the GPIO0-GPIO4 pins of the RK3568 microprocessor, which adopts a quad-core Cortex-A55 architecture and NPU computing power of 0.8TOPS. After isolation voltage ≥2500V by the single-channel TLP521 opto-isolation chip, the intrinsically safe G6K-2P-Y relay module for mining is driven to close, with a pull-in current ≤50mA. The relay contacts trigger a red audible and visual alarm (specifically, a BBJ-36V audible and visual alarm, meeting the Ex dI Mb explosion-proof rating), and drive the external KBD-127 explosion-proof power-off device (parameters: 127V / 5A, Ex dI Mb explosion-proof rating) to directly cut off the power supply to the equipment in the area. At the same time, the spray dust suppression device is activated, and the robot is controlled to autonomously evacuate (evacuation speed ≥ 0.5m / s). An emergency alarm is sent to the ground dispatch center, realizing automated linkage from early warning to disposal, with an overall emergency response time of ≤ 10 seconds.

[0139] The two-level hazard threshold determination logic model processes data and integrates with hardware in the following steps: The microprocessor acquires real-time ambient temperature and real-time hazardous gas concentration values ​​from each sensor node at a set sampling period, and substitutes them into the calculation formula to obtain the current comprehensive environmental risk index.

[0140] The microprocessor compares the comprehensive environmental risk index for the current period with the first-level warning threshold and the second-level power outage threshold stored in memory.

[0141] If the comprehensive environmental risk index is greater than or equal to the first-level warning threshold and less than the second-level power outage threshold, the microprocessor executes the software-level speed limiting scheduling program and outputs a low-frequency pulse signal to the audible and visual alarm.

[0142] If the comprehensive environmental risk index is greater than or equal to the level 2 power outage threshold, the microprocessor bypasses the operating system's regular task scheduling, directly pulls the level of the general-purpose input / output pins high to drive the external safety relay to disconnect the power supply, and triggers the network controller's emergency message transmission interrupt task.

[0143] For the specific anti-jitter implementation of the aforementioned hysteresis comparator logic at the code level and the network layer protocol encapsulation of the UDP emergency broadcast message, those skilled in the art can refer to the embedded system programming specifications. The basic software development is a well-known technology in this field and will not be elaborated here.

[0144] Support plate 2 is equipped with a hardware-level, second-level emergency execution module. Under extreme disaster conditions in underground coal mines, the software operating system may fall into an infinite loop due to strong electromagnetic interference or core process crashes. When the system experiences a serious failure or receives a high-risk environment alarm, the hardware-level, second-level emergency execution module does not rely on the execution of software code, but directly cuts off the power supply at the physical level and achieves mechanical braking.

[0145] The hardware-level, second-level emergency execution module includes a hardware watchdog circuit, redundant safety relays, and an energy-consuming braking and discharge circuit.

[0146] The hardware watchdog circuit specifically uses the MAX706 independent monitoring chip to monitor the operating status of the microprocessor. If the microprocessor fails to send a feed pulse signal through a general-purpose input / output pin within the preset timeout period (set to 1.6 seconds), the hardware watchdog circuit directly outputs a hardware reset signal to the microprocessor and simultaneously outputs a low-level signal to trigger the redundant safety relay.

[0147] The redundant safety relays employ a dual-channel series relay group with forced-guided contacts. This physical structure ensures that even if one relay's contacts fail due to welding, the other relay's contacts can still reliably disconnect the power bus of the moving drive mechanism, meeting the single-fault tolerance requirements of intrinsically safe equipment.

[0148] The energy-saving braking discharge circuit is connected in parallel across the motor busbar of the mobile drive mechanism, and consists of a high-power wire-wound discharge resistor connected in series with an insulated-gate bipolar transistor (IGBT). At the instant the power supply is cut off, the IGBT conducts, converting the rotor's inertial kinetic energy and the winding's back electromotive force into heat energy for rapid dissipation. This prevents the chassis from slipping in sloping tunnels and suppresses potential electrical sparks generated by voltage overshoot on the busbar.

[0149] Let the equivalent inductance of the motor winding of the moving drive mechanism be... The total equivalent discharge resistance of the energy-consuming braking discharge circuit is The initial operating current at the moment of power failure is The set safe current threshold is Emergency braking time The calculation formula is as follows: ; in: For emergency braking time; The equivalent inductance of the motor windings for the motion drive mechanism; The total equivalent discharge resistance of the energy-consuming braking discharge circuit; This is the initial operating current at the moment of power failure; The set safe current threshold.

[0150] In the system hardware design, taking an equivalent inductance of 2mH for the motor windings, an initial operating current of 10A, and a safe current threshold of 0.1A as an example, a high-power wire-wound bleeder resistor with a resistance of 5Ω and a rated power of 100W is selected as the total equivalent bleeder resistor, ensuring that the calculated emergency braking time is less than 0.5 seconds. This parameter setting ensures that when the robot detects signs of an impending disaster or experiences a system crash, it completes the entire process from physical power failure to complete mechanical stop, avoiding secondary collisions caused by inertia.

[0151] The physical action process of the hardware-level, second-level emergency execution module is performed according to the following steps: The logic gate circuits of the hardware-level, second-level emergency execution module simultaneously monitor the secondary power-off level signal issued by the microprocessor and the timeout fault signal output by the hardware watchdog circuit.

[0152] When the logic gate circuit receives any kind of trigger signal, it immediately cuts off the coil drive power supply of the redundant safety relay, uses the mechanical spring force to release the forced guide contact, and disconnects the 24V power input main circuit of the moving drive mechanism.

[0153] The moment the power input is disconnected, the detection terminal of the energy consumption braking discharge circuit detects the bus voltage drop and immediately outputs a high-level signal to drive the insulated gate bipolar transistor to conduct.

[0154] The motor winding current is forced into a high-power wire-wound bleeder resistor for rapid dissipation, causing the chassis to stop moving quickly during the emergency braking time. After braking is completed, the motor rotor is in an electromagnetic lock-up state.

[0155] For the gate drive isolation circuit design of the above-mentioned insulated gate bipolar transistor and the heat dissipation layout structure of the high-power wire-wound bleeder resistor, those skilled in the art can refer to the hardware design specifications of general motor controllers. Its power loop topology is a well-known technology in the field and will not be described in detail here.

[0156] The central control module 21 is equipped with a 5G communication module and a UWB positioning module on its internal circuit board. Specifically, the 5G communication module uses the MH5000-31 mining-grade 5G RF module, meeting the Ex ia I Mb explosion-proof requirements and operating in the Sub-6GHz band. The UWB positioning module uses the DWM1000 module, also meeting the Ex ia I Mb explosion-proof requirements. Together with the spatial binding algorithm of the aforementioned FPGA chip, they ensure that the bidirectional interactive latency for uploading environmental data, images, videos, and inspection reports is ≤100ms. The underground environment of coal mines lacks satellite positioning signals, and long-distance roadways have communication blind spots caused by metal supports and nonlinear space. The 5G communication module and the UWB positioning module work collaboratively to meet the needs of high-capacity data backhaul and spatial positioning.

[0157] The 5G communication module specifically adopts a mining 5G communication terminal with an integrated Sub-6GHz band baseband processor. The Sub-6GHz band has strong electromagnetic wave diffraction capabilities, which are suitable for the physical space underground with many bends and large coal mining machines that block the view, and is used to establish uplink high-bandwidth data links and downlink low-latency control links.

[0158] The UWB positioning module specifically uses the DW1000 ultra-wideband radio frequency chip, configured to operate in the 6.5GHz to 8GHz frequency range. This frequency band is physically isolated from the operating frequency band of the 5G communication module, avoiding increased bit error rate due to co-channel interference. The UWB positioning module uses an asymmetric bilateral bidirectional ranging algorithm to communicate with a positioning base station pre-deployed at the top of the tunnel to obtain high-precision distance parameters. The microprocessor embeds the spatial coordinate information calculated by the UWB positioning module into the header of the data frame transmitted by the 5G communication module, ensuring that every frame of video stream and point cloud data transmitted back to the ground via the 5G network carries a precise location tag.

[0159] Let the UWB ranging distance between the UWB positioning module and the positioning base station be... The speed of radio wave propagation is In the asymmetric bilateral two-way ranging interaction process, let the time for the first polling of the tag to be sent to the receiver be... The base station's first response processing time is The second round of polling by the base station was sent to the receiving end at a time when... The second response processing time for the tag is UWB ranging distance The calculation formula is as follows: ; in: The UWB ranging distance between the UWB positioning module and the positioning base station; The tag is sent to the receiving time in the first round of polling; This is the first response processing time for the base station; The time for the base station to send the second round of polling to the receiving time; This refers to the second response processing time for the tag.

[0160] During system hardware configuration, the microprocessor calls a high-frequency hardware timer within the UWB positioning module to record the aforementioned timestamp data. The timer's resolution is set to the 15.65 picosecond level. By substituting the above asymmetric bilateral bidirectional formula into the calculations, the accumulated ranging error caused by crystal oscillator clock frequency drift between the UWB positioning module and the external positioning base station is offset, controlling the physical error of the calculated UWB ranging distance to within 10 centimeters. This provides a precise spatial basis for subsequent environmental modeling and disaster point location.

[0161] The operation of the 5G and UWB collaborative data interaction and positioning mechanism is carried out in the following steps: The UWB positioning module broadcasts polling ranging pulse messages with an initial timestamp to the positioning base station in the alley via its radio frequency antenna, and writes the message transmission time into its internal register.

[0162] After receiving the polling ranging pulse message, the positioning base station returns a response message to the UWB positioning module after a fixed processing delay at the hardware level. The UWB positioning module receives and parses the message to complete two rounds of complete timestamp interaction.

[0163] The microprocessor reads the tag's first polling transmission time, base station's first response processing time, base station's second polling transmission time, and tag's second response processing time from the UWB positioning module register through the serial peripheral interface, substitutes them into the calculation formula to obtain the current UWB ranging distance, and solves the device's three-dimensional spatial coordinates by calling the polygonal positioning algorithm.

[0164] The microprocessor aligns the obtained 3D spatial coordinates with the infrared video stream, lidar point cloud data, and concentration data from the hazardous gas detector in the current acquisition cycle, and packages them into the direct memory access transmission buffer of the 5G communication module according to a preset custom communication protocol frame format. This preset custom communication protocol frame format specifically includes a 2-byte synchronization header, a 12-byte 3D spatial coordinate segment, a variable-length sensor data payload segment, and a 2-byte cyclic redundancy check tail.

[0165] The 5G communication module extracts data packets from the buffer, performs digital modulation and power amplification through the internal radio frequency front end, and then sends them to the nearby underground 5G micro base station. The data packets are then transmitted to the server in the ground dispatch center via the mine's fiber optic backbone ring network.

[0166] For the solution process of the geometric equations in three-dimensional space of the above-mentioned multilateral positioning algorithm, as well as the channel coding and resource block mapping configuration of the underlying physical layer of the 5G communication module, those skilled in the art can consult the wireless sensor network positioning theory and 3GPP standard protocol documents. The mathematical model and wireless radio frequency transmission principle are well-known technologies in this field and will not be elaborated here.

[0167] Specific application examples: This specific application example uses the daily operation of a fully mechanized tunneling face in a high-gas coal mine as an example. During the automatic inspection task, the robot body 1 moves in the roadway at a preset speed. At this time, the coal cutter at the working face generates a large amount of suspended dust, and the local dust concentration reaches 800 mg / m³. 3 The sensing tower atop the robot rotates and scans at an angular velocity of 10° / s. The laser scanning radar receives strong scattered echoes from the high-concentration dust, generating a large amount of isolated noise data. After receiving the raw point cloud data, the field-programmable gate array in the central control module calls a pre-configured point cloud filtering algorithm. By judging the reflection intensity threshold of 15 and calculating the distance to spatial nearest neighbors, the system removes discrete suspended dust noise, preserving the effective point cloud contours of the real tunnel rock wall and the trackless rubber-wheeled vehicle parked ahead, guiding the chassis to successfully avoid obstacles.

[0168] As the robot continued its journey to the tail end of the conveyor belt, the infrared thermal imager detected an abnormal rise in the local ambient temperature to 85°C, while the hazardous gas detector detected a gradual increase in carbon monoxide concentration to 45 ppm. The microprocessor within the central control module ran a multi-source disaster precursor fusion identification model based on DS evidence theory. If relying solely on the single temperature rise data, conventional judgment logic might identify it as mechanical friction heating from the conveyor belt idler roller bearings. This system jointly calculated the temperature data with the basic probability allocation function of the carbon monoxide concentration, eliminating the possibility of simple metal mechanical friction, and calculated the probability of a coal spontaneous combustion hazard to be 0.92. Based on this, the microprocessor determined that the comprehensive environmental risk index exceeded the level two power outage threshold of 0.9.

[0169] Upon triggering the threshold, the central control module 21 directly raises its output signal to the hardware-level, second-level emergency execution module. The redundant safety relay instantly cuts off the main power input circuit of the mobile drive mechanism, and the energy-consuming braking discharge circuit is activated, converting the back electromotive force of the motor windings into heat dissipation within 0.5 seconds. The robot remains firmly stationary on the sloped tunnel floor without losing control and sliding downhill. Simultaneously, the 5G communication module extracts the three-dimensional spatial coordinates calculated by the UWB positioning module, packages them with the disaster alarm message, and sends them to the ground dispatch center's server within 80 milliseconds, completing the early detection and reporting of the hazard.

[0170] Experimental verification and effect comparison: Test 1: Inside the simulated mine environment chamber: Standard coal dust was gradually introduced using a dust generator, and the ambient concentration was calibrated using a dust concentration detector. The dust concentration test gradient was set to 200 mg / m³. 3 400mg / m 3 600mg / m 3 800mg / m 3 and 1000mg / m 3 The system runs continuously for 10 minutes at each concentration gradient, extracting single-frame echo point cloud data from the lidar of both traditional mining robots and the robot of this invention. After removing invalid discrete noise points, the ratio of effective entity reflection points to total data points is calculated to obtain the point cloud efficiency.

[0171] Table 1. Comparison Test Data of Point Cloud Efficiency in High-Dust Environments

[0172] in conclusion: Combined with Table 1 and Appendix Figure 5 The line graph comparing point cloud efficiency in high-dust environments is shown. The horizontal axis represents dust concentration, and the vertical axis represents point cloud efficiency. The curve characteristics clearly show the comparison results: for traditional robots marked with gray dashed lines and hollow circles, the point cloud efficiency exhibits a steep downward trend as the dust concentration in the environment increases, reaching a maximum of 1000 mg / m³. 3 The efficiency had dropped to 72.0%, but the fitting curve of the device of this invention, marked with a black solid line and a black solid square, remained at a stable high level. Even in an extremely dusty environment, the efficiency was still close to the 100% baseline (98.2%), which directly proves that the point cloud filtering algorithm of this invention has extremely strong robustness against dust interference.

[0173] Test 2: Comparison of accuracy rates in disaster hazard identification: Two sets of test heat sources were set up in the experimental area. The first set simulated the overheating from roller friction, using a heating resistor to stabilize the metal surface temperature at 80°C, while maintaining a surrounding carbon monoxide concentration of 0 ppm. The second set simulated the spontaneous combustion source of coal seam oxidation, raising the coal sample surface temperature to 80°C and using a gas generator to maintain a surrounding carbon monoxide concentration of 50 ppm. Each set of equipment conducted 50 independent proximity identification tests on the two heat sources. The final hazard assessment results output by the system were recorded, and the accuracy rate of identifying real spontaneous combustion sources and the false alarm rate of mechanical heat sources were statistically analyzed.

[0174] Table 2. Comparison Test Data on Accuracy of Disaster Hazard Identification

[0175] in conclusion: Combined with Table 2 and Appendix Figure 6 The performance comparison bar chart for disaster identification is shown below. Dark gray bars represent the accuracy rate of identifying genuine spontaneous combustion sources, while light gray bars represent the false alarm rate of mechanical heat sources. This bar chart reveals the performance of both methods under multi-source interference: the traditional robot's false alarm rate for the light gray bars remains high at 64%, while the false alarm rate for the light gray bars of this invention completely disappears, dropping to 0%, while its accuracy rate for the dark gray bars reaches as high as 98%. This fully demonstrates that the device of this invention achieves neither missed nor false alarms in disaster assessment, exhibiting significant advantages in anti-interference performance.

[0176] Test 3: To compare emergency braking response performance, a simulated track surface with a 20-degree inclination was constructed in the laboratory. Two sets of equipment traveled downhill on the slope at rated speed. The system received a secondary power-off command triggered by an external force. A digital storage oscilloscope was used to synchronously record the moment the trigger command was issued, the bus voltage drop waveform, and the current discharge waveform. A high-frame-rate camera was installed on the side of the track to measure the physical displacement of the chassis during the period from the issuance of the command to complete stillness.

[0177] Table 3. Comparative Test Data of Emergency Braking Response Performance

[0178] in conclusion: Combined with Table 3 and Appendix Figure 7 The emergency braking performance comparison line graph is shown. The graph contains two sub-graphs (a) and (b) with the same time axis (horizontal axis). In the graph, the gray dashed line represents traditional software braking, and the pure black solid line represents the hardware-level braking of the present invention.

[0179] Sub-figure (a) shows the bus current decay line graph, indicating that after the system receives the command, the bus discharge current of the black solid line in this invention decreases exponentially and is completely decayed to the safety threshold at 0.42s marked by the auxiliary dashed line, thus completing hardware braking. In contrast, the current decay process of the gray dashed line in the traditional robot is extremely slow.

[0180] Sub-figure (b) shows the slope displacement evolution broken line diagram, which further confirms that due to the physical constraints of rapid current discharge and electromagnetic locking, the transient slope displacement of the black solid line in this invention is strictly suppressed below the auxiliary dashed line of the 0.04m safe displacement upper limit in the figure and reaches a steady state. In contrast, the gray dashed line of the traditional equipment shows that due to the software response delay, its displacement continues to rise over time, eventually producing a dangerous slip of up to 0.85m.

Claims

1. A mine underground pipe installation robot, characterized in that, include: Robot body (1), support plate (2) and shell (3); The support plate (2) is fixed to the upper end of the robot body (1); The housing (3) covers the upper end of the support plate (2) and forms a sealed cavity inside; The upper end of the support plate (2) is equipped with a central control module (21), an emergency execution module (4) and an explosion-proof lithium battery power module (5). A sensing module (31) is provided on the top of the housing (3). The sensing module (31) is communicatively connected to the central control module (21) and is used to collect external environmental information and send it to the central control module (21). The central control module (21) is electrically connected to the emergency execution module (4) and is used to perform calculations and judgments on the external environment information, and issue control commands to the emergency execution module (4) based on the judgment results.

2. The mine underground pipe installation robot according to claim 1, characterized in that, The bottom of the robot body (1) is provided with a tracked chassis structure; The robot body (1) is equipped with a mobile drive mechanism inside; The mobile drive mechanism includes a DC motor, a motor drive module, and an incremental motor encoder. The motor drive module receives the digital control signal from the central control module (21) and outputs a drive voltage to the DC motor; The output shaft of the DC motor is connected to the drive wheel of the tracked chassis structure via a transmission mechanism. The incremental motor encoder is coaxially mounted on the output end of the DC motor, collects the rotation information of the DC motor, and feeds back the speed pulse signal to the central control module (21).

3. The mine underground pipe installation robot according to claim 1, characterized in that, The upper edge of the support plate (2) is provided with an annular sealing groove; A high-elastic silicone O-ring is installed inside the annular sealing groove; The bottom edge of the housing (3) is attached to the high-elastic silicone O-ring; The support plate (2) and the shell (3) are mechanically fastened together by explosion-proof bolts; The outer surface of the housing (3) is coated with a polytetrafluoroethylene anti-stick coating; The sensing module (31) enters the sealed cavity through the housing (3) via a cable, and an explosion-proof gland connector is installed at the cable hole.

4. The mine underground pipe installation robot according to claim 1, characterized in that, The sensing module (31) is mounted on a rotatable sensing tower. The bottom of the sensing tower is equipped with a tower base, a slewing support bearing, an explosion-proof stepper motor, a gear transmission assembly, an explosion-proof conductive slip ring, and an absolute encoder. The tower base is fixed to the top panel of the housing (3); The explosion-proof stepper motor is installed inside the sealed cavity; The output shaft of the explosion-proof stepper motor is connected to the rotary support bearing through the gear transmission assembly. The absolute encoder is coaxially mounted on the output end of the explosion-proof stepper motor; The explosion-proof conductive slip ring is coaxially mounted at the rotation axis of the sensing tower.

5. A mine underground pipe installation robot according to claim 4, characterized in that, A multimodal explosion-proof sensor array is installed on the sensing tower. The multimodal explosion-proof sensor array includes an infrared thermal imager, a hazardous gas detector, and a laser scanning radar; The infrared thermal imager extracts the maximum temperature value of the current field of view and converts it into a basic probability allocation function for infrared temperature. The harmful gas detector outputs a carbon monoxide concentration value and converts it into a basic probability distribution function for carbon monoxide concentration; The accident hazard identification model (212) inside the central control module (21) integrates the infrared temperature basic probability allocation function and the carbon monoxide concentration basic probability allocation function to output the safety status result.

6. A mine underground pipe installation robot according to claim 1, characterized in that, The explosion-proof lithium battery power module (5) is equipped with a lithium iron phosphate battery pack. The output terminal of the lithium iron phosphate battery pack is connected in series with an explosion-proof current limiting circuit and a dual overcurrent protector. The explosion-proof current limiting circuit and the dual overcurrent protector are integrally cast into an epoxy resin insulating curing layer. The explosion-proof lithium battery power module (5) is connected to a power management and distribution circuit; The power management distribution circuit includes a voltage regulation branch and a voltage boosting branch, which output operating voltage to the central control module (21) and the sensing module (31), respectively.

7. A mine underground pipe installation robot according to claim 5, characterized in that, The central control module (21) is equipped with a microcontroller, a field-programmable gate array and a microprocessor; The microcontroller reads the carbon monoxide concentration value and packages it into a status data frame; The field-programmable gate array receives laser point cloud data collected by the laser scanning radar, executes a voxel filtering algorithm to remove noise points, and moves the noise-reduced point cloud data to the system memory address space of the microprocessor. The microprocessor performs feature extraction operations on the multimodal data in the system memory address space and outputs the safety status result and motion planning instructions. The microprocessor sends the motion planning instructions to the microcontroller.

8. A mine underground pipe installation robot according to claim 7, characterized in that, The emergency execution module (4) includes a hardware watchdog circuit, redundant safety relays, and energy consumption braking discharge circuit. The hardware watchdog circuit monitors the operating status of the microprocessor and outputs a low-level signal to trigger the redundant safety relay to operate. The redundant safety relay disconnects the main power input circuit; The energy consumption braking discharge circuit is connected in parallel to both ends of the busbar of the motor of the moving drive mechanism. The energy-consuming braking discharge circuit is activated the instant the main power input circuit is disconnected, converting the back electromotive force of the winding into heat energy dissipation.

9. A mine underground pipe installation robot according to claim 7, characterized in that, The circuit board inside the central control module (21) is equipped with a 5G communication module and a UWB positioning module. The UWB positioning module communicates with the positioning base station to obtain distance parameters; The microprocessor acquires the distance parameter and calculates the spatial coordinate information, and embeds the spatial coordinate information into the header of the data frame transmitted by the 5G communication module; The 5G communication module extracts data frames containing spatial coordinate information, performs digital modulation and power amplification via the radio frequency front end, and sends them to the underground micro base station.

10. A control system for a mine underground pipe installation robot, characterized in that, The mine underground pipe-laying robot according to any one of claims 1-9 comprises: The sensing module (31) is used to acquire raw sensing data from the mine and transmit the raw sensing data to the central control module (21). The central control module (21) is used to receive the raw sensing data, perform noise reduction processing and spatial coordinate system transformation, and output a fused data stream; The central control module (21) includes a mining artificial intelligence algorithm unit (211) and an emergency linkage logic unit (213). The mining artificial intelligence algorithm unit (211) is used to acquire the fused data stream and perform feature extraction operations, and output the safety status result; The emergency linkage logic unit (213) is used to determine the alarm level based on the mine safety status result and generate corresponding emergency control instructions. Emergency execution module (4) is used to receive the emergency control command and output the corresponding drive electrical signal.

Images