Rail transport remote control and safety monitoring system

Abstract

This invention discloses a remote control and safety monitoring system for a rail transport machine. Two identical driven wheels are mounted on the bottom of the frame below the cargo box to assist the transport machine's movement. Fuel is supplied to the engine from a fuel tank, and the electricity generated by the engine is transmitted to the power unit, enabling the power unit to travel on the track. The power unit drives the frame, which in turn moves the cargo in the cargo box, thus transporting the goods. Front radar, front camera, rear camera, and rear radar transmit the collected data to a control box. The control box fuses this information to identify obstacles, allowing the transport machine to brake in time to avoid accidents. This data is then transmitted to a monitoring center via 5G technology. The monitoring center can remotely control the transport machine's operation via 5G technology, controlling the power unit to drive the transport machine and controlling the braking device to apply brakes.

Description

Technical Field

[0001] This invention relates to the field of rail transport machine control and monitoring, and in particular to a remote control and safety monitoring system for rail transport machines. Background Technology

[0002] The main reason why many mountainous areas have untapped mountain resources is the inconvenient transportation and extreme difficulties in transporting goods. The widespread use of agricultural machinery such as mountain rail transport machines has greatly reduced the labor intensity of farmers and opened up a shortcut to prosperity. Mountain rail transport machines are used to solve the transportation problems of orchards, harvested fruits, and crops in hilly and mountainous areas, as well as the transportation of fertilizers and pesticides during daily management, and the transport of small tillage implements, spraying equipment, and weeding machinery. This significantly helps farmers reduce transportation costs, improve work efficiency, and enhances their enthusiasm for crop cultivation.

[0003] In the complex terrain and environment of mountainous areas, rail transport vehicles must be able to detect obstacles on the road ahead and their own defects in order to take timely measures to avoid collisions and accidents. However, current detection technology for mountain rail transport vehicles needs further improvement and refinement to enhance the accuracy of detecting obstacles ahead and the surrounding environment.

[0004] In existing technologies, the operational status of rail transport vehicles is mainly detected through manual inspection and ultrasonic testing. This method is susceptible to subjective judgment and experience of the operators, potentially leading to bias. Different operators may have different opinions on the same issue, resulting in inconsistent results. Large rail transport vehicles typically have complex structures and numerous components, making comprehensive inspection entirely through manual inspection extremely time-consuming and labor-intensive. Using ultrasonic sensors to measure the propagation and echo signals of ultrasonic waves on the surface of the rail transport vehicle to detect defects, cracks, or material fatigue is a simple and real-time method, but it may have limitations for deeply buried or inaccessible areas. Furthermore, during transport, rail transport vehicles need real-time awareness of obstacles ahead to avoid collisions and accidents. In addition, the positioning and monitoring of the transport vehicle itself are also crucial for effectively understanding its operational status. Summary of the Invention

[0005] To address the aforementioned problems, the present invention aims to provide a remote control and safety monitoring system for rail transport vehicles. This system utilizes 5G technology to achieve remote control of the rail transport vehicle, including its speed, direction, and braking. It employs multi-sensor technology for real-time safety monitoring, including combining video images and radar to detect obstacles ahead for automatic obstacle avoidance, using infrared thermal imagers to detect defects in gears and racks, and using accelerometers to detect vehicle swaying. This system can improve the operational flexibility, safety, and efficiency of rail transport vehicles, reduce accident risks, and provide crucial support and guidance for maintenance and management.

[0006] To achieve the above objectives, the present invention provides a remote control and safety monitoring system for a rail transport vehicle, implemented as follows:

[0007] A remote control and safety monitoring system for a rail transport machine includes a support column, a track, a front radar, a front camera, an engine, a control box, a fuel tank, a frame, a cargo box, a rear camera, a rear radar, a braking device, driven wheels, and a power unit. The track is mounted above the support column. The transport machine, composed of the engine, control box, fuel tank, frame, cargo box, braking device, driven wheels, and power unit, is mounted on the upper front of the frame. The cargo box is mounted on the upper rear of the frame. Two identical driven wheels are mounted on the bottom of the frame below the cargo box to assist the transport machine's movement. The power unit is mounted on the lower front of the frame to drive the transport machine on the track. The front radar and front camera are mounted on the front of the frame to detect obstacles ahead. The rear camera and rear radar are mounted on the rear of the frame. The braking device is mounted on the bottom of the frame below the rear of the cargo box to provide braking. The transport aircraft is equipped with brakes, fuel tanks supply fuel to the engine, and the electricity generated by the engine is transmitted to the power unit, enabling the power unit to travel on the track. The power unit drives the chassis, which in turn moves the cargo box, thus transporting the goods. With the assistance of driven wheels, the transport aircraft can travel more smoothly on the track. Front radar, front camera, rear camera, and rear radar transmit the collected data to the control box. The control box integrates this information to identify obstacles, allowing the transport aircraft to brake in time to avoid accidents. This data is also transmitted to the monitoring center via 5G technology. The monitoring center can remotely control the transport aircraft's operation via 5G technology, including its speed, direction, and braking. The control box controls the power unit to drive the transport aircraft and controls the braking system to apply the brakes.

[0008] The track of the present invention includes a groove and a rack. The groove is cut into the track to provide support points and guidance for the transporter. The rack is set below the track so that the power unit can mesh with the rack to provide forward or backward power for the transporter.

[0009] The power unit of this invention includes a servo motor, a reducer, a vibration sensor, an upper gear, a bearing bracket, a main gear, a drive gear, a triaxial accelerometer, an infrared thermal imager, a spur gear, a lower gear, a rotating shaft, bearings, and a speed sensor. The servo motor's shaft is connected to the power input end of the reducer, and the main gear is connected to the power output end of the reducer. The bearing bracket is mounted together with the reducer. The upper gear and drive gear are connected together by a rotating shaft and mounted on the bearing bracket. The lower gear and spur gear are connected together by a rotating shaft and mounted on the bearing bracket. The rotating shaft and bearing bracket are connected by bearings. The servo motor drives the reducer, which in turn drives the main gear to rotate. The main gear drives the upper and lower gears, thereby driving the drive gear to move in the groove of the track and driving the spur gear to move on the rack of the track. The vibration sensor is installed inside the bearing bracket for... The system monitors for defects in the reducer, bearings, upper gear, main gear, and lower gear. A triaxial accelerometer and an infrared thermal imager are installed on the outer surface of the bearing bracket. The triaxial accelerometer detects the vibration of the power unit, reflecting the overall vibration of the transport vehicle. The infrared thermal imager detects defects in the spur gears and the rack on the track. A speed sensor is installed at the power output end of the reducer to detect its output speed. The converted speed value is used as the operating speed of the transport vehicle. Data collected by the vibration sensor, triaxial accelerometer, infrared thermal imager, and speed sensor are transmitted to the control box. The control box then transmits this data to the monitoring center for real-time monitoring. Upon receiving the speed information from the speed sensor, the control box adjusts the speed of the servo motor to keep the transport vehicle within a preset speed range.

[0010] The present invention employs two drive wheels and two spur gears, with each drive wheel connected to an upper gear via a rotating shaft, which is then mounted on a bearing bracket; each spur gear is connected to a lower gear via a rotating shaft, which is also mounted on a bearing bracket. This dual-gear design improves the stability, transmission efficiency, and load-bearing capacity of the transport vehicle, while reducing the load on each drive wheel and enhancing the reliability and durability of the system.

[0011] The driven wheel of this invention is an iron caster. The iron caster is connected to the frame by hydraulic shock absorption. The hydraulic shock absorption can absorb and reduce the energy generated by vibration and impact, thereby reducing the impact on the frame.

[0012] The braking device of the present invention includes a support box, a telescopic motor, and brake shoes. The support box is installed at the bottom of the frame, the telescopic motor is installed in the support box, and the brake shoes are connected to the telescopic rod of the telescopic motor. When braking of the transport machine is required, the control box controls the servo motor to stop rotating, and the meshing friction between the spur gear and the rack achieves primary braking, slowing down the speed of the transport machine. Then, the control box controls the telescopic motor to extend the telescopic rod, pushing the brake shoes onto the caster wheel to achieve secondary braking. Through the contact between the brake shoes and the caster wheel, a greater braking force is generated, further slowing down or even stopping the transport machine.

[0013] The control box of this invention includes an aluminum metal box, a voltage regulator, a telescopic motor drive board, a servo motor drive board, a control circuit board, and a 5G module. The voltage regulator, telescopic motor drive board, servo motor drive board, control circuit board, and 5G module are all installed in the aluminum metal box. The antenna of the 5G module is placed on the outer surface of the aluminum metal box. The voltage regulator is connected to the generator to stabilize the power voltage generated by the generator within its set value range, enabling the servo motor and telescopic motor to operate normally under their rated operating voltage. The telescopic motor drive board and servo motor drive board are electrically connected to the control circuit board and the voltage regulator. The power supply regulated by the voltage regulator is input to the telescopic motor drive board and servo motor drive board respectively, and then the telescopic motor drive board and servo motor drive board are connected to the telescopic motor and servo motor respectively. The control circuit board controls the output of the telescopic motor drive board and servo motor drive board, thereby controlling the rotation of the telescopic motor and servo motor. The 5G module is connected to the control circuit board, and the front camera, front radar, rear camera, rear radar, vibration sensor, triaxial accelerometer, infrared thermal imager, and speed sensor are electrically connected to the control circuit board, and the collected data is transmitted to the front and rear of the device. Data is transmitted to the control circuit board, which processes it and then controls the 5G module to transmit this data to the PC in the monitoring center for real-time monitoring. The PC displays real-time video images of the environment before and after the transport vehicle, as well as information on the reducer, bearings, upper gear, main gear, lower gear, spur gear, and rack, monitoring for defects in these components. The PC also records the location of any shaking in the transport vehicle, allowing staff to check for broken racks. After receiving speed information from the speed sensor, the control circuit board adjusts the output of the servo motor drive board to regulate the servo motor's speed, keeping the transport vehicle within a preset speed range. Furthermore, staff can send speed, direction, and braking commands to the 5G module via the PC. The 5G module sends these commands to the control circuit board, which then controls the output of the servo motor drive board to adjust the servo motor's rotation and speed. Forward rotation of the servo motor controls the transport vehicle's forward movement, while reverse rotation controls its backward movement. The control circuit board also controls the telescopic motor drive board to control the telescopic motor's brake shoes for braking.

[0014] This invention uses video images and radar data to perform decision fusion to detect obstacles in the forward or backward direction of a transport aircraft. For video image detection, a target tracking algorithm is used to detect obstacles; for radar detection, a support vector machine algorithm is used. When both detection algorithms identify an obstacle, a radar-visual fusion decision is made. When the target is successfully matched, an alarm is output. The radar-visual fusion decision scheme is as follows:

[0015] S1. Time Alignment: The front and rear radars, along with the front and rear cameras, monitor the orbital perimeter environment in real time. The collected radar data and video data are fed into the target tracking algorithm and support vector machine algorithm, respectively, and real-time detection is performed in parallel. The radar refresh rate is 12 frames / second, and the video frame rate is 36 frames / second. The video acquisition rate is 3 times that of the radar. After alignment at the detection time start point, the video data is sampled every frame according to the timestamp of the radar data acquisition to achieve time synchronization.

[0016] S2. Spatial alignment: Because radar detection data contains coordinate information, when it detects a target, it simultaneously extracts the target's world coordinates. The video algorithm detects the target and obtains the target's centroid pixel coordinates, which are then converted to imaging coordinates. Finally, the imaging coordinates are converted to world coordinates. The front radar and front camera are installed side by side at the same height, so they can be approximated as having the same world coordinates. Similarly, the rear radar and rear camera are installed side by side at the same height, so they can also be approximated as having the same world coordinates. After the two algorithms detect the target, they simultaneously extract and convert the target coordinates, unifying them in one world coordinate system, thus completing spatial synchronization.

[0017] S3. Target Decision: After the two detection algorithms identify the target, they calculate and extract the target's world coordinates, input the identification results into the decision fusion module, and the module makes a decision and alarm based on the logical fusion judgment rules.

[0018] S4. Fusion and comparison: Based on the radar recognition time and video recognition time, calculate the time difference T. The time difference T is then compared with a set threshold T0. S Perform a comparison; if T < T S Assuming the identification results from both methods occurred within the same time period, the world coordinates of the radar-identified target and the video-identified target are further extracted, the distance between the targets is calculated, and compared with a distance threshold D. S Compare, if D < D S If the millimeter-wave radar and video fusion detection target are successfully compared, a fusion alarm result will be output; if T>TS or D>D S If the comparison fails, the video and the millimeter-wave radar identification results will be treated as separate detection cases, and the process will proceed to steps S5 and S6.

[0019] S5. When the video detects a target alone, the waiting time Tw If no new result appears within the waiting time, the judgment ends. If a new recognition result appears, the recognition method is determined. If the target is detected by radar, proceed to step S4; if it is still a video alarm, perform cumulative calculation and determine the time period T. D If the number of internal alarms exceeds the threshold S, the current time and screen brightness are obtained to determine whether it is an extremely low light scene at night. If it is not the scene, an alarm will be output; otherwise, manual review will be initiated.

[0020] S6. When the millimeter-wave radar detects a target alone, the waiting time T w If no new result appears during the waiting time, the judgment ends. If a new recognition result appears, the recognition method is determined. If it is video, the process jumps to step S4. If it is still a millimeter-wave radar alarm, the cumulative calculation is performed. If the number of TD alarms exceeds S within the time period, the alarm result is output.

[0021] The present invention employs a target tracking algorithm to achieve obstacle detection in video as follows:

[0022] S1. Video capture: Record real-time video of the surrounding environment using front and rear camera capture devices.

[0023] S2. Preprocessing: The video is preprocessed, including noise reduction, image enhancement, brightness and contrast adjustment, etc., to improve the effect of subsequent detection algorithms.

[0024] S3. Object Detection: Use convolutional neural networks to process the video and identify obstacles in the video frames.

[0025] S4. Target Tracking: For consecutive video frames, target tracking algorithms are used to track the position and trajectory of obstacles in order to determine their correlation between different frames.

[0026] S5. Obstacle Classification: The random forest algorithm is used to classify the detected obstacles to distinguish different types of obstacles.

[0027] The present invention employs a support vector machine algorithm to implement radar obstacle detection as follows:

[0028] S1. Data Acquisition and Preprocessing: Front and rear radars are used to acquire signal data from the surrounding environment. The acquired raw data is preprocessed, including noise removal, filtering, and standardization, to improve data quality.

[0029] S2. Feature Extraction: Extract Doppler shift features from the preprocessed data. These features should better describe the characteristics of the obstacle.

[0030] S3. Data labeling: Label the data according to the actual situation, and divide the data into two categories: obstacles and non-obstacles.

[0031] S4. Dataset partitioning: The labeled dataset is divided into training and test sets, usually using cross-validation to evaluate the algorithm's performance.

[0032] S5. Feature Selection: The feature selection method based on analysis of variance is used to select features that have a significant impact on obstacle detection performance.

[0033] S6. Support Vector Machine Model Training: Using the training set data and labels, a classification model is trained using the support vector machine algorithm. The support vector machine maps the data to a high-dimensional space and constructs an optimal hyperplane to achieve classification.

[0034] S7. Model Evaluation: The trained support vector machine model is evaluated using a test set, and metrics such as accuracy, recall, and F1 score are calculated to assess the algorithm's performance.

[0035] S8. Parameter optimization: Adjust the kernel function type, kernel function parameters, and penalty coefficient according to the actual situation to improve the generalization ability of the model.

[0036] S9. Obstacle Detection: Use a trained support vector machine model to detect obstacles in new radar data. Input the extracted features into the support vector machine model, and judge and classify obstacles based on the model's output.

[0037] Because this invention employs 5G technology for remote control of the rail transport vehicle and utilizes multi-sensor technology for real-time safety monitoring of the transport vehicle, the following beneficial effects can be achieved:

[0038] 1. This invention utilizes 5G technology to enable remote control of rail transport vehicles. This means that operators can remotely control the speed, direction, and braking of the transport vehicle via a network connection without needing to physically approach it. This not only improves operational flexibility but also reduces operational risks.

[0039] 2. Through multi-sensor technology, obstacles in front of the transport aircraft can be monitored in real time, and automatic obstacle avoidance can be performed. When the system detects an obstacle ahead, the transport aircraft can automatically stop or adjust its speed and direction to avoid collisions with the obstacle and ensure transportation safety. Attached Figure Description

[0040] Figure 1 This is a schematic diagram of the overall structure of a remote control and safety monitoring system for a rail transport vehicle according to the present invention;

[0041] Figure 2This is a schematic diagram of the track structure of a remote control and safety monitoring system for a rail transport vehicle according to the present invention;

[0042] Figure 3 , Figure 4 This is a schematic diagram of the power unit of a remote control and safety monitoring system for a rail transport machine according to the present invention;

[0043] Figure 5 This is a schematic diagram of the installation structure of the driven wheel in the remote control and safety monitoring system for a rail transport machine according to the present invention;

[0044] Figure 6 This is a schematic diagram of the braking device of a remote control and safety monitoring system for a rail transport machine according to the present invention;

[0045] Figure 7 This is a schematic diagram of the control box of a remote control and safety monitoring system for a rail transport machine according to the present invention;

[0046] Figure 8 This is a schematic diagram illustrating the working principle of a remote control and safety monitoring system for a rail transport vehicle according to the present invention.

[0047] Figure 9 This is a diagram of a radar-visual fusion decision-making scheme for a remote control and safety monitoring system for rail transport vehicles according to the present invention.

[0048] Explanation of symbols for key components.

[0049]

[0050] Detailed Implementation

[0051] The present invention will now be described in further detail with reference to the embodiments and the accompanying drawings.

[0052] Please see Figures 1 to 9 The diagram shows a remote control and safety monitoring system for a rail transport machine according to the present invention, including a support column 1, a track 2, a front radar 3, a front camera 4, an engine 5, a control box 6, an oil tank 7, a frame 8, a cargo box 9, a rear camera 10, a rear radar 11, a braking device 12, a driven wheel 13, and a power unit 14.

[0053] like Figure 1As shown, the track 2 is installed above the support column 1. The transport vehicle consists of an engine 5, control box 6, fuel tank 7, frame 8, cargo box 9, braking device 12, driven wheels 13, and power unit 14. The engine 5, control box 6, and fuel tank 7 are installed above the front of the frame 8, and the cargo box 9 is installed above the rear of the frame 8. Two identical driven wheels 13 are installed at the bottom of the frame 8 below the cargo box 9 to assist the transport vehicle's movement. The power unit 14 is installed below the front of the frame 8 to drive the transport vehicle on the track 2. A front radar 3 and a front camera 4 are installed at the front of the frame 8 to detect obstacles ahead. A rear camera 10 and a rear radar 11 are installed at the rear of the frame 8. The braking device 12 is installed at the bottom of the frame 8 below the rear of the cargo box 9 to provide braking for the transport vehicle. The fuel tank 7 supplies fuel to the engine 5, and the electricity generated by the engine 5 is transmitted to the power unit 14, causing the power unit 14 to move on the track 2. When the power unit 14 moves, it drives the frame 8. The transport vehicle is driven by the frame 8, which moves the cargo box 9 to transport the goods. With the assistance of the driven wheels 13, the transport vehicle can travel more smoothly on the track 2. The front radar 3, front camera 4, rear camera 10, and rear radar 11 transmit the collected data to the control box 6. In the control box 6, this information is fused to identify obstacles so that the transport vehicle can brake in time to avoid accidents. This data is also transmitted to the monitoring center via 5G technology. At the monitoring center, the operation of the transport vehicle can be remotely controlled via 5G technology, including the running speed, running direction, and braking. The control box 6 controls the power unit 14 to drive the transport vehicle and controls the braking device 12 to brake the transport vehicle. Multi-sensor technology is used to realize real-time safety monitoring of the transport vehicle, including detecting obstacles in front of the transport vehicle in the running direction to realize automatic obstacle avoidance, detecting gears and racks 16 on the transport vehicle and track 2, and detecting the shaking of the transport vehicle. This remote control and safety monitoring system for the rail transport aircraft has advantages such as automatic obstacle avoidance, remote control, smooth driving, real-time safety monitoring, and efficient transportation, which helps to improve the safety and efficiency of the transport aircraft.

[0054] like Figure 2 As shown, the track 2 includes a groove 15 and a rack 16. The groove 15 is cut into the track 2 to provide support points and guidance for the transport vehicle. The rack 16 is positioned below the track 2 to facilitate the engagement of the power unit 14 with the rack 16, providing forward or backward power to the transport vehicle. This track 2 design ensures stable operation of the transport vehicle on the track 2 and provides more precise control and guidance for forward or backward movement based on the engagement of the power unit 14 and the rack 16. Through this design, the transport vehicle can operate efficiently on the track 2, achieving safe and reliable cargo transportation.

[0055] like Figure 3As shown, the power unit 14 includes a servo motor 17, a reducer 18, a vibration sensor 19, an upper gear 20, a bearing bracket 21, a main gear 22, a drive gear 23, a triaxial accelerometer 24, an infrared thermal imager 25, a spur gear 26, a lower gear 27, a rotating shaft 28, a bearing 29, and a speed sensor 30. The rotating shaft of the servo motor 17 is connected to the power input end of the reducer 18, the main gear 22 is connected to the power output end of the reducer 18, the bearing bracket 21 is mounted together with the reducer 18, the upper gear 20 and the drive gear 23 are connected together by the rotating shaft 28 and mounted on the bearing bracket 21, and the lower gear 27 and the spur gear 26 are connected together by the rotating shaft 28 and mounted on the shaft. On the bearing bracket 21, the rotating shaft 28 is connected to the bearing bracket 21 by a bearing 29. A servo motor 17 drives a reducer 18, which in turn drives the main gear 22 to rotate. The main gear 22 drives the upper gear 20 and the lower gear 27, which in turn drives the drive wheel 23 to move in the groove 15 of the track 2, and drives the spur gear 26 to move on the rack 16 of the track 2. A vibration sensor 19 is installed inside the bearing bracket 21 to monitor for defects in the reducer 18, bearing 29, upper gear 20, main gear 22, and lower gear 27. This allows for real-time monitoring of the vibration status of these components, enabling early detection of defective or potentially faulty parts and the implementation of appropriate repair measures to prevent more serious mechanical failures. This contributes to improving the reliability and safety of the equipment. A triaxial accelerometer 24 and an infrared thermal imager 25 are mounted on the outer surface of the bearing bracket 21. The triaxial accelerometer 24 is used to detect the swaying of the power unit 14, reflecting the overall swaying of the transport machine. Since the accelerometer 30 can measure the acceleration changes of an object in three axes, when the power unit 14 sways, the sensor detects the acceleration changes caused by the vibration. This acceleration data can be analyzed to determine the swaying of the transport machine. The infrared thermal imager 25 is used to detect defects in the spur gear 26 and the rack 16 on the track 2. Because the spur gear 26 generates heat due to friction after meshing with the rack 16, defective or faulty spur gear 26 or rack 16 components usually produce different heat dissipation patterns. By using the infrared thermal imager 25, the difference in heat dissipation between the spur gear 26 and the rack 16 on the track 2 can be detected, thereby discovering potential defects or faults. Speed ​​sensor 30 is installed at the power output end of reducer 18 to detect the output speed of reducer 18. The converted speed value is used as the running speed of the transport vehicle. The data information collected by vibration sensor 19, triaxial accelerometer 24, infrared thermal imager 25 and speed sensor 30 are transmitted to control box 6. Control box 6 transmits this data information to the monitoring center for real-time monitoring. After receiving the speed information from speed sensor 30, control box 6 adjusts the speed of servo motor 17 to keep the transport vehicle traveling within the preset speed range.

[0056] like Figure 4 As shown, two drive wheels 23 and two spur gears 26 are used, and each drive wheel 23 is connected to an upper gear 20 through a rotating shaft 28, which is then mounted on a bearing bracket 21; each spur gear 26 is connected to a lower gear 27 through a rotating shaft 28, which is then mounted on a bearing bracket 21. The dual-gear design can improve the stability, transmission efficiency and load-bearing capacity of the transport machine, while reducing the load on each drive wheel 23 and improving the reliability and durability of the system.

[0057] like Figure 5 As shown, the driven wheel 13 is a steel caster 31, which is connected to the frame 8 via a hydraulic damper 32. The hydraulic damper 32 absorbs and mitigates the energy generated by vibration and impact, thereby reducing the impact on the frame 8. This helps protect the structure of the transport vehicle and provides smoother operation. Furthermore, the hydraulic damper 32 connection can compensate for the position of the steel caster 31 by adjusting the fluid pressure, ensuring that the steel caster 31 always maintains good contact with the slide 15, providing stable travel.

[0058] like Figure 6 As shown, the braking device 12 includes a support box 33, a telescopic motor 34, and brake shoes 35. The support box 33 is installed at the bottom of the frame 8, and the telescopic motor 34 is installed in the support box 33. The support box 33 provides support and fixation, while the telescopic motor 34 drives the operation of the braking device 12. The brake shoes 35 are connected to the telescopic rod of the telescopic motor 34. When braking is initiated, the telescopic motor 34 controls the movement of the telescopic rod, causing the brake shoes 35 to contact the surface of the caster wheel 31. When braking of the transport machine is required, the control box 6 controls the servo motor 17 to stop rotating. The meshing friction between the spur gear 26 and the rack 16 achieves primary braking, slowing down the transport machine. Immediately afterwards, the control box 6 controls the telescopic motor 34 to extend the telescopic rod, pushing the brake shoes 35 against the caster wheel 31 to achieve secondary braking. Through the contact between the brake shoes 35 and the caster wheel 31, a greater braking force is generated, further slowing down or even stopping the transport machine. The braking device 12 achieves two-stage braking, providing a more reliable and flexible braking effect, ensuring the safety and stability of the transport machine.

[0059] like Figure 7As shown, the control box 6 includes an aluminum metal box 36, a voltage regulator 37, a telescopic motor drive board 38, a servo motor drive board 39, a control circuit board 40, and a 5G module 41. The voltage regulator 37, telescopic motor drive board 38, servo motor drive board 39, control circuit board 40, and 5G module 41 are all installed in the aluminum metal box 36. The antenna of the 5G module 41 is placed on the outer surface of the aluminum metal box 36. The voltage regulator 37 is connected to the generator and is used to stabilize the power voltage generated by the generator within its set value range, enabling the servo motor 17 and the telescopic motor 34 to operate normally under their rated operating voltage. The telescopic motor drive board 38 and the servo motor drive board 39 are both electrically connected to the control circuit board 40 and the voltage regulator 37. The voltage after being regulated by the voltage regulator 37... The power sources are input to the telescopic motor drive board 38 and the servo motor drive board 39, respectively. The telescopic motor drive board 38 and the servo motor drive board 39 are then connected to the telescopic motor 34 and the servo motor 17, respectively. The control circuit board 40 controls the output of the telescopic motor drive board 38 and the servo motor drive board 39, thereby controlling the rotation of the telescopic motor 34 and the servo motor 17. The 5G module 41 is connected to the control circuit board 40. The front camera 4, front radar 3, rear camera 10, rear radar 11, vibration sensor 19, triaxial accelerometer 24, infrared thermal imager 25, and speed sensor 30 are electrically connected to the control circuit board 40 and transmit the collected data to the control circuit board 40. After processing by the control circuit board 40, the 5G module 41 controls these components. Data is transmitted to a PC in the monitoring center for real-time monitoring. The PC displays real-time video images of the environment before and after the transport vehicle, as well as information on the reducer 18, bearing 29, upper gear 20, main gear 22, lower gear 27, spur gear 26, and rack 16. This monitors for defects in these components. The PC also records the location of any shaking in the transport vehicle, allowing staff to check for broken rack 16. Furthermore, after receiving speed information from the speed sensor 30, the control circuit board 40 adjusts the output of the servo motor drive board 39 to regulate the speed of the servo motor 17, ensuring the transport vehicle travels within a preset speed range. Additionally, staff can send speed, direction, and braking data to the 5G module 41 via the PC. The 5G module 41 sends instructions to the control circuit board 40, which then controls the output of the servo motor drive board 39 to adjust the rotation and speed of the servo motor 17. The forward rotation of the servo motor 17 controls the transport vehicle's forward movement, while the reverse rotation controls its backward movement. When the front radar 3 and the front camera 4 or the rear camera 10 and rear radar 11 detect an obstacle, the control circuit board 40 controls the output of the servo motor drive board 39 to stop the servo motor 17 from rotating. The meshing friction between the spur gear 26 and the rack 16 achieves primary braking, slowing the transport vehicle. Immediately afterwards, the control circuit board 40 controls the telescopic motor drive board 38 to control the telescopic motor 34 to push the brake shoe 35, achieving secondary braking.

[0060] like Figure 9 As shown, this invention uses video images and radar data to perform decision fusion to detect obstacles in the forward or backward direction of a transport aircraft. For video image detection, a target tracking algorithm is used to detect obstacles; for radar detection, a support vector machine algorithm is used. When both detection algorithms identify an obstacle, a radar-visual fusion decision is made. When the target matching is successful, an alarm is output. The radar-visual fusion decision scheme is as follows:

[0061] S1. Time Alignment: Front radar 3, rear radar 11, front camera 4, and rear camera 10 monitor the perimeter environment of orbit 2 in real time. The collected radar data and video data are respectively fed into the target tracking algorithm and support vector machine algorithm, and real-time detection is started in parallel. The radar refresh rate is 12 frames / second, the video frame rate is 36 frames / second, and the video acquisition rate is 3 times that of the radar. After alignment at the detection time start point, the video data is sampled every frame according to the timestamp of the radar data acquisition to achieve time synchronization.

[0062] S2. Spatial alignment: Because radar detection data contains coordinate information, when a target is detected, the world coordinates of the target are extracted synchronously. The video algorithm detects the target and obtains the centroid pixel coordinates of the target, which are then converted to imaging coordinates. Finally, the imaging coordinates are converted to world coordinates. The front radar 3 and the front camera 4 are installed side by side at the same height, which can be approximated as the same world coordinate system. The rear radar 11 and the rear camera 10 are also installed side by side at the same height, which can be approximated as the same world coordinate system. After the two algorithms detect the target, they simultaneously extract and convert the target coordinates, unifying them in one world coordinate system to complete spatial synchronization.

[0063] S3. Target Decision: After the two detection algorithms identify the target, they calculate and extract the target's world coordinates, input the identification results into the decision fusion module, and the module makes a decision and alarm based on the logical fusion judgment rules.

[0064] S4. Fusion and comparison: Based on the radar recognition time and video recognition time, calculate the time difference T. The time difference T is then compared with a set threshold T0. S Perform a comparison; if T < T S Assuming the identification results from both methods occurred within the same time period, the world coordinates of the radar-identified target and the video-identified target are further extracted, the distance between the targets is calculated, and compared with a distance threshold D. S Compare, if D < D S If the millimeter-wave radar and video fusion detection target are successfully compared, a fusion alarm result will be output; if T>TS or D>D SIf the comparison fails, the video and the millimeter-wave radar identification results will be treated as separately detected cases, and the process will proceed to steps S5 and S6.

[0065] S5. When the video detects a target alone, the waiting time T w If no new result appears within the waiting time, the judgment ends. If a new recognition result appears, the recognition method is determined. If the target is detected by radar, proceed to step S4; if it is still a video alarm, perform cumulative calculation and determine the time period T. D If the number of internal alarms exceeds the threshold S, the current time and screen brightness are obtained to determine whether it is an extremely low light scene at night. If it is not the scene, an alarm will be output; otherwise, manual review will be initiated.

[0066] S6. When the millimeter-wave radar detects a target alone, the waiting time T w If no new result appears within the waiting time, the judgment ends. If a new recognition result appears, the recognition method is determined. If it is video, proceed to step S4; if it is still a millimeter-wave radar alarm, perform cumulative calculation within the time period T. D If the number of alarms exceeds S, the alarm result will be output.

[0067] The present invention employs a target tracking algorithm to achieve obstacle detection in video as follows:

[0068] S1. Video capture: Real-time video of the surrounding environment is recorded using front camera 4 and rear camera 10.

[0069] S2. Preprocessing: The video is preprocessed, including noise reduction, image enhancement, brightness and contrast adjustment, etc., to improve the effect of subsequent detection algorithms.

[0070] S3. Object Detection: Use convolutional neural networks to process the video and identify obstacles in the video frames.

[0071] S4. Target Tracking: For consecutive video frames, target tracking algorithms are used to track the position and trajectory of obstacles in order to determine their correlation between different frames.

[0072] S5. Obstacle Classification: The random forest algorithm is used to classify the detected obstacles to distinguish different types of obstacles.

[0073] The present invention employs a support vector machine algorithm to implement radar obstacle detection as follows:

[0074] S1. Data Acquisition and Preprocessing: The front radar 3 and rear radar 11 are used to acquire signal data from the surrounding environment. The acquired raw data is preprocessed, including noise removal, filtering, and standardization, to improve data quality.

[0075] S2. Feature Extraction: Extract Doppler shift features from the preprocessed data. These features should better describe the characteristics of the obstacle.

[0076] S3. Data labeling: Label the data according to the actual situation, and divide the data into two categories: obstacles and non-obstacles.

[0077] S4. Dataset partitioning: The labeled dataset is divided into training and test sets, usually using cross-validation to evaluate the algorithm's performance.

[0078] S5. Feature Selection: The feature selection method based on analysis of variance is used to select features that have a significant impact on obstacle detection performance.

[0079] S6. Support Vector Machine Model Training: Using the training set data and labels, a classification model is trained using the support vector machine algorithm. The support vector machine maps the data to a high-dimensional space and constructs an optimal hyperplane to achieve classification.

[0080] S7. Model Evaluation: The trained support vector machine model is evaluated using a test set, and metrics such as accuracy, recall, and F1 score are calculated to assess the algorithm's performance.

[0081] S8. Parameter optimization: Adjust the kernel function type, kernel function parameters, and penalty coefficient according to the actual situation to improve the generalization ability of the model.

[0082] S9. Obstacle Detection: Use a trained support vector machine model to detect obstacles in new radar data. Input the extracted features into the support vector machine model, and judge and classify obstacles based on the model's output.

[0083] The working principle and process of this invention are as follows:

[0084] The control circuit board 40 controls the output of the telescopic motor drive board 38 and the servo motor drive board 39, thereby controlling the rotation of the telescopic motor 34 and the servo motor 17. The front camera 4, front radar 3, rear camera 10, rear radar 11, vibration sensor 19, three-axis accelerometer 24, infrared thermal imager 25, and speed sensor 30 transmit the collected data to the control circuit board 40. After processing by the control circuit board 40, the 5G module transmits this data to the PC in the monitoring center for real-time monitoring. The PC displays real-time environmental video images before and after the transport vehicle, as well as information on the reducer 18, bearing 29, upper gear 20, main gear 22, lower gear 27, spur gear 26, and rack 16, monitoring for defects in these components. The PC also records the position of the transport vehicle's swaying. The rack 16 is positioned to facilitate staff to check if it is broken. After receiving speed information from the speed sensor 30, the control circuit board 40 adjusts the output of the servo motor drive board 39 to regulate the speed of the servo motor 17, keeping the transport vehicle within a preset speed range. In addition, staff can send speed, direction, and braking commands to the 5G module 41 via PC. The 5G module 41 sends the commands to the control circuit board 40, which then controls the output of the servo motor drive board 39 to adjust the rotation and speed of the servo motor 17. The forward rotation of the servo motor 17 controls the transport vehicle to move forward, and the reverse rotation of the servo motor 17 controls the transport vehicle to move backward. The control circuit board 40 controls the telescopic motor drive board 38 to control the telescopic motor 34 to push the brake shoe 35 to achieve braking.

Claims

1. A method for remote control and safety monitoring of a rail transport vehicle, characterized in that: This system, used for remote control and safety monitoring of rail transport vehicles, includes a support column, track, front radar, front camera, engine, control box, fuel tank, frame, cargo box, rear camera, rear radar, braking device, driven wheels, and power unit. The track is mounted above the support column. The transport vehicle, composed of the engine, control box, fuel tank, frame, cargo box, braking device, driven wheels, and power unit, is mounted on the upper front of the frame. The cargo box is mounted on the upper rear of the frame. Two identical driven wheels are mounted on the bottom of the frame below the cargo box to assist the transport vehicle's movement. The power unit is mounted on the lower front of the frame to propel the transport vehicle along the track. The front radar and front camera are mounted on the front of the frame to detect obstacles. The rear camera and rear radar are mounted on the rear of the frame. The braking device is mounted on the bottom of the frame below the rear of the cargo box. The system provides braking for the transport aircraft. Fuel is supplied to the engine by the fuel tank, and the electricity generated by the engine is transmitted to the power unit, enabling the power unit to travel on the track. When the power unit is traveling, it drives the frame, which in turn drives the cargo in the cargo box to move, thus realizing the transport of goods. With the assistance of the driven wheels, the transport aircraft can travel more smoothly on the track. The front radar, front camera, rear camera, and rear radar transmit the collected data to the control box. In the control box, this information is fused to identify obstacles, so that the transport aircraft can brake in time to avoid safety accidents. This data is also transmitted to the monitoring center via 5G technology. At the monitoring center, the operation of the transport aircraft can be remotely controlled via 5G technology, including the running speed, running direction, and braking. The control box controls the power unit to drive the transport aircraft and controls the braking device to perform braking operations on the transport aircraft. This method uses video images and radar data to perform decision fusion to detect obstacles in the forward or backward direction of a transport aircraft. For video image detection, a target tracking algorithm is used; for radar detection, a support vector machine algorithm is used. When both detection algorithms identify an obstacle, a radar-visual fusion decision is made. When the target match is successful, an alarm is output. The radar-visual fusion decision scheme is as follows: S1. Time alignment: The front and rear radars, along with the front and rear cameras, monitor the orbital perimeter environment in real time. The collected radar data and video data are fed into the target tracking algorithm and support vector machine algorithm, respectively, and real-time detection is performed in parallel. The radar refresh rate is 12 frames / second, and the video frame rate is 36 frames / second. The video acquisition rate is 3 times that of the radar. After alignment at the starting point of the detection time, the video data is sampled every frame according to the timestamp of the radar data acquisition to achieve time synchronization. S2. Spatial alignment: Because radar detection data contains coordinate information, when it detects a target, it simultaneously extracts the target's world coordinates. The video algorithm detects the target and obtains the target's centroid pixel coordinates, which are then converted to imaging coordinates. Finally, the imaging coordinates are converted to world coordinates. The front radar and front camera are installed side by side at the same height, so they can be approximated as having the same world coordinates. Similarly, the rear radar and rear camera are installed side by side at the same height, so they can also be approximated as having the same world coordinates. After the two algorithms detect the target, they simultaneously extract and convert the target coordinates, unifying them in one world coordinate system to complete spatial synchronization. S3. Target Decision: After the two detection algorithms identify the target, the world coordinates of the target are calculated and extracted. The identification result information is input into the decision fusion module, which makes a decision and alarm based on the logical fusion judgment rules. S4. Fusion and comparison: Based on the radar recognition time and video recognition time, calculate the time difference T, and compare the time difference T with a set threshold T. S Perform a comparison; if T < T S Assuming the identification results from both methods occurred within the same time period, the world coordinates of the radar-identified target and the video-identified target are further extracted, the distance between the targets is calculated, and compared with a distance threshold D. S Compare, if D < D S If the millimeter-wave radar and video fusion detection target are successfully compared, a fusion alarm result will be output; if T > T S Or D > D S If the comparison fails, the video and the millimeter-wave radar identification results will be treated as separate detection cases, and the process will proceed to steps S5 and S6 respectively. S5. When the video detects a target alone, the waiting time T w If no new result appears within the waiting time, the judgment ends. If a new recognition result appears, the recognition method is determined. If the target is detected by radar, proceed to step S4; if it is still a video alarm, perform cumulative calculation and determine the time period T. D If the number of internal alarms exceeds the threshold S, the current time and screen brightness are obtained to determine whether it is an extremely low light scene at night. If it is not the scene, an alarm will be output. If it is, manual review will be initiated. S6. When the millimeter-wave radar detects a target alone, the waiting time T w If no new result appears within the waiting time, the judgment ends. If a new recognition result appears, the recognition method is determined. If it is video, proceed to step S4; if it is still a millimeter-wave radar alarm, perform cumulative calculation within the time period T. D If the number of alarms exceeds the threshold S, the alarm result will be output.

2. The remote control and safety monitoring method for rail transport machines according to claim 1, characterized in that: The power unit includes a servo motor, a reducer, a vibration sensor, an upper gear, a bearing bracket, a main gear, a drive gear, a triaxial accelerometer, an infrared thermal imager, a spur gear, a lower gear, a rotating shaft, bearings, and a speed sensor. The servo motor's shaft is connected to the reducer's power input end, and the main gear is connected to the reducer's power output end. The bearing bracket is mounted on the reducer. The upper gear and drive gear are connected together by a rotating shaft and mounted on the bearing bracket. The lower gear and spur gear are also connected together by a rotating shaft and mounted on the bearing bracket. The rotating shaft and bearing bracket are connected by bearings. The servo motor drives the reducer, which in turn drives the main gear to rotate. The main gear drives the upper and lower gears, which in turn drive the drive gear to move in the track's groove and the spur gear to move on the track's rack. The vibration sensor is installed inside the bearing bracket to monitor for defects in the reducer, bearings, upper gear, main gear, and lower gear. The triaxial accelerometer and infrared thermal imager are mounted on the outer surface of the bearing bracket. The triaxial accelerometer is used to detect the sway of the power unit. The system monitors the movement of the entire transport vehicle by detecting the vibration of the power unit. An infrared thermal imager detects defects in the spur gears and racks on the track. A speed sensor, installed at the power output end of the reducer, detects the output speed. The converted speed value is used as the transport vehicle's operating speed. Data collected by the vibration sensor, triaxial accelerometer, infrared thermal imager, and speed sensor are transmitted to the control box, which then transmits this data to the monitoring center for real-time monitoring. Upon receiving speed information from the speed sensor, the control box adjusts the servo motor's speed to keep the transport vehicle within a preset speed range. Two drive wheels and two spur gears are used, with each drive wheel connected to an upper gear via a rotating shaft mounted on a bearing bracket. Each spur gear is connected to a lower gear via a rotating shaft also mounted on a bearing bracket. This dual-gear design improves the transport vehicle's stability, transmission efficiency, and load-bearing capacity, while reducing the load on each drive wheel, thus enhancing the system's reliability and durability.

3. The remote control and safety monitoring method for rail transport machines according to claim 1, characterized in that: The control box includes an aluminum metal box, a voltage regulator, a telescopic motor drive board, a servo motor drive board, a control circuit board, and a 5G module. The voltage regulator, telescopic motor drive board, servo motor drive board, control circuit board, and 5G module are all installed in the aluminum metal box. The antenna of the 5G module is placed on the outer surface of the aluminum metal box. The voltage regulator is connected to the generator to stabilize the power voltage generated by the generator within its set range, enabling the servo motor and telescopic motor to operate normally under their rated operating voltage. The telescopic motor drive board and servo motor drive board are electrically connected to the control circuit board and the voltage regulator. The power regulated by the voltage regulator is input to the telescopic motor drive board and servo motor drive board respectively, and then the telescopic motor drive board and servo motor drive board are connected to the telescopic motor and servo motor respectively. The control circuit board controls the output of the telescopic motor drive board and servo motor drive board, thereby controlling the rotation of the telescopic motor and servo motor. The 5G module is connected to the control circuit board, and the front camera, front radar, rear camera, rear radar, vibration sensor, triaxial accelerometer, infrared thermal imager, and speed sensor are electrically connected to the control circuit board, transmitting the collected data... The information is transmitted to the control circuit board, which processes it and then controls the 5G module to transmit this data to the PC in the monitoring center for real-time monitoring. The PC displays real-time video images of the environment before and after the transport vehicle, as well as information on the reducer, bearings, upper gear, main gear, lower gear, spur gear, and rack, monitoring for defects in these components. The PC also records the location of any shaking in the transport vehicle, allowing staff to check for broken racks. After receiving speed information from the speed sensor, the control circuit board adjusts the output of the servo motor drive board to regulate the servo motor's speed, keeping the transport vehicle within a preset speed range. Furthermore, staff can send speed, direction, and braking commands to the 5G module via the PC. The 5G module sends these commands to the control circuit board, which then controls the output of the servo motor drive board to adjust the servo motor's rotation and speed. Forward rotation of the servo motor controls the transport vehicle's forward movement, while reverse rotation controls its backward movement. The control circuit board also controls the telescopic motor drive board to control the telescopic motor's brake shoes for braking.

4. The remote control and safety monitoring method for rail transport machines according to claim 1, characterized in that: The method for detecting obstacles in video using a target tracking algorithm is as follows: S1. Video capture: Record real-time video of the surrounding environment using front and rear camera capture devices; S2. Preprocessing: The video is preprocessed, including noise reduction, image enhancement, brightness adjustment, and contrast adjustment, in order to improve the performance of subsequent detection algorithms; S3. Object Detection: Use convolutional neural networks to process the video and identify obstacles in the video frames; S4. Target Tracking: For consecutive video frames, target tracking algorithms are used to track the position and motion trajectory of obstacles in order to determine their correlation between different frames; S5. Obstacle Classification: The random forest algorithm is used to classify the detected obstacles to distinguish different types of obstacles.

5. The remote control and safety monitoring method for rail transport machines according to claim 1, characterized in that: The solution for radar obstacle detection using the support vector machine algorithm is as follows: S1. Data Acquisition and Preprocessing: Use front and rear radars to acquire signal data from the surrounding environment, and preprocess the acquired raw data, including noise removal, filtering, and standardization, to improve data quality. S2. Feature Extraction: Extract Doppler shift features from the preprocessed data. These features can better describe the characteristics of obstacles. S3. Data labeling: Label the data according to the actual situation, and divide the data into two categories: obstacles and non-obstacles; S4. Dataset partitioning: The labeled dataset is divided into training and test sets, usually using cross-validation to evaluate the algorithm's performance; S5. Feature Selection: The feature selection method based on analysis of variance is used to select features that have a significant impact on obstacle detection performance; S6. Support Vector Machine Model Training: Using training set data and labels, a classification model is trained using the support vector machine algorithm. The support vector machine maps the data to a high-dimensional space and constructs an optimal hyperplane to achieve classification. S7. Model Evaluation: The trained support vector machine model is evaluated using a test set, and the recognition accuracy, recall, and F1 score are calculated to evaluate the algorithm performance. S8. Parameter optimization: Adjust the kernel function type, kernel function parameters, and penalty coefficient according to the actual situation to improve the generalization ability of the model; S9. Obstacle Detection: Use a trained support vector machine model to detect obstacles in new radar data. Input the extracted features into the support vector machine model and judge and classify obstacles based on the model's output.

Images