Energy self-sufficiency type intelligent tunnel inspection device and system
By introducing CCD fisheye cameras and three-axis fill lights into the tunnel patrol robot, combining tunnel wind power generation and dual battery swap mode, the tunnel detection blind spots and energy waste problems are solved, and efficient and continuous tunnel detection is achieved.
Patent Information
- Application Number
- CN202510475607.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-16
- Publication Date
- 2025-08-26
AI Technical Summary
The existing tunnel inspection robot lacks a lighting system, which leads to insufficient light in long tunnels affecting detection accuracy. It is also heavy, fast power consumption, long charging time, and cannot detect the safety status of the tunnel in a timely manner. It has high cost of use and maintenance, and is not suitable for conventional short-distance tunnels.
An energy-sufficient intelligent tunnel inspection device was designed, using CCD fisheye camera to identify four-directional live situations, paired with three-axis adaptive fill lights, using tunnel wind energy to generate electricity and adopting dual battery swap mode, combining NFC locator and intelligent cloud management system to realize adaptive inspection and efficient energy management.
It solves the problems of tunnel detection blind spots and insufficient light, improves detection accuracy, reduces energy waste, realizes efficient and continuous tunnel detection, and reduces cost and resource waste.
Smart Images
Figure CN120544290A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of tunnel inspection, and in particular to an energy-self-sufficient intelligent tunnel inspection device and system. Background Art
[0002] In recent years, the development of intelligent inspection devices in highway tunnels has evolved rapidly, using a variety of advanced technologies, including intelligent sensors, data processing and analysis systems, and robot inspections, which have greatly improved the real-time monitoring capabilities of tunnel internal parameters, can timely identify abnormal conditions in the tunnel, and realize remote monitoring and fault diagnosis. In terms of intelligent sensor technology, the inspection device realizes real-time monitoring of environmental parameters such as temperature, humidity and gas in the tunnel by assembling various sensors, such as infrared sensors and temperature sensors, into robots. By comparing with known normal data, it can quickly detect abnormalities and respond. In terms of data processing and analysis systems, the inspection device collects information through intelligent sensors, processes and analyzes the data based on algorithms such as YOLO, and converts it into operating status reports and early warning information that can be known to the tunnel management department, discovers problems in time, and provides a basis for making scientific decisions for tunnel safety management. The use of data processing technology can make tunnel inspections more scientific and accurate, and provide guarantees for tunnel safe operation. In terms of robot inspections, there are currently available on the market. There are relatively few tunnel inspection products, most of which use crawler and roller methods for movement, which have the advantages of simple structure and good obstacle crossing performance. In addition, there are other inspection methods such as track inspection. Among them, the track inspection method can avoid occupying emergency tracks by building running tracks, reducing the possibility of inspection robots being unable to move forward due to obstacles, and has good environmental adaptability. With the increasing development of artificial intelligence technology and robotics technology, inspection robots are widely used in various fields. In response to safety issues in highway tunnel inspection work and the current situation of excessive power loss when using highway tunnels, autonomous energy storage intelligent track inspection devices that can perform visual analysis in highway tunnels are developed to improve tunnel maintenance efficiency, reduce energy consumption, and realize the intelligent development of tunnel traffic safety detection. At present, most tunnel inspection robots, including the three types mentioned above, are installed on the tunnel wall or drive directly on the road surface. Due to the close contact with the tunnel wall, there is a detection blind spot on the side close to the tunnel wall; and driving directly on the road surface may affect the normal use of the tunnel and affect the traffic flow during tunnel inspection.
[0003] However, traditional inspection devices have the following disadvantages:
[0004] (1) Tunnel inspection robots proposed on the market do not involve lighting systems. However, in long tunnels and some old tunnels, the internal lighting is relatively weak. In addition, the lights of motor vehicles in the tunnel will also have a certain impact on the detection equipment, reducing its response rate and increasing the detection error rate. For example, the robot may misjudge some dark areas or miss some smaller damaged areas.
[0005] (2) The top inspection robots on the market now have relatively complete detection functions, so they are heavy and consume power quickly. Most of these robots are charged by charging piles, which takes a long time and requires pulling additional cables, increasing the cost of robot deployment. In addition, they cannot detect the safety status in the tunnel in a timely manner when charging;
[0006] (3) The robot carries a large number of sensors, detectors and other equipment, which results in high costs for use and maintenance. Therefore, it is not suitable for conventional, short-distance ordinary tunnels, which will lead to an "oversupply" of its functions and cause a certain amount of waste of resources. Summary of the Invention
[0007] The object of the present invention is to provide an energy-self-sufficient intelligent tunnel inspection device and system to solve the problem mentioned in the above background technology that the tunnel inspection robots proposed on the market do not involve a lighting system. However, in long tunnels and some old tunnels, the internal lighting is relatively weak, and the lights of motor vehicles in the tunnel will also have a certain impact on the detection equipment, reducing its reaction rate and increasing the detection error rate. For example, the robot may misjudge some dark areas or miss some smaller damaged areas; the top inspection robots on the market now have relatively complete detection functions, so they are heavy and consume power quickly. Such robots are mostly charged by charging piles, which takes a long time to charge and requires pulling additional cables, which increases the cost of robot layout, and the safety status in the tunnel cannot be detected in time when charging; the robots carry a large number of sensors and detectors, so the use and maintenance costs are high, so they are not suitable for conventional, short-distance ordinary tunnels, which will lead to an "oversupply" of their functions and cause a certain waste of resources.
[0008] To achieve the above-mentioned objectives, the present invention provides the following technical solutions: an energy-self-sufficient intelligent tunnel inspection device and system, comprising an inspection device, the inspection device comprising a motion positioning device, an energy storage device, an intelligent cloud management system, a detection and identification system, an intelligent auxiliary device and a host computer, the motion positioning device being connected to the energy storage device, the motion positioning device, the energy storage device and the host computer being all connected to the intelligent cloud management system, the intelligent cloud management system and the intelligent auxiliary device being both connected to the detection and identification system, the motion positioning device comprising a servo motion motor and an NFC locator, the NFC locator being connected to the servo motion motor, the detection and identification system comprising a photographic imaging target detection component, a first YOLOv5 target detection module and an image preprocessing module, the first YOLOv5 target detection module and the image preprocessing module being both connected to the photographic imaging target detection component, the intelligent cloud management system comprising a second YOLOv5 target detection module, a detection and identification data module, a communication module and an audible and visual alarm, the second YOLOv5 target detection module being connected to the detection and identification data module, the detection and identification data module being connected to the communication module, and the communication module being connected to the audible and visual alarm.
[0009] As a preferred technical solution of the present invention, the image preprocessing module includes a Mosaic data enhancement submodule, an adaptive anchor frame calculation submodule and an adaptive picture scaling submodule;
[0010] The Mosaic data enhancement submodule uses four pictures and splices them together by random scaling, random cropping, and random arrangement. Each picture has its corresponding frame. After splicing the four pictures, a new picture is obtained, and the frame corresponding to this picture is also obtained. Then a new picture is passed into the neural network for learning, which is equivalent to passing in four pictures for learning at once. This method greatly enriches the background of the detected object. In the YOLO series of algorithms, the adaptive anchor box calculation submodule needs to set an anchor box of specific length and width for different data sets. In the network training stage, the model outputs the corresponding prediction box based on the initial anchor box in the initial stage, calculates the gap between it and the GT box, and performs a reverse update operation to update the parameters of the entire network. In the target detection algorithm, the adaptive picture scaling submodule has different lengths and widths of different pictures, so the common method is to uniformly scale the original picture to a standard size and then send it to the detection network.
[0011] As an optimal technical solution of the present invention, the servo motion motor includes a DC power supply, a control drive circuit, a motor body and a position sensor, the DC power supply and the position sensor are both connected to the control drive circuit, the control drive circuit is connected to the motor body, and the motor body is connected to the position sensor.
[0012] As a preferred technical solution of the present invention, the control drive circuit includes an Arduino development board, a power battery and a motor speed control board. The Arduino development board and the power battery are both connected to the motor speed control board, and the motor speed control board is connected to the motor body. The written motor operation program is implanted into the Arduino development board, and then the Arduino development board is connected to the motor speed control board. At the same time, the battery is also connected to the motor speed control board. Finally, the speed control board is connected to the motor to control its rotation.
[0013] As a preferred technical solution of the present invention, the NFC locator includes an NFC initiator device and an NFC target device. The NFC initiator device is connected to the NFC target device. The robot receives the task category of the tunnel block through the NFC signal reader to determine different inspection methods. For example, a slow and careful inspection is adopted for a tunnel block that is old, has undergone multiple repairs, and has frequent accidents. A quick and simple inspection is adopted for a tunnel block that is new, has a short tunnel length, and has not had any accidents. On this basis, the image recognition module is used to judge whether the current tunnel environment is "good" or "bad" to send the corresponding signal to the tunnel block. Input the information into the industrial computer and change the motion state, causing the robot to stop or slow down to carefully inspect the problem area. If the "bad" signal appears multiple times, the result is uploaded to the cloud. After cloud processing, the result is fed back to the robot. The feedback result determines the subsequent operation strategy and the motion state is changed again. If the feedback result is a more serious accident, such as a large car accident or obvious cracks in the tunnel, the robot will continue to detect slowly or reciprocally, and the full picture of the accident will be fed back to the cloud. If the feedback result is: a general accident, a small vehicle rear-end collision, or slight peeling of the tunnel wall, the robot will be restored to its original motion state. This cycle will be repeated to achieve efficient and intelligent detection of tunnels.
[0014] As an optimal technical solution of the present invention, the energy storage device includes a battery replacement component, an energy storage battery and a breeze power generation system energy storage module, the battery replacement component is connected to the energy storage battery, the energy storage battery is connected to the breeze power generation system energy storage module, the NFC locator and the energy storage battery are both connected to the second YOLOv5 target detection module, the servo motion motor is connected to the battery replacement component, the detection and identification data module is connected to the photo imaging target detection component, and the communication module is connected to the host computer. By setting five groups of vertical axis wind turbines on the charging pile, the wind energy generated by the narrow tube effect and the vehicle-driven wind phenomenon in the tunnel is utilized. Due to the natural narrow-caliber characteristics of the tunnel, the natural wind flowing through the tunnel will be significantly amplified. At the same time, the traffic speed in the highway tunnel is 80km / s, which will generate a vehicle-driven wind of 22m / s. The device consists of two parts: a tunnel wind generator and an energy storage bin. The tunnel wind flows through the outer sleeve. The wind speed will be amplified to 1.6-2 times of the original. After the tunnel wind enters the sleeve, it will drive the blades to rotate, and then increase the rotation speed through the speed increaser to drive the main shaft and rotor to move to generate electricity. The electricity is then stored in the battery pack in the energy storage compartment at the bottom of the charging pile for storage and standby. The inspection robot is equipped with an "automatic return to the nest" function to automatically return to the charging pile when the battery is low. The cloud system dynamically detects the battery power to determine whether the robot's power is lower than the specified value in the current state; when the power management unit detects that the battery power is lower than the specified value, it sends a command to the control unit, commanding the inspection device to move to a charging pile with sufficient power, and realizes automatic battery replacement through the "motor + guide rail" drive mode and the setting of dual battery slots; when the battery replacement is completed, information is sent to the control unit, and the inspection device checks whether the new battery can work normally and waits for the next work instruction.
[0015] As a preferred technical solution of the present invention, the intelligent auxiliary device includes a first rotating shaft and a second rotating shaft, a first micro motor is fixedly installed on the bottom end of the inner wall of the first rotating shaft, the output end of the first micro motor is fixedly connected to the bottom end of the second rotating shaft, a detection frame is fixedly installed on the top end of the second rotating shaft, the internal rotation of the detection frame is connected to the third rotating shaft, a camera frame is fixedly installed in the middle of the third rotating shaft, a CCD fisheye camera is fixedly installed in the middle of one side of the camera frame, a fill aperture located outside the CCD fisheye camera is fixedly installed on the camera frame, a second micro motor that drives the third rotating shaft to rotate is fixedly installed on the surface of the detection frame, the first micro motor starts after being powered on, the first micro motor drives the second rotating shaft to rotate, adjusts the direction of the CCD fisheye camera and the fill aperture, and the second micro motor starts after being powered on The second micro motor drives the third rotating shaft to adjust the angle of the CCD fisheye camera and the fill aperture. When the device enters the detection area, it reads the type of tunnel. The tunnel categories include tunnel areas with high / low accident rates, tunnel areas of different construction years, tunnel areas with different maintenance records, tunnel areas of different lengths, tunnel areas with different directions, inspection tasks and real-time brightness and darkness. At this time, the three-axis adaptive fill light drives the motor according to the specific circumstances of the task and the light and shadow environment, according to the pre-burned program, and rotates the three rotating shafts to supplement the lighting of the detection area. At the same time, the light intensity is adjusted according to the light sensor. In the process of movement, it continuously adjusts the lighting angle according to the change of the movement position to achieve precise fill lighting of the detection area. There is a limit on the movement of the rotating shaft to ensure that the light does not directly shine on the road surface, causing an impact on the vehicle driver.
[0016] As a preferred technical solution of the present invention, the bottom end of the first rotating shaft is connected to the detection and identification system, and the intelligent auxiliary device is installed on the detection and identification system through the first rotating shaft.
[0017] Compared with the prior art, the present invention has the following beneficial effects:
[0018] 1. To address the issues of lateral blind spots and road surface detection affecting tunnel traffic, this design uses a downward-mounted CCD fisheye camera to identify the real-time situation in the four directions of the tunnel: front, top, front, bottom, left, and right. By hanging on the track at the top of the tunnel, it can observe the real-time safety issues of the entire tunnel, thus avoiding the occurrence of the above two problems.
[0019] 2. To address the problem of insufficient lighting inside long, old, and curved tunnels, which can easily affect image recognition, this design uses a set of three-axis movable three-axis adaptive fill lights. The servo drives the angle to achieve follow-up fill light and brightness adjustment functions. When the device enters the detection area, it reads the tunnel type, inspection task, and real-time brightness and darkness. At this time, the three-axis adaptive fill light drives the motor according to the specific situation of the task and the lighting environment, according to the pre-programmed program, rotates the three shafts, and supplements the illumination of the detection area. At the same time, the light intensity is adjusted according to the light sensor. During the movement, the light angle is continuously adjusted according to the change of the movement position, achieving precise fill light for the detection area, thereby more clearly and comprehensively detecting the details of the tunnel. The servo rotation is limited to prevent the road surface from being illuminated and affecting the driving of the vehicle.
[0020] 3. In order to solve the problems of long charging time, high cost and inability to continue inspection during charging of the inspection robot, this design sets up a set of vertical axis power generation structures, utilizes the tunnel wind energy generated by the narrow channel effect, and the relatively large number of vehicle-borne wind resources, to construct a tunnel wind power generation-energy storage structure, and adopts a dual-battery replacement working mode, that is, when one battery is charging at the charging pile, the other battery is installed on the robot to work. Two power battery compartments for battery replacement are set in the robot, and two groups of corresponding charging slots are set in the charging pile. When the battery power of the inspection robot is lower than the threshold, the automatic return program is started. After entering the predetermined charging position, the battery is unlocked, and the external feeding device gently pushes the low-power battery into the charging pile. The battery replacement device locks the low-power battery and unlocks it at the same time and uses the spring to release the fully charged battery to complete the directional movement of the battery during the battery replacement process. The inspection robot then continues to work, thereby improving its continuous high-intensity working ability to cope with sudden power outages in the tunnel;
[0021] 4. To address the problem of inspection robots adapting to various tunnels, this design first comprehensively distinguishes different tunnel types in terms of algorithm application and inspection task division. Each tunnel is then divided into multiple blocks based on its type. NFC chips are installed on the tracks of each block. Different chips are loaded with inspection tasks and location information corresponding to different areas. This allows the robot's optimal operating conditions, the working mode of the three-axis adaptive fill light, and the degree of need for multiple sensors to be calculated in advance, thus greatly reducing the problem of energy and resource waste.
[0022] 5. This design is equipped with a set of three-axis movable three-axis adaptive fill light. The angle is adjusted by driving the servo to achieve follow-up fill light and brightness adjustment functions, so as to more clearly and comprehensively detect the details in the tunnel. The servo angle is limited so that it will not illuminate the road surface and affect the driving of the vehicle.
[0023] 6. This design sets up a set of vertical axis power generation structures, utilizing the tunnel wind energy generated by the narrow channel effect and the abundant vehicle-borne wind resources to construct a tunnel wind power generation and energy storage structure. It also adopts a dual-battery swapping working mode. That is, when one battery is charging at the charging pile, the other battery is installed on the robot to work. After entering the charging position, the charging pile receives the low-battery battery of the robot and unlocks the fully charged battery to release it for use by the robot. The robot continues to work, thereby improving its continuous high-intensity work capacity and coping with sudden power outages in the tunnel.
[0024] 7. In terms of algorithm application and inspection task division, this design first distinguishes the categories of different tunnels as a whole, then divides each tunnel into multiple blocks according to its type, and sets corresponding inspection tasks and location information in each block, so as to pre-calculate the optimal operating conditions of the robot, the working mode of the three-axis adaptive fill light, and the demand for multiple sensors, thereby largely solving the problem of energy and resource waste. BRIEF DESCRIPTION OF THE DRAWINGS
[0025] Figure 1 Schematic diagram of the structure of the inspection device of the present invention;
[0026] Figure 2 is a cross-sectional view of the intelligent assistance device of the present invention;
[0027] Figure 3 Schematic diagram of the structure of the image preprocessing module of the present invention;
[0028] Figure 4 This is a schematic diagram of the structure of the NFC locator of the present invention;
[0029] Figure 5 Schematic diagram of the structure of the servo motor of the present invention;
[0030] Figure 6 Schematic diagram of the structure of the control drive circuit of the present invention;
[0031] Figure 7 This is a working diagram of the energy storage device of the present invention;
[0032] Figure 8 This is a flow chart of power monitoring and homing triggering of the present invention.
[0033] In the figure: 1. Inspection device; 2. Motion positioning device; 21. Servo motion motor; 211. DC power supply; 212. Control drive circuit; 2121. Arduino development board; 2122. Power battery; 2123. Motor speed control board; 213. Motor body; 214. Position sensor; 22. NFC locator; 221. NFC initiator; 222. NFC target device; 3. Energy storage device; 31. Battery replacement component; 32. Energy storage battery; 33. Breeze power generation system energy storage module; 4. Intelligent cloud management system; 41. Second YOLOv5 target detection module; 42. Detection recognition Identification data module; 43. Communication module; 44. Sound and light alarm; 5. Detection and recognition system; 51. Photo imaging target detection component; 52. First YOLOv5 target detection module; 53. Image preprocessing module; 531. Mosaic data enhancement submodule; 532. Adaptive anchor frame calculation submodule; 533. Adaptive image scaling submodule; 6. Intelligent auxiliary device; 61. First rotating shaft; 62. First micro motor; 63. Second rotating shaft; 64. Detection frame; 65. Second micro motor; 66. Third rotating shaft; 67. Camera frame; 68. CCD fisheye camera; 69. Fill aperture; 7. Host computer. DETAILED DESCRIPTION
[0034] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.
[0035] See also Figure 1-8The present invention provides an energy-self-sufficient intelligent tunnel inspection device and system, including an inspection device 1, the inspection device 1 includes a motion positioning device 2, an energy storage device 3, an intelligent cloud management system 4, a detection and identification system 5, an intelligent auxiliary device 6 and a host computer 7, the motion positioning device 2 is connected to the energy storage device 3, the motion positioning device 2, the energy storage device 3 and the host computer 7 are all connected to the intelligent cloud management system 4, the intelligent cloud management system 4 and the intelligent auxiliary device 6 are all connected to the detection and identification system 5, the motion positioning device 2 includes a servo motion motor 21 and an NFC locator 22, the NFC locator 22 is connected to the servo motion motor 21 Then, the detection and recognition system 5 includes a photographic imaging target detection component 51, a first YOLOv5 target detection module 52 and an image preprocessing module 53. The first YOLOv5 target detection module 52 and the image preprocessing module 53 are both connected to the photographic imaging target detection component 51. The intelligent cloud management system 4 includes a second YOLOv5 target detection module 41, a detection and recognition data module 42, a communication module 43 and an audible and visual alarm 44. The second YOLOv5 target detection module 41 is connected to the detection and recognition data module 42, the detection and recognition data module 42 is connected to the communication module 43, and the communication module 43 is connected to the audible and visual alarm 44.
[0036] The image preprocessing module 53 includes a mosaic data enhancement submodule 531, an adaptive anchor frame calculation submodule 532, and an adaptive picture scaling submodule 533;
[0037] The Mosaic data enhancement submodule 531 uses four pictures and splices them together by random scaling, random cropping and random arrangement. Each picture has its corresponding frame. After splicing the four pictures, a new picture is obtained, and the frame corresponding to this picture is also obtained. Then the new picture is passed into the neural network for learning, which is equivalent to passing in four pictures for learning at once. This method greatly enriches the background of the detected object. In the YOLO series of algorithms, the adaptive anchor frame calculation submodule 532 needs to set an anchor frame of specific length and width for different data sets. In the network training stage, the model outputs the corresponding prediction frame based on the initial anchor frame in the initial stage, calculates the difference between it and the GT frame, and performs a reverse update operation to update the parameters of the entire network. In the target detection algorithm, the length and width of different pictures are different, so the common method is to uniformly scale the original picture to a standard size and then send it to the detection network.
[0038] The servo motor 21 includes a DC power supply 211, a control drive circuit 212, a motor body 213 and a position sensor 214. The DC power supply 211 and the position sensor 214 are both connected to the control drive circuit 212, the control drive circuit 212 is connected to the motor body 213, and the motor body 213 is connected to the position sensor 214.
[0039] The control drive circuit 212 includes an Arduino development board 2121, a power battery 2122 and a motor speed control board 2123. The Arduino development board 2121 and the power battery 2122 are both connected to the motor speed control board 2123, and the motor speed control board 2123 is connected to the motor body 213. The written motor operation program is implanted into the Arduino development board 2121, and then the Arduino development board 2121 is connected to the motor speed control board 2123. At the same time, the battery is also connected to the motor speed control board 2123. Finally, the speed control board is connected to the motor to control its rotation.
[0040] The NFC locator 22 includes an NFC initiator device 221 and an NFC target device 222. The NFC initiator device 221 is connected to the NFC target device 222. The robot receives the task category of the tunnel block through the NFC signal reader to determine different inspection methods. For example, a slow and careful inspection is adopted for a tunnel block that is old, has been repaired many times, and has many accidents. A quick and simple inspection is adopted for a tunnel block that is new, has a short tunnel length, and has not had any accidents. On this basis, the image recognition module is used to judge whether the current tunnel environment is "good" or "bad", and the corresponding signal is input into the work The control machine changes the motion state, causing the robot to stop or slow down to carefully inspect the problem area; if the "bad" signal appears multiple times, the result is uploaded to the cloud, and after cloud processing, the result is fed back to the robot. The feedback result determines the subsequent operation strategy and the motion state is changed again. If the feedback result is a more serious accident such as a large car accident or obvious cracks in the tunnel, the robot will continue to detect slowly or reciprocally, and the full picture of the accident will be fed back to the cloud. If the feedback result is: a general accident such as a small vehicle rear-end collision or slight peeling of the tunnel wall, the robot will return to its original motion state. This cycle is repeated to achieve efficient and intelligent detection of tunnels.
[0041] The energy storage device 3 includes a battery replacement component 31, an energy storage battery 32 and a breeze power generation system energy storage module 33. The battery replacement component 31 is connected to the energy storage battery 32, and the energy storage battery 32 is connected to the breeze power generation system energy storage module 33. The NFC locator 22 and the energy storage battery 32 are both connected to the second YOLOv5 target detection module 41, the servo motion motor 21 is connected to the battery replacement component 31, the detection and identification data module 42 is connected to the photo imaging target detection component 51, and the communication module 43 is connected to the host computer 7. By setting five groups of vertical axis wind turbines on the charging pile, the wind energy generated by the narrow tube effect and the vehicle-driven wind phenomenon in the tunnel is utilized. Due to the natural narrow diameter characteristics of the tunnel, the natural wind flowing through the tunnel will be significantly amplified. At the same time, the traffic speed in the highway tunnel is 80km / s, which will generate a vehicle-driven wind of 22m / s. The device consists of two parts: a tunnel wind generator and an energy storage bin. The tunnel wind flows through the outer sleeve. The wind speed will be amplified to 1.6-2 times of the original. After the tunnel wind enters the sleeve, it will drive the blades to rotate, and then increase the rotation speed through the speed increaser to drive the main shaft and rotor to move to generate electricity. The electricity is then stored in the battery pack in the energy storage compartment at the bottom of the charging pile for storage and standby. The inspection robot is equipped with an "automatic return to the nest" function to automatically return to the charging pile when the battery is low. The cloud system dynamically detects the battery power to determine whether the robot's power is lower than the specified value in the current state; when the power management unit detects that the battery power is lower than the specified value, it sends an instruction to the control unit, commanding the inspection device 1 to move to a charging pile with sufficient power, and realizes automatic battery replacement through the "motor + guide rail" drive mode and the setting of dual battery slots; when the battery replacement is completed, information is sent to the control unit, and the inspection device 1 checks whether the new battery can work normally and waits for the next work instruction.
[0042] The intelligent auxiliary device 6 includes a first rotating shaft 61 and a second rotating shaft 63. A first micro motor 62 is fixedly installed at the bottom end of the inner wall of the first rotating shaft 61. The output end of the first micro motor 62 is fixedly connected to the bottom end of the second rotating shaft 63. A detection frame 64 is fixedly installed on the top of the second rotating shaft 63. The inner rotation of the detection frame 64 is connected to the third rotating shaft 66. A camera frame 67 is fixedly installed in the middle of the third rotating shaft 66. A CCD fisheye camera 68 is fixedly installed in the middle of one side of the camera frame 67. A fill aperture 69 located outside the CCD fisheye camera 68 is fixedly installed on the camera frame 67. A second micro motor 65 that drives the third rotating shaft 66 to rotate is fixedly installed on the surface of the detection frame 64. The first micro motor 62 starts after being energized. The first micro motor 62 drives the second rotating shaft 63 to rotate, adjusts the directions of the CCD fisheye camera 68 and the fill aperture 69, and the second micro motor 65 After power is turned on, the second micro motor 65 drives the third shaft 66 to rotate, adjusting the angles of the CCD fisheye camera 68 and the fill aperture 69. When the device enters the detection area, it reads the type of tunnel. Tunnel categories include tunnel areas with high / low accident rates, tunnel areas of different construction years, tunnel areas with different maintenance records, tunnel areas of different lengths, tunnel areas with different directions, inspection tasks and real-time brightness and darkness. At this time, the three-axis adaptive fill light drives the motor according to the pre-programmed program based on the specific circumstances of the task and the light and shadow environment, rotates the three shafts, supplements the lighting to the detection area, and adjusts the light intensity according to the photosensor. During the movement, it continuously adjusts the light irradiation angle according to the change of the movement position to achieve precise fill lighting for the detection area. There is a limit on the movement of the shaft to ensure that the light does not directly shine on the road surface, causing an impact on the vehicle driver.
[0043] The bottom end of the first rotating shaft 61 is connected to the detection and identification system 5 , and the intelligent auxiliary device 6 is installed on the detection and identification system 5 through the first rotating shaft 61 .
[0044] In the present invention, the intelligent cloud management system 4 is used for the collection, processing, transportation and overall control of the operating logic of the working environment of the integrated data, which is mainly composed of the host computer and the Zigbee communication module 43. The specific implementation path is as follows: the cloud system collects the image information and traffic flow monitoring sent back by the robot through the Zigbee module, and the host computer 7 analyzes the returned image information to determine whether the tunnel environment is normal. When the feedback is normal, the robot continues to work under normal conditions; when relevant problems and faults are detected, the cloud will update the robot's motion state, so that the robot can adopt different detection methods according to different situations, and timely process the traffic flow, accident details, and tunnel information obtained by the visual analysis of the robot system, turn on the sound and light warning in the tunnel, and at the same time feedback the image data to the traffic management department or the department responsible for tunnel safety. On this basis, the cloud system will also upload the fault image information to the robot's data center for enriching the existing data and deep learning of visual recognition, and try to match it with existing cases, combined with the attached manual Figure 7 , automatic battery replacement function: adopts dual battery replacement working mode, that is, when one battery is charging in the charging pile, the other battery is installed on the robot for work. Two power battery compartments for battery replacement are set in the robot, and two sets of corresponding charging slots are set in the charging pile. When the battery power of the inspection robot is lower than the threshold, the automatic return program is started. After entering the predetermined charging position, the battery is unlocked. The external feeding device gently pushes the low-power battery into the charging pile. The battery replacement device locks the low-power battery and unlocks it at the same time and uses the spring to release the fully charged battery to complete the directional movement of the battery during the battery replacement process. Then the inspection robot continues to work. The design of this battery replacement device saves more than 95% of the charging time and improves the robot's endurance. Combined with the instructions attached Figure 8 , set the return conditions according to the current position and the remaining distance, and judge whether the inspection robot returns to the nest. Considering that the coulomb counter can only calculate the change of SOC when the single-chip microcomputer monitors the charging and discharging current of the lithium battery, when there is no detected current, the SOC measurement will have errors. When the time is very long, the error measured according to the formula will become larger. We use the voltage correction algorithm to correct it - according to the corresponding points of power and voltage during the actual charging and discharging process of the lithium battery pack, the corresponding points are collected and saved through the algorithm. When the SOC value and the corresponding value of voltage are too large, the SOC is corrected through the BMS software. The calculation formula is as follows:
[0045] SOC(n+1)=SOC(n)-I(n)*t(n),
[0046] Iteratively update the power, where SOC(n) is the remaining power, in coulombs; I(n) is the discharge current function, in amperes; t(n) is the discharge time function, in seconds. The specific process of adaptive image scaling is to calculate the scaling ratio based on the original image size and the image size input to the network, calculate the scaled image size based on the original image size and the scaling ratio, and calculate the black edge filling value, where 416 represents the image width required by the YOLO v5 network, and 312 represents the width of the scaled image. First, a subtraction operation is performed to obtain the black edge length 104 to be filled; then a remainder operation is performed on the value, i.e. 104%32=8. 32 is used because the entire YOLO v5 network performs 5 downsampling operations. Finally, the value is divided by 2, which disperses the filled area to both sides, thus reducing the 416*416 image to 416*320, thereby greatly improving the inference speed of the algorithm.
[0047] The robot uses a track-based operation mode, that is, a slide rail is installed on the top of the tunnel, and the robot relies on a motor to drive the active wheel to drive the robot to run on the slide rail. During the operation of the robot, optimizing path planning and reducing the mechanical and energy losses caused by repeated operations of the robot are of great significance to the extension of the robot's life. The specific implementation path is as follows: adopting block division work technology, that is, first distinguishing the categories of different tunnels as a whole, and then dividing each tunnel into multiple blocks according to its type, and installing NFC chips on the track of each block. Different chips are loaded with inspection tasks and location information of different areas. According to different inspection tasks, the best acceleration and deceleration operation mode of the robot is calculated in advance, and the motor program is written through kile5 and input to the inspection robot to make it run according to the set optimized path to reduce losses. This project uses NFC technology and radar antenna positioning tracker to realize the motion positioning of the inspection robot. The motor operation control program adopts the "planning + tracking" mode, that is, the robot first drives according to the preset ideal trajectory, and then uses the instruction tracking controller to eliminate the tracking error. NFC signal readers are installed on the wheels of the inspection robot to make the machine The robot reads its location in the block and then uses a positioning tracker to track the robot's real-time position. The control algorithm module performs tasks including coordinate system establishment, path planning, control command generation, external condition constraints, and follow-up control. The STM32 microcontroller controls the stepper motor and controls the motion of the stepper motor by inputting PWM waves. The program module includes a full-duplex serial port communication module, which allows data to be transmitted in both directions simultaneously. Therefore, full-duplex communication is a combination of two simplex communication methods. It requires independent receiving and transmitting capabilities for both the sending and receiving devices, a key input module, and a motor drive module. The robot receives the task category of the tunnel block through the NFC signal reader to determine different inspection methods. For example, a slow and careful inspection is performed in an old tunnel block that has undergone multiple repairs and is prone to accidents. A quick and simple inspection is performed in a newer tunnel block that is short and has not experienced accidents. Based on this, the image recognition module determines whether the current tunnel environment is "good" or "bad". The corresponding signal is input into the industrial computer and the motion state is changed, causing the robot to stop or slow down to carefully inspect the problem area.If "bad" signals appear multiple times, the results are uploaded to the cloud, and after being processed by the cloud, the results are fed back to the robot. The feedback results determine the subsequent operation strategy and change the motion state again. If the feedback result is a more serious accident, such as a large car accident or obvious cracks in the tunnel, the robot will continue to detect slowly or reciprocally, and the full picture of the accident will be fed back to the cloud. If the feedback result is: a general accident with a small vehicle rear-ending or slight peeling of the tunnel wall, the robot will return to its original motion state, and repeat this process to achieve efficient and intelligent detection of tunnels. The detection system with the YOLO single-stage target detection algorithm model as the core includes a combination of Mosaic data, SURF feature point extraction and Opencv image distortion correction modules. The system combines 1080P high-definition cameras and edge detection to achieve accurate and rapid identification of cable damage, wall cracks, and vault leaks in tunnels, greatly To reduce the consumption of manpower and material resources, a block-based work technique, short-range wireless communication technology, and motion equations are combined to determine the position of the inspection robot for track positioning and motion control. The device's servo motor circuit control system is designed to accurately control the device's position and operation. The robot's motion plans are calculated and planned for different road sections and working environments, and working time and energy output are rationally allocated for different working conditions, maximizing work efficiency and improving the detection rate of defects. Regarding energy supply, a set of vertical axis power generation structures is installed, utilizing tunnel wind energy generated by the narrow channel effect and the abundant vehicle-borne wind resources to construct a tunnel wind power generation and energy storage structure. A fully automatic battery replacement device is used to achieve intelligent and automated operation of the entire device, enhancing its endurance and improving work efficiency. The device's survivability in extreme accidents is also improved to cope with sudden power outages in the tunnel.
[0048] Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.
Claims
1. An energy-self-sufficient intelligent tunnel inspection device and system, comprising an inspection device (1), characterized in that: The inspection device (1) comprises a motion positioning device (2), an energy storage device (3), an intelligent cloud management system (4), a detection and identification system (5), an intelligent auxiliary device (6) and a host computer (7); the motion positioning device (2) is connected to the energy storage device (3); the motion positioning device (2), the energy storage device (3) and the host computer (7) are all connected to the intelligent cloud management system (4); the intelligent cloud management system (4) and the intelligent auxiliary device (6) are all connected to the detection and identification system (5); the motion positioning device (2) comprises a servo motion motor (21) and an NFC locator (22); the NFC locator (22) is connected to the servo motion motor (21); the detection and identification system (5) comprises a servo motion motor (21) and an NFC locator (22); the NFC locator (22) is connected to the servo motion motor (21); and the detection and identification system (5) comprises a servo motion motor (21) and an NFC locator (22). The system comprises a photographic imaging target detection component (51), a first YOLOv5 target detection module (52) and an image preprocessing module (53), wherein the first YOLOv5 target detection module (52) and the image preprocessing module (53) are both connected to the photographic imaging target detection component (51), and the intelligent cloud management system (4) comprises a second YOLOv5 target detection module (41), a detection and identification data module (42), a communication module (43) and an audible and visual alarm (44), wherein the second YOLOv5 target detection module (41) is connected to the detection and identification data module (42), the detection and identification data module (42) is connected to the communication module (43), and the communication module (43) is connected to the audible and visual alarm (44).
2. The energy-self-sufficient intelligent tunnel inspection device and system according to claim 1, characterized in that: The image preprocessing module (53) includes a mosaic data enhancement submodule (531), an adaptive anchor frame calculation submodule (532) and an adaptive picture scaling submodule (533); The Mosaic data enhancement submodule (531) uses four pictures and splices them together in a random scaling, random cropping and random arrangement manner. Each picture has its corresponding frame. After splicing the four pictures, a new picture is obtained, and the frame corresponding to this picture is also obtained. Then the new picture is passed into the neural network for learning, which is equivalent to passing four pictures for learning at once. This method greatly enriches the background of the detected object. In the YOLO series of algorithms, the adaptive anchor frame calculation submodule (532) needs to set a specific length and width anchor frame for different data sets. In the network training stage, the model outputs the corresponding prediction frame based on the initial anchor frame in the initial stage, calculates the difference between it and the GT frame, and performs a reverse update operation to update the parameters of the entire network. In the target detection algorithm, the length and width of different pictures are different, so the common method is to uniformly scale the original picture to a standard size and then send it to the detection network.
3. The energy-self-sufficient intelligent tunnel inspection device and system according to claim 1, characterized in that: The servo motion motor (21) comprises a DC power supply (211), a control drive circuit (212), a motor body (213) and a position sensor (214); the DC power supply (211) and the position sensor (214) are both connected to the control drive circuit (212); the control drive circuit (212) is connected to the motor body (213); and the motor body (213) is connected to the position sensor (214).
4. The energy-self-sufficient intelligent tunnel inspection device and system according to claim 3, characterized in that: The control drive circuit (212) comprises an Arduino development board (2121), a power battery (2122) and a motor speed regulating board (2123); the Arduino development board (2121) and the power battery (2122) are both connected to the motor speed regulating board (2123); and the motor speed regulating board (2123) is connected to the motor body (213).
5. The energy-self-sufficient intelligent tunnel inspection device and system according to claim 1, characterized in that: The NFC locator (22) comprises an NFC initiator device (221) and an NFC target device (222), wherein the NFC initiator device (221) is connected to the NFC target device (222).
6. The energy-self-sufficient intelligent tunnel inspection device and system according to claim 1, characterized in that: The energy storage device (3) includes a battery replacement component (31), an energy storage battery (32) and a breeze power generation system energy storage module (33); the battery replacement component (31) is connected to the energy storage battery (32); the energy storage battery (32) is connected to the breeze power generation system energy storage module (33); the NFC locator (22) and the energy storage battery (32) are both connected to a second YOLOv5 target detection module (41); the servo motion motor (21) is connected to the battery replacement component (31); the detection and identification data module (42) is connected to a photographic imaging target detection component (51); and the communication module (43) is connected to a host computer (7).
7. The energy-self-sufficient intelligent tunnel inspection device and system according to claim 1, characterized in that: The intelligent auxiliary device (6) comprises a first rotating shaft (61) and a second rotating shaft (63), wherein a first micro motor (62) is fixedly mounted on the bottom end of the inner wall of the first rotating shaft (61), an output end of the first micro motor (62) is fixedly connected to the bottom end of the second rotating shaft (63), a detection frame (64) is fixedly mounted on the top end of the second rotating shaft (63), a third rotating shaft (66) is rotatably connected to the inside of the detection frame (64), a camera frame (67) is fixedly mounted on the middle part of the third rotating shaft (66), a CCD fisheye camera (68) is fixedly mounted on the middle part of one side of the camera frame (67), a fill aperture (69) located outside the CCD fisheye camera (68) is fixedly mounted on the camera frame (67), and a second micro motor (65) for driving the third rotating shaft (66) to rotate is fixedly mounted on the surface of the detection frame (64).
8. The energy-self-sufficient intelligent tunnel inspection device and system according to claim 7, characterized in that: The bottom end of the first rotating shaft (61) is connected to the detection and identification system (5).