Intelligent all-terrain embankment hidden danger rapid diagnosis crawler and operation method

Abstract

The application discloses an intelligent all-terrain embankment hidden danger rapid diagnosis crawler and a working method.The crawler comprises a vehicle body, a four-wheel independent suspension crawler structure, a geological radar detection module, a transient electromagnetic detection module, a multi-modal navigation system, an intelligent control module and a remote control platform;the intelligent all-terrain embankment hidden danger rapid diagnosis crawler is realized based on the multi-modal navigation system, the intelligent control module and the remote control platform to realize all-terrain path planning, autonomous obstacle avoidance and collaborative detection;through the four-wheel independent suspension crawler structure, double-mode collaborative detection, real-time data transmission and inversion and other technologies, the detection range is significantly expanded, the hidden danger identification accuracy and reliability are improved, the interval between the field diagnosis and the emergency response time is shortened, the bottleneck problems of the prior art in terms of terrain adaptability, detection accuracy, timeliness and the like are solved, and the comprehensiveness, accuracy and emergency response speed of the embankment hidden danger diagnosis are significantly improved.

Description

Technical Field

[0001] This invention relates to the field of dike engineering detection technology, specifically to an intelligent all-terrain dike hazard rapid diagnosis tracked vehicle and its operation method. Background Technology

[0002] As the core barrier for flood control, the accurate detection and rapid diagnosis of internal hazards in dikes are directly related to the safety of people's livelihoods. While geophysical exploration technology is currently used in dike inspection, significant technical bottlenecks remain, leading to low efficiency and unreliable results in hazard identification. Specifically, these bottlenecks manifest in the following key issues: First, the automation level of detection equipment is low, and its terrain adaptability is poor: Existing detection equipment largely relies on manual operation, resulting in low efficiency and discontinuous data collection, making it difficult to meet the demands of high-intensity patrols during the flood season. Conventional detection vehicles, limited by their wheeled chassis structure and power performance, are ill-suited to complex terrains such as steep dike slopes and muddy, rugged terrain, leading to insufficient detection coverage and a significant risk of missed detections. Second, traditional detection methods have prominent limitations and low accuracy: The dynamic changes in the physical properties of the dike medium and the cross-interference of physical parameters make single geophysical methods susceptible to geological conditions and serious ambiguity issues; existing inversion methods lack multi-source data fusion mechanisms, resulting in excessively large errors in identifying cavities. Third, there is a lack of on-site assessment capabilities and a lag in emergency response: the current geophysical exploration equipment requires offline processing of the detection data, and the time from data collection to the generation of results reports is too long, which cannot support real-time decision-making during the flood season.

[0003] Existing technologies propose multimodal detection schemes, but their vehicle-mounted designs are limited to flat terrain and do not solve the data fusion problem. Technological breakthroughs are urgently needed in the following areas: ① Developing an intelligent all-terrain detection platform to overcome the limitations imposed by complex terrain on equipment; ② Constructing a multi-physics joint inversion model to improve the accuracy of hazard identification; ③ Establishing a remote transmission architecture to achieve rapid on-site diagnosis and emergency response. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle and its operation method. The purpose is to improve the hazard detection capability in complex environments and achieve rapid hazard diagnosis and emergency response by using technologies such as four-wheel independent suspension track structure, dual-modal collaborative detection, real-time data transmission and inversion.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an intelligent all-terrain tracked vehicle for rapid diagnosis of embankment hazards, comprising:

[0006] Vehicle body;

[0007] The four-wheel independent suspension track structure is located at the bottom of the vehicle body and is used to move the vehicle body.

[0008] The power system, which is located inside the vehicle body, is used to transmit power to the four-wheel independent suspension track structure;

[0009] A ground-penetrating radar detection module is installed on the underside of the vehicle body and is used to collect radar data for geological structure detection.

[0010] A transient electromagnetic detection module, which is installed on the vehicle body, is used to collect transient electromagnetic data for geological structure detection;

[0011] A multimodal navigation system, which is mounted on the vehicle body, is used to generate environmental maps in real time, take real-time photos, and detect obstacles.

[0012] The intelligent control module, located within the vehicle body, is used to receive and respond to signals; the signal response includes vehicle motion control based on a multimodal navigation system and transmission of collected data; the collected data includes geological structure detection radar data and geological structure detection transient electromagnetic data;

[0013] The remote control platform is used to control the intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle, receive collected data, preprocess the collected data, perform collaborative inversion calculations, and output the hazard area.

[0014] Based on a multimodal navigation system, intelligent control module, and remote control platform, the tracked vehicle enables rapid diagnosis of hidden dangers in all-terrain embankments through all-terrain path planning, autonomous obstacle avoidance, and collaborative detection.

[0015] Furthermore, the multimodal navigation system includes:

[0016] An integrated RTK-GNSS differential positioning module is installed inside the vehicle body to provide global coordinates.

[0017] A lidar, installed inside the vehicle, is used to scan the environment in real time and generate three-dimensional point cloud data;

[0018] A binocular vision camera, which is mounted on the front of the vehicle body, is used to capture real-time images;

[0019] An IMU (Inertial Measurement Unit) is installed inside the vehicle body to perform vehicle attitude measurement, navigation and positioning, and motion monitoring.

[0020] A wheel speed encoder is mounted on the wheel assembly of a four-wheel independent suspension track structure and is used to measure the wheel assembly speed.

[0021] Furthermore, the intelligent control module includes:

[0022] A wireless communication unit, which is installed inside the vehicle body, is used to receive or transmit signals;

[0023] The central control unit is located inside the vehicle body and is connected to the wireless communication unit to receive signals transmitted by the wireless communication unit and respond accordingly.

[0024] The wireless communication unit includes a multi-band router and a 5G communication module installed on the top of the vehicle body, which are used to receive and transmit signals.

[0025] Furthermore, the remote control platform integrates an intelligent vehicle remote control system, a data fusion processing system, and a display device; the intelligent vehicle remote control system is used to control the intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle; the data fusion processing system is used to preprocess the collected data, then perform collaborative inversion calculations, and output the hazard area.

[0026] Furthermore, the central control unit is electrically connected to the ground-penetrating radar detection module, transient electromagnetic detection module, integrated RTK-GNSS differential positioning module, lidar, binocular vision camera, IMU inertial navigation unit, and wheel speed encoder.

[0027] A method for operating a tracked vehicle for rapid diagnosis of hidden dangers in intelligent all-terrain embankments includes the following steps:

[0028] Step S1: The intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle arrives at the target embankment section operation area;

[0029] Step S2: The intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle performs all-terrain path planning, autonomous obstacle avoidance, and collaborative detection in the target embankment section operation area;

[0030] Step S3: The geological structure detection radar data and geological structure detection transient electromagnetic data collected by the detection are transmitted to the data fusion processing system for processing. When a hidden danger is detected in the target embankment section operation area, the location coordinates of the hidden danger are output through the multi-modal navigation system, and a hidden danger distribution map is generated.

[0031] Step S4: Push the location coordinates and distribution map of the hidden danger to the command center. Based on the pushed location coordinates and distribution map of the hidden danger, the command center determines the area where the hidden danger exists and promptly carries out emergency treatment on the section of the embankment with hidden danger.

[0032] Furthermore, the specific process of the intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle performing all-terrain path planning, autonomous obstacle avoidance, and collaborative detection in the target embankment section operation area is as follows:

[0033] The intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle scans the environment in real time with LiDAR to generate three-dimensional point cloud data. Combined with the global coordinates provided by the integrated RTK-GNSS differential positioning module, it generates an electronic map. The electronic map is then transmitted to a remote control platform through an intelligent control module for path planning.

[0034] The intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle travels along a planned path. By integrating an RTK-GNSS differential positioning module, an IMU inertial navigation unit, and a wheel speed encoder, it generates the vehicle's position and posture during travel and compensates for positioning drift caused by vehicle bumps. It also detects obstacles and slope inclination in the direction of travel using lidar and a binocular vision camera.

[0035] The feasible speed space of the intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle is evaluated in real time based on the dynamic window method. Based on the feasible speed space, obstacles and slope inclination, the driving path of the intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle is dynamically adjusted and the optimal path is selected.

[0036] The intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle collects movement / stop signals through wheel speed encoders. Based on the movement / stop signals, the ground-penetrating radar detection module and transient electromagnetic detection module are controlled to synchronously perform / stop data acquisition according to the optimal path. The collected data is transmitted to the remote control platform through a wireless communication unit.

[0037] Furthermore, the dynamic window method selects the optimal speed pair for the tracked vehicle used in the rapid diagnosis of hidden dangers in intelligent all-terrain embankments through a multi-objective weighted evaluation function, including linear velocity. and angular velocity , represented as:

[0038] ;

[0039] In the formula, This represents the cost function used to evaluate linear velocity. and angular velocity The overall performance of the vehicle during operation; Indicates the heading angle term; Indicates the distance to obstacles; Indicates the velocity term; This represents the weighting coefficient of the heading angle term, used to adjust the proportion of the heading angle term in the cost function; This represents the weighting coefficient of the obstacle distance term, used to adjust the proportion of the obstacle distance term in the cost function; This represents the weighting coefficient of the velocity term, used to adjust the proportion of the velocity term in the cost function.

[0040] Furthermore, the collected data is transmitted to the data fusion processing system for processing. When a potential hazard is detected in the target embankment section's work area, the multimodal navigation system outputs the hazard's location coordinates and generates a hazard distribution map. The specific process is as follows:

[0041] Collect radar data and transient electromagnetic data for geological structure detection, and preprocess the collected radar data and transient electromagnetic data for geological structure detection.

[0042] Based on prior information or experience, set the initial resistivity. and dielectric constant Distribution, based on initial resistivity and dielectric constant Distributed forward modeling;

[0043] Based on the resistivity in the forward model Distribution, generating forward transient electromagnetic response data; based on the dielectric constant in the forward model. Distribution, generating forward-modeled ground-penetrating radar response data;

[0044] The forward-modeled transient electromagnetic response data were compared with the acquired transient electromagnetic data from geological structure exploration, and the electromagnetic data fitting difference was calculated. The forward-modeled ground-penetrating radar response data is compared with the acquired geological structure detection radar data to calculate the radar data fitting difference. ;

[0045] Based on the current resistivity and dielectric constant Distribution, calculation of cross gradient constraint terms :

[0046] ;

[0047] In the formula, For gradient operators;

[0048] Will , and Combining these elements, we construct the objective function for a joint inversion model based on a Bayesian framework. :

[0049] ;

[0050] In the formula, Represents model parameters, including resistivity and dielectric constant α, β, and λ are the weighting coefficients of each item;

[0051] Gradient descent is used to minimize the objective function. To optimize the inversion model parameters, iteratively update the inversion model parameters:

[0052] ;

[0053] In the formula, This represents the model parameter vector after the (k+1)th iteration; Indicates the learning rate;

[0054] After each iteration, check whether the rate of change of the objective function is less than the preset rate of change or whether the number of iterations has reached the preset maximum value. If yes, stop the iteration; otherwise, continue iterating to update the model parameters.

[0055] By combining the optimized model parameters with the digital elevation model, a hazard distribution map is generated, and the hazard areas are visualized and annotated in three dimensions.

[0056] Compared with existing technologies, the present invention has the following advantages:

[0057] (1) This invention integrates four-wheel independent suspension track structure, dual-modal collaborative detection, real-time data transmission and inversion and other technologies into the vehicle body, which significantly expands the detection range of tracked vehicles, improves the accuracy and reliability of hidden danger identification, shortens the time interval between on-site diagnosis and emergency response, systematically solves the bottleneck problems of existing technologies in terms of terrain adaptability, detection accuracy and timeliness, and greatly improves the comprehensiveness, accuracy and emergency response speed of embankment hidden danger diagnosis, and has significant technological advancement and engineering practical value.

[0058] (2) The present invention can control the tracked vehicle to perform path planning, autonomous obstacle avoidance and collaborative detection through a multimodal navigation system, intelligent control module and remote control platform set on the tracked vehicle, realizing intelligent and automated detection of dike hazards in complex environments and reducing the consumption of manpower and material resources; by collecting multi-physical field data to complement each other, the efficiency of hazard detection operation and the accuracy of results are significantly improved.

[0059] (3) This invention can push the location coordinates of hidden dangers and the distribution map of hidden dangers to the rear command center, so that experts and emergency rescue teams can grasp the state of the dike, quickly determine the area where hidden dangers exist, and carry out emergency treatment on the dike section with hidden dangers in a timely manner, so as to realize the rapid diagnosis and emergency response of dike hidden dangers. Attached Figure Description

[0060] Figure 1 This is an exterior view of the tracked vehicle for rapid diagnosis of hidden dangers in intelligent all-terrain embankments according to the present invention.

[0061] Figure 2 This is an internal structural diagram of the tracked vehicle for rapid diagnosis of hidden dangers in intelligent all-terrain embankments according to the present invention.

[0062] Figure 3 This is a diagram of the internal rear structure of the tracked vehicle for rapid diagnosis of hidden dangers in intelligent all-terrain embankments according to the present invention.

[0063] Figure 4 This is a flowchart of the method of the present invention.

[0064] Figure 5 This is a schematic diagram of the tracked vehicle for rapid diagnosis of hidden dangers in embankments according to the present invention, in real-time operation.

[0065] In the diagram, 110 is a four-wheel independent suspension track structure; 120 is a lithium battery pack; 210 is a radar antenna; 310 is an electromagnetic coil; 330 is a tilting support arm; 410 is an integrated RTK-GNSS differential positioning module; 420 is a lidar; 430 is a binocular vision camera; 440 is an IMU inertial navigation unit; 450 is a wheel speed encoder; 511 is a multi-band router; 512 is a 5G communication module; and 520 is a central control unit. Detailed Implementation

[0066] like Figures 1-3 As shown, the present invention provides a technical solution: an intelligent all-terrain tracked vehicle for rapid diagnosis of embankment hazards, comprising:

[0067] Vehicle body.

[0068] The four-wheel independent suspension track structure 110 is located at the bottom of the vehicle body and is used to drive the vehicle body to move; the four-wheel independent suspension track structure 110 can adapt to complex terrain with a slope of ≤30°.

[0069] The power system, located within the vehicle body, is used to transmit power to the four-wheel independent suspension track structure.

[0070] The ground-penetrating radar detection module is installed on the underside of the vehicle body, close to the ground, and is used to collect radar data for geological structure detection.

[0071] A transient electromagnetic detection module, which is installed on the vehicle body, is used to collect transient electromagnetic data for geological structure detection.

[0072] A multimodal navigation system, installed on the vehicle body, is used to generate high-precision environmental maps in real time and plan driving routes. During operation, it can photograph the surrounding environment, detect obstacles, and autonomously avoid obstacles, thus enabling autonomous operation in complex environments.

[0073] The intelligent control module, located within the vehicle body, is used to receive and respond to signals (for vehicle motion control based on a multimodal navigation system and for transmitting collected data); the collected data includes geological structure detection radar data and geological structure detection transient electromagnetic data.

[0074] The radar antenna 210 of the ground-penetrating radar detection module is installed on the top of the vehicle body. The radar antenna 210 can be the GX160 high dynamic shielded antenna from the Swedish company MALA. The transmission cable of the ground-penetrating radar detection module can be the RG-214 / U double-shielded cable.

[0075] The electromagnetic coil 310 of the transient electromagnetic detection module is mounted on the front of the vehicle body via a flip-up support arm 330; the electromagnetic coil 310 can be a high-current multi-turn loop coil with a diameter of 60cm; the communication cable of the transient electromagnetic detection module can be a high-protection cable of model LMR-400-DB; the flip-up support arm 330 can be made of carbon fiber.

[0076] The multimodal navigation system includes:

[0077] An integrated RTK-GNSS differential positioning module 410 is installed in the vehicle body at the rear end to provide global coordinates; the integrated RTK-GNSS differential positioning module 410 can be the Hexin Xingtong UM960 model.

[0078] The lidar 420 is installed inside the vehicle body and is used to scan the environment in real time to generate three-dimensional point cloud data; the lidar 420 can be the Fashi lidar S30.

[0079] A binocular vision camera 430 is mounted on the front of the vehicle body and is used to capture real-time images; the binocular vision camera 430 can be a Stereolalabs ZED 2 binocular vision stereo camera.

[0080] The IMU inertial navigation unit 440 is installed inside the vehicle body and is used to realize vehicle attitude measurement, navigation positioning and motion monitoring; the IMU inertial navigation unit 440 can be the VectorNav VN-100 inertial navigation system.

[0081] Wheel speed encoder 450 is mounted on the wheel set of the four-wheel independent suspension track structure 110 and is used to measure the wheel speed; the wheel speed encoder 450 can be an Omron E6B2-CWZ3E incremental rotary encoder.

[0082] The intelligent control module includes:

[0083] A wireless communication unit is installed inside the vehicle body and is used to receive or transmit signals; the wireless communication unit may be a SIM800L module.

[0084] The central control unit 520 is located inside the vehicle body and is connected to the wireless communication unit for receiving signals transmitted by the wireless communication unit and responding accordingly. The central control unit 520 may be a Siemens S7-1200 PLC.

[0085] The wireless communication unit includes a multi-band router 511 and a 5G communication module 512 installed on the top of the vehicle body. The multi-band router 511 and the 5G communication module 512 are used for wireless connection to receive and transmit signals. The multi-band router 511 can be a TP-Link Archer AX6000 router. The 5G communication module 512 can use a Qualcomm Snapdragon X55 chip.

[0086] This includes a 120-cell lithium battery pack for powering the tracked vehicle used for rapid diagnosis of hidden dangers in intelligent all-terrain embankments, with a range of ≥8 hours, an operating speed of 5.4 km / h, and an operating efficiency of 5000 m³ / h. 2 / h, meeting the detection efficiency requirements.

[0087] This also includes a remote control platform for controlling tracked vehicles used for rapid diagnosis of potential hazards in intelligent all-terrain embankments.

[0088] The remote control platform integrates an intelligent vehicle remote control system, a data fusion processing system, and a display device. The intelligent vehicle remote control system controls the intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle. The data fusion processing system preprocesses the collected data, performs collaborative inversion calculations, and outputs the hazard area. The display device displays the status information of the intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle, an electronic map marking the vehicle's location, real-time collected data, and hazard detection results. The display device also supports interactive operation. The intelligent vehicle remote control system can be a Raspberry Pi 4 computer, and the data fusion processing system can be an NVIDIA Jetson embedded system.

[0089] The central control unit 520 is electrically connected to the ground-penetrating radar detection module, transient electromagnetic detection module, integrated RTK-GNSS differential positioning module 410, lidar 420, binocular vision camera 430, IMU inertial navigation unit 440, and wheel speed encoder 450.

[0090] The vehicle body is made of carbon steel, and the four-wheel independent suspension track structure 110 is equipped with anti-slip rubber tracks and a hydraulic leveling module for leveling the four-wheel independent suspension track structure 110.

[0091] Among them, the tracked vehicle, which is based on a multimodal navigation system, intelligent control module and remote control platform, realizes all-terrain path planning, autonomous obstacle avoidance and collaborative detection of the intelligent all-terrain embankment hidden danger rapid diagnosis vehicle.

[0092] like Figure 4 As shown, a method for operating a tracked vehicle for rapid diagnosis of hidden dangers in intelligent all-terrain embankments includes the following steps:

[0093] Step S1: The intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle arrives at the target embankment section operation area, such as... Figure 5 As shown.

[0094] Step S2: The intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle performs all-terrain path planning, autonomous obstacle avoidance, and collaborative detection in the target embankment section operation area.

[0095] Step S3: The geological structure detection radar data and geological structure detection transient electromagnetic data collected by the detection are transmitted to the data fusion processing system for processing. When a hidden danger is detected in the target embankment section operation area, the location coordinates of the hidden danger are output through the multi-modal navigation system, and a hidden danger distribution map is generated.

[0096] Step S4: Push the location coordinates and distribution map of the hidden danger to the command center. Based on the pushed location coordinates and distribution map of the hidden danger, the command center determines the area where the hidden danger exists and promptly carries out emergency treatment on the section of the embankment with hidden danger.

[0097] The specific process of the intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle performing all-terrain path planning, autonomous obstacle avoidance, and collaborative detection in the target embankment section operation area is as follows:

[0098] 1. The intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle scans the environment in real time using a LiDAR 420 to generate three-dimensional point cloud data. Combined with the global coordinates provided by the integrated RTK-GNSS differential positioning module 410, it generates a high-precision electronic map. The electronic map is then transmitted to a remote control platform via an intelligent control module, and path planning is performed (setting parameters such as detection route, detection spacing, and speed).

[0099] 2. The intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle travels along the planned path. By integrating the RTK-GNSS differential positioning module 410, IMU inertial navigation unit 440, and wheel speed encoder 450, the vehicle's position and posture during travel are generated, and the positioning drift caused by the vehicle's bumps is compensated. The LiDAR 420 and binocular vision camera 430 detect obstacles and slope inclination in the direction of travel. The LiDAR 420 detects obstacles at higher positions (height > 20cm), and the binocular vision camera 430 detects obstacles at lower positions (height < 20cm) and slope inclination.

[0100] 3. The feasible speed space of the intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle is evaluated in real time based on the dynamic window method (DWA). Based on the feasible speed space, obstacles and slope inclination, the driving path of the intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle is dynamically adjusted and the optimal path is selected to ensure that the intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle travels along the planned path.

[0101] Among them, the Dynamic Window Method (DWA) selects the optimal speed pair (linear velocity) for the tracked vehicle for rapid diagnosis of hidden dangers in intelligent all-terrain embankments through a multi-objective weighted evaluation function. and angular velocity ), represented as:

[0102] ;

[0103] In the formula, This represents the cost function, used to evaluate different combinations of speeds (linear velocity). and angular velocity The overall performance of the vehicle under these speed combinations is represented by a smaller value, indicating a better combination of speeds. Indicates the heading angle term; Indicates the distance to obstacles; Indicates the velocity term; This represents the weighting coefficient of the heading angle term, used to adjust the proportion of the heading angle term in the cost function; This represents the weighting coefficient of the obstacle distance term, used to adjust the proportion of the obstacle distance term in the cost function; The weighting coefficient for the velocity term is used to adjust its proportion in the cost function; in this embodiment, , , .

[0104] 4. The wheel speed encoder 450 collects the movement / stop signals of the tracked vehicle for rapid diagnosis of hidden dangers in intelligent all-terrain embankments. Based on the movement / stop signals, the ground-penetrating radar detection module and transient electromagnetic detection module are controlled to synchronously start / stop data acquisition according to the optimal path (or commands can be issued manually through the intelligent vehicle remote control system). The collected data is transmitted to the remote control platform through the wireless communication unit. The operation sequence of the ground-penetrating radar detection module and transient electromagnetic detection module is aligned with GPS timing (time deviation ≤ 1ms). The data timestamp is bound to the positioning coordinates of the integrated RTK-GNSS differential positioning module 410 (the installation distance between the ground-penetrating radar detection module and transient electromagnetic detection module is ≥ 2m) to ensure spatial consistency.

[0105] The process involves transmitting the collected data (geological structure detection radar data and geological structure detection transient electromagnetic data) to a data fusion processing system for processing. When a potential hazard is detected in the target embankment section's work area, the multi-modal navigation system outputs the hazard's location coordinates and generates a hazard distribution map.

[0106] 1. Data collection and preprocessing.

[0107] Collect radar data and transient electromagnetic data for geological structure detection. Perform preprocessing on the collected radar data and transient electromagnetic data for geological structure detection, including time synchronization, spatial registration and noise suppression.

[0108] 2. Initial model establishment.

[0109] Based on prior information or experience, set the initial resistivity. and dielectric constant Distribution, based on initial resistivity and dielectric constant A forward model is constructed using a distributed approach to generate forward response data.

[0110] 3. Forward modeling.

[0111] Based on the resistivity in the forward model Distribution, generating forward transient electromagnetic response data; based on the dielectric constant in the forward model. Distribution, generating forward modeling ground-penetrating radar response data.

[0112] 4. Calculate the fit difference.

[0113] The forward-modeled transient electromagnetic response data were compared with the acquired transient electromagnetic data from geological structure exploration, and the electromagnetic data fitting difference was calculated. The forward-modeled ground-penetrating radar response data is compared with the acquired geological structure detection radar data to calculate the radar data fitting difference. .

[0114] 5. Calculation of cross gradient constraints.

[0115] Based on the current resistivity and dielectric constant Distribution, calculation of cross gradient constraint terms :

[0116] ;

[0117] In the formula, This is the gradient operator.

[0118] 6. Construct the objective function.

[0119] Will , and Combining these elements, we construct the objective function for a joint inversion model based on a Bayesian framework. :

[0120] ;

[0121] In the formula, Represents model parameters, including resistivity and dielectric constant α=0.6, β=0.3, and λ=0.1 are the weight coefficients of each item, which are determined through cross-validation of the model data.

[0122] 7. Optimize model parameters.

[0123] Gradient descent is used to minimize the objective function. To optimize the inversion model parameters, iteratively update the inversion model parameters according to the following formula:

[0124] ;

[0125] In the formula, This represents the model parameter vector after the (k+1)th iteration; This represents the learning rate.

[0126] After each iteration, check whether the rate of change of the objective function is less than 1% or whether the number of iterations has reached the maximum value (e.g., 50 times). If the termination condition is met, stop the iteration; otherwise, continue iterating to update the model parameters.

[0127] 8. Output the inversion results.

[0128] The optimized model parameters (resistivity) and dielectric constant By combining it with a digital elevation model (DEM), a hazard distribution map is generated, and hazard areas are visualized and annotated in three dimensions.

[0129] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A tracked vehicle for rapid diagnosis of hidden dangers in intelligent all-terrain embankments, characterized in that, include: Vehicle body; The four-wheel independent suspension track structure is located at the bottom of the vehicle body and is used to move the vehicle body. The power system, which is located inside the vehicle body, is used to transmit power to the four-wheel independent suspension track structure; A ground-penetrating radar detection module is installed on the underside of the vehicle body and is used to collect radar data for geological structure detection. A transient electromagnetic detection module, which is installed on the vehicle body, is used to collect transient electromagnetic data for geological structure detection; A multimodal navigation system, which is mounted on the vehicle body, is used to generate environmental maps in real time, take real-time photos, and detect obstacles. The intelligent control module, located within the vehicle body, is used to receive and respond to signals; the signal response includes vehicle motion control based on a multimodal navigation system and transmission of collected data; the collected data includes geological structure detection radar data and geological structure detection transient electromagnetic data; The remote control platform is used to control the intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle, receive collected data, preprocess the collected data, perform collaborative inversion calculations, and output the hazard area. The multimodal navigation system includes: An integrated RTK-GNSS differential positioning module is installed inside the vehicle body to provide global coordinates. A lidar, installed inside the vehicle, is used to scan the environment in real time and generate three-dimensional point cloud data; A binocular vision camera, which is mounted on the front of the vehicle body, is used to capture real-time images; An IMU (Inertial Measurement Unit) is installed inside the vehicle body to perform vehicle attitude measurement, navigation and positioning, and motion monitoring. A wheel speed encoder, which is mounted on the wheel set of a four-wheel independent suspension track structure, is used to measure the wheel set rotation speed; The intelligent control module includes: A wireless communication unit, which is installed inside the vehicle body, is used to receive or transmit signals; The central control unit is located inside the vehicle body and is connected to the wireless communication unit to receive signals transmitted by the wireless communication unit and respond accordingly. The remote control platform integrates an intelligent vehicle remote control system, a data fusion processing system, and a display device; the intelligent vehicle remote control system is used to control an intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle; the data fusion processing system is used to preprocess the collected data, then perform collaborative inversion calculations, and output the hazard area. The intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle scans the environment in real time with LiDAR to generate three-dimensional point cloud data. Combined with the global coordinates provided by the integrated RTK-GNSS differential positioning module, it generates an electronic map. The electronic map is then transmitted to a remote control platform through an intelligent control module for path planning. The intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle travels along a planned path. By integrating an RTK-GNSS differential positioning module, an IMU inertial navigation unit, and a wheel speed encoder, it generates the vehicle's position and posture during travel and compensates for positioning drift caused by vehicle bumps. It also detects obstacles and slope inclination in the direction of travel using lidar and a binocular vision camera. The feasible speed space of the intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle is evaluated in real time based on the dynamic window method. Based on the feasible speed space, obstacles and slope inclination, the driving path of the intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle is dynamically adjusted and the optimal path is selected. The intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle collects movement / stop signals by wheel speed encoder. Based on the movement / stop signals, the ground radar detection module and transient electromagnetic detection module are controlled to synchronously perform / stop data acquisition according to the optimal path. The collected data is transmitted to the remote control platform through wireless communication unit. The dynamic window method selects the optimal speed pair for the tracked vehicle used in the rapid diagnosis of hidden dangers in intelligent all-terrain embankments through a multi-objective weighted evaluation function, including linear velocity. and angular velocity , represented as: ； In the formula, This represents the cost function used to evaluate linear velocity. and angular velocity The overall performance of the vehicle during operation; Indicates the heading angle term; Indicates the distance to obstacles; Indicates the velocity term; This represents the weighting coefficient of the heading angle term, used to adjust the proportion of the heading angle term in the cost function; This represents the weighting coefficient of the obstacle distance term, used to adjust the proportion of the obstacle distance term in the cost function; The weighting coefficient for the velocity term is used to adjust the proportion of the velocity term in the cost function. The data fusion processing system preprocesses the collected radar data and transient electromagnetic data from geological structure detection. Based on prior information or experience, set the initial resistivity. and dielectric constant Distribution, based on initial resistivity and dielectric constant Distributed forward modeling; Based on the resistivity in the forward model Distribution, generating forward transient electromagnetic response data; based on the dielectric constant in the forward model. Distribution, generating forward-modeled ground-penetrating radar response data; The forward-modeled transient electromagnetic response data were compared with the acquired transient electromagnetic data from geological structure exploration, and the electromagnetic data fitting difference was calculated. The forward-modeled ground-penetrating radar response data is compared with the acquired geological structure detection radar data to calculate the radar data fitting difference. ; Based on the current resistivity and dielectric constant Distribution, calculation of cross gradient constraint terms : ； In the formula, For gradient operators; Will , and Combining, constructing the objective function of the joint inversion model based on the Bayesian framework : ； In the formula, Represents model parameters, including resistivity and dielectric constant α, β, and λ are the weighting coefficients of each term; Gradient descent is used to minimize the objective function. To optimize the inversion model parameters, iteratively update the inversion model parameters: ； In the formula, This represents the model parameter vector after the (k+1)th iteration; Indicates the learning rate; After each iteration, check whether the rate of change of the objective function is less than the preset rate of change or whether the number of iterations has reached the preset maximum value. If yes, stop the iteration; otherwise, continue iterating to update the model parameters. By combining the optimized model parameters with the digital elevation model, a hazard distribution map is generated, and the hazard areas are visualized and annotated in three dimensions.

2. The intelligent all-terrain tracked vehicle for rapid diagnosis of embankment hazards according to claim 1, characterized in that: The wireless communication unit includes a multi-band router and a 5G communication module installed on the top of the vehicle body, which are used to receive and transmit signals.

3. The intelligent all-terrain tracked vehicle for rapid diagnosis of embankment hazards according to claim 2, characterized in that: The central control unit is electrically connected to the ground-penetrating radar detection module, transient electromagnetic detection module, integrated RTK-GNSS differential positioning module, lidar, binocular vision camera, IMU inertial navigation unit, and wheel speed encoder.

4. A method for operating a tracked vehicle for rapid diagnosis of hidden dangers in intelligent all-terrain embankments as described in any one of claims 1-3, characterized in that, Includes the following steps: Step S1: The intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle arrives at the target embankment section operation area; Step S2: The intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle performs all-terrain path planning, autonomous obstacle avoidance, and collaborative detection in the target embankment section operation area; Step S3: The geological structure detection radar data and geological structure detection transient electromagnetic data collected by the detection are transmitted to the data fusion processing system for processing. When a hidden danger is detected in the target embankment section operation area, the location coordinates of the hidden danger are output through the multi-modal navigation system, and a hidden danger distribution map is generated. Step S4: Push the location coordinates and distribution map of the hidden danger to the command center. Based on the pushed location coordinates and distribution map of the hidden danger, the command center determines the area where the hidden danger exists and promptly carries out emergency treatment on the section of the embankment with hidden danger.

5. The operating method of a tracked vehicle for rapid diagnosis of hidden dangers in intelligent all-terrain embankments according to claim 4, characterized in that: The specific process of the intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle performing all-terrain path planning, autonomous obstacle avoidance, and collaborative detection in the target embankment section operation area is as follows: The intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle scans the environment in real time with LiDAR to generate three-dimensional point cloud data. Combined with the global coordinates provided by the integrated RTK-GNSS differential positioning module, it generates an electronic map. The electronic map is then transmitted to a remote control platform through an intelligent control module for path planning. The intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle travels along a planned path. By integrating an RTK-GNSS differential positioning module, an IMU inertial navigation unit, and a wheel speed encoder, it generates the vehicle's position and posture during travel and compensates for positioning drift caused by vehicle bumps. It also detects obstacles and slope inclination in the direction of travel using lidar and a binocular vision camera. The feasible speed space of the intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle is evaluated in real time based on the dynamic window method. Based on the feasible speed space, obstacles and slope inclination, the driving path of the intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle is dynamically adjusted and the optimal path is selected. The intelligent all-terrain embankment hazard rapid diagnosis tracked vehicle collects movement / stop signals through wheel speed encoders. Based on the movement / stop signals, the ground-penetrating radar detection module and transient electromagnetic detection module are controlled to synchronously perform / stop data acquisition according to the optimal path. The collected data is transmitted to the remote control platform through a wireless communication unit.

6. The operating method of a tracked vehicle for rapid diagnosis of hidden dangers in intelligent all-terrain embankments according to claim 5, characterized in that: The dynamic window method selects the optimal speed pair for the tracked vehicle used in the rapid diagnosis of hidden dangers in intelligent all-terrain embankments through a multi-objective weighted evaluation function, including linear velocity. and angular velocity , represented as: ； In the formula, This represents the cost function used to evaluate linear velocity. and angular velocity The overall performance of the vehicle during operation; Indicates the heading angle term; Indicates the distance to obstacles; Indicates the velocity term; This represents the weighting coefficient of the heading angle term, used to adjust the proportion of the heading angle term in the cost function; This represents the weighting coefficient of the obstacle distance term, used to adjust the proportion of the obstacle distance term in the cost function; This represents the weighting coefficient of the velocity term, used to adjust the proportion of the velocity term in the cost function.

7. The operating method of a tracked vehicle for rapid diagnosis of hidden dangers in intelligent all-terrain embankments according to claim 6, characterized in that: The data collected during detection is transmitted to the data fusion processing system for processing. When a potential hazard is detected in the target embankment section's work area, the multimodal navigation system outputs the hazard's location coordinates and generates a hazard distribution map. The specific process is as follows: Collect radar data and transient electromagnetic data for geological structure detection, and preprocess the collected radar data and transient electromagnetic data for geological structure detection. Based on prior information or experience, set the initial resistivity. and dielectric constant Distribution, based on initial resistivity and dielectric constant Distributed forward modeling; Based on the resistivity in the forward model Distribute and generate forward-modeled transient electromagnetic response data; Based on the dielectric constant in the forward model Distribution, generating forward-modeled ground-penetrating radar response data; The forward-modeled transient electromagnetic response data were compared with the acquired transient electromagnetic data from geological structure exploration, and the electromagnetic data fitting difference was calculated. ; The forward-modeled ground-penetrating radar response data is compared with the acquired geological structure detection radar data to calculate the radar data fitting difference. ; Based on the current resistivity and dielectric constant Distribution, calculation of cross gradient constraint terms : ； In the formula, For gradient operators; Will , and Combining, constructing the objective function of the joint inversion model based on the Bayesian framework : ； In the formula, Represents model parameters, including resistivity and dielectric constant α, β, and λ are the weighting coefficients of each item; Gradient descent is used to minimize the objective function. To optimize the inversion model parameters, iteratively update the inversion model parameters: ； In the formula, This represents the model parameter vector after the (k+1)th iteration; Indicates the learning rate; After each iteration, check whether the rate of change of the objective function is less than the preset rate of change or whether the number of iterations has reached the preset maximum value. If yes, stop the iteration; otherwise, continue iterating to update the model parameters. By combining the optimized model parameters with the digital elevation model, a hazard distribution map is generated, and the hazard areas are visualized and annotated in three dimensions.

Images