Lightweight tracking detection system and method for submarine electromagnetic cable

Abstract

The invention discloses a lightweight tracking detection system and method for a submarine electromagnetic cable. According to the method, firstly, an autonomous underwater robot is placed in a proper sea area and surrounds the sea area by a circle, a grid map is established for the sea area, electromagnetic intensity detection is carried out on the detected sea area by utilizing an electromagnetic inductor while the map is established, detection is started from places with high electromagnetic intensity, areas without electromagnetic induction are eliminated, and time is saved. After a global map is generated, the electromagnetic cable is detected, a path is planned, the electromagnetic cable is tracked after the optimal path is planned, the longitude and latitude of the electromagnetic cable are synchronized when the electromagnetic sensor detects the electromagnetic cable, and when the visual sensor detects a damaged part, the damaged part is quantitatively expressed. According to the invention, the autonomous underwater robot can work in the seabed for a long time, and the working efficiency of seabed electromagnetic cable detection is remarkably improved.

Description

Technical Field

[0001] This invention belongs to the field of underwater unmanned system target detection, specifically involving a lightweight tracking and detection system and method for submarine electromagnetic cables. It includes functions such as optimal path planning, autonomous hibernation, damage area detection and quantification, and autonomous charging, and is applied to autonomous underwater robots. Background Technology

[0002] Submarine electromagnetic cables serve as the "main arteries" of global communication and energy transmission, and their stable operation is crucial. However, these cables, laid in complex seabed environments, are susceptible to damage from natural and human factors such as ocean activity and anchor towing, and repair costs are extremely high. Traditional underwater detection and inspection mainly rely on manual diving or remotely operated vehicles (ROVs), which are inefficient, costly, and pose safety risks to operators, especially in harsh deep-sea environments. Although existing autonomous underwater vehicles (AUVs) have attempted to detect targets using acoustic devices, acoustic technology has limited detection capabilities for cables buried beneath the seabed or with small diameters, and is susceptible to underwater noise interference, resulting in insufficient identification accuracy. Therefore, developing an intelligent submarine cable detection system capable of precise, autonomous, and long-term operation is an urgent need. Summary of the Invention

[0003] Objective: This invention proposes a lightweight tracking and detection system and method for submarine electromagnetic cables. The aim is to improve the accuracy and robustness of autonomous underwater vehicles (AUVs) in identifying electromagnetic cables in low-texture environments on the seabed, and to quantify the degree of cable damage. The core of this method lies in solving how to track and identify all electromagnetic cables within a specified sea area in a shorter time, quantify the degree of damage to damaged cables while marking the latitude and longitude information of the damage location, and activate a sleep mode to save power and computing power when no electromagnetic cables are detected, and autonomously return to a fixed charging platform to recharge when the battery is low.

[0004] Technical solution:

[0005] This invention first provides a lightweight tracking and detection method for submarine electromagnetic cables, comprising the following steps:

[0006] (1) Build a hardware system. The autonomous underwater robot is equipped with an electromagnetic sensor, a vision camera, a Doppler velocimeter and an inertial sensor; it is equipped with a lithium battery, a buoyancy drive system and a thruster.

[0007] (2) The detection of submarine electromagnetic cables begins. An autonomous underwater vehicle (AUV) is placed in the area to be detected. Once it sinks to the seabed, the AUV will begin its detection work. First, the AUV will circle the area to be detected and simultaneously activate its electromagnetic sensors to detect the electromagnetic field and create a grid map. The area is divided into grids, and each grid has an electromagnetic intensity value. Based on the electromagnetic intensity, high-intensity grids correspond to the location of electromagnetic cables, while low-intensity areas close to zero are eliminated. Then, optimal path planning is performed to calculate the optimal path for submarine cable detection, quickly locate the area where electromagnetic cables are clustered, and reduce redundant searches.

[0008] (3) A confidence model for the existence of electromagnetic cables is established by integrating data from multiple sensors. Based on the dynamic threshold, the autonomous underwater robot can automatically enter a low-power sleep state when there is no electromagnetic cable target. The confidence level must be stable, the duration must meet the standard, and the body must be stationary. When specific signals such as sudden changes in water flow or electromagnetic anomalies are detected, a graded strategy is adopted to gradually wake up the sensors and processing modules, thereby improving the endurance and work efficiency of the autonomous underwater robot during long-term underwater operations.

[0009] (4) After successfully tracing the electromagnetic cable, the automated detection and quantitative assessment of electromagnetic cable damage is initiated. First, the damage point is initially identified and confirmed by fusing the local texture anomaly index of the visual camera image and the distortion factor of the electromagnetic sensor. Then, the three-dimensional area of ​​the damaged area is quantified using stereo vision and convex hull algorithm. Next, the confidence level is introduced to assess the credibility of the measurement results. If the credibility is insufficient, a remeasurement is triggered. Finally, the damage level is calculated by combining the damaged area and the electromagnetic distortion factor to provide a priority basis for maintenance. Finally, the location information of the confirmed damage location is synchronized, and after the current point assessment is completed, it automatically moves along the electromagnetic cable for continuous detection.

[0010] (5) During the detection process, if the autonomous underwater vehicle's battery level is below 20%, the autonomous underwater vehicle will automatically navigate to the charging platform for contact charging. After charging to 100%, the autonomous underwater vehicle will automatically return to the place where the last detection was interrupted and continue to detect the electromagnetic cable. This process continues until all the electromagnetic cables in the detected sea area have been scanned and detected. The autonomous underwater vehicle will then return to the location where it first entered the water and wait for retrieval.

[0011] The present invention also provides a lightweight tracking and detection system for submarine electromagnetic cables, which implements the above-mentioned method. The system includes an electromagnetic sensor, a visual camera, a Doppler velocimeter, and an inertial sensor; and is equipped with a lithium battery, a buoyancy drive system, and a propulsion unit.

[0012] Beneficial effects:

[0013] 1. This invention breaks through the limitations of traditional acoustic detection by designing a lightweight autonomous underwater robot that integrates multimodal sensors, including electromagnetic sensors, optical cameras, inertial sensors, and Doppler velocimeters. The core detection technology employs passive electromagnetic induction, capturing the cable's own radiation through a highly sensitive electromagnetic sensor and sensing the resulting electromagnetic field signal. This allows for effective detection even when the cable is partially buried, overcoming the blind spots of purely acoustic or optical methods in turbid water or buried environments. This multi-sensor fusion framework achieves precise positioning through a combination of inertial navigation and Doppler velocimeters, while combining visual sensors to record the cable's apparent state, providing data support for subsequent damage detection. Compared to other studies, the advantages of this invention lie in its integrated and intelligent functional design. Firstly, in terms of path planning, the system does not simply preset a route but uses a magnetic signal gradient-based guidance strategy to achieve adaptive tracking of the cable in three-dimensional space, significantly improving detection efficiency and accuracy. Secondly, the system's integrated autonomous sleep and charging functions address the bottleneck of limited endurance in traditional autonomous underwater vehicles. Through energy management, it enters a low-power state during non-operational periods and can be recharged at the charging platform, extending continuous operating time. Furthermore, the designed damage area detection and quantification function, through image recognition and electromagnetic signal anomaly analysis, can automatically identify cable damage points and assess the degree of damage, achieving a leap from "detection" to "diagnosis," facilitating quick and targeted repairs by maintenance personnel. This system is an intelligent operation and maintenance platform integrating detection, diagnosis, and maintenance.

[0014] 2. This autonomous underwater vehicle (AUV) is capable of fully autonomous tracking and damage detection of submarine electromagnetic cables. Its designed electromagnetic grid-optimized path tracking method efficiently tracks the cables; moreover, the design accurately identifies damaged sections of the cable, quantifies them, and marks the latitude and longitude of the damaged area, facilitating repair personnel in locating the damage and carrying appropriate repair materials. Furthermore, the design includes an autonomous contact charging platform, ensuring the AUV can operate long-term on the seabed, significantly improving the efficiency of submarine electromagnetic cable inspection. Attached Figure Description

[0015] Figure 1 This is a system flowchart of the present invention; Figure 2 Schematic diagram for tracking and detecting electromagnetic cables; Figure 3 A diagram showing the trajectory of an autonomous underwater robot tracking a cable. Figure 4 A diagram showing the trajectory of an autonomous underwater robot connected to a charging platform. Figure 5 A 3D trajectory map of the driving path when connecting to the charging platform. Detailed Implementation

[0016] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0017] As shown in Figure 1, a lightweight tracking and detection method for submarine electromagnetic cables includes: establishing a grid map of the detected sea area; planning the optimal path; tracking the electromagnetic cable; locating the damaged parts and quantifying the degree of damage; and autonomous hibernation and recharging of an autonomous underwater vehicle. Specifically, it includes the following steps:

[0018] 1. After debugging the autonomous underwater vehicle (AUV), place the AUV on the surface of the water in the area to be explored. Once the AUV sinks into the water, the preliminary preparations are complete and the exploration work is ready to begin.

[0019] 2. The autonomous underwater robot begins its underwater exploration mission by first circling the area to be explored to create a grid map of the area and simultaneously collecting electromagnetic intensity data.

[0020] Based on the collected electromagnetic intensity data and the divided grid map, the electromagnetic intensity value of each grid area is calculated. Each grid is categorized according to its detected electromagnetic intensity value, and a threshold is set for it. This threshold needs to be dynamically changed; grids with intensity above the threshold are retained, while areas with intensity below the threshold are removed. The retained grids are then aggregated into a grid set. Optimal path planning is then performed on the retained grids, and the optimal path for submarine cable detection is calculated using the path cost formula. This allows for rapid location of electromagnetic cable clusters and reduces redundant searches.

[0021] 3. The autonomous hibernation mode is used throughout the submarine cable detection process. The purpose is that when the autonomous underwater robot fails to detect the submarine cable or encounters other special circumstances, the autonomous underwater robot can autonomously adjust to the hibernation mode, effectively reducing the robot's power consumption and laying the foundation for the autonomous underwater robot to work underwater for a long time.

[0022] First, an electromagnetic cable existence confidence model is established by combining electromagnetic detection data, visual features, and motion status, and the feature vector is adjusted in real time using dynamic weighting coefficients and environmental adaptive algorithms. Next, a spatiotemporal correlation function is defined to calculate the confidence level within a time window and is updated online based on historical templates. Then, a dynamic threshold adjustment mechanism is introduced, involving environmental noise attenuation factors and visual contribution coefficients, to set sleep trigger conditions. These conditions require that the confidence level be within the threshold, the duration exceed the adaptive idle time, and the motion status be below the threshold, thereby entering a low-power sleep state. At the same time, a multi-level wake-up strategy is adopted. When the Doppler velocimeter detects a sudden change in flow velocity, a primary wake-up is triggered to switch to shallow sleep. If an electromagnetic anomaly is detected, a deep wake-up is performed to efficiently manage energy consumption.

[0023] 4. After successfully tracing the electromagnetic cable, the electromagnetic cable damage detection and quantification module begins to work. Based on multi-sensor fusion, it moves along the electromagnetic cable, maintains the optimal detection distance, and runs vision and electromagnetic detection algorithms simultaneously. When damage is detected, the autonomous underwater robot hovers and performs a fine scan. It synchronizes the location information of the damage point, stores the damage image, electromagnetic data, area quantification results, and generates the damage level.

[0024] First, the probability of damage is initially determined by analyzing the local texture anomaly index of the images captured by the camera and the electromagnetic distortion factor detected by the electromagnetic sensor. The basic probability allocation function is then used to fuse and confirm the visual and electromagnetic detection results. After confirming the damage, the system uses stereo vision measurement and coordinate transformation formulas to calculate the actual coordinates of the damaged area in three-dimensional space and quantifies the damaged area. At the same time, the confidence level of the area measurement is considered for depth measurement error and pixel error. If it is lower than the threshold, a remeasurement is triggered. Finally, the system calculates the damage level by combining the damaged area and the electromagnetic distortion factor, and records the geographical location information in time to plan the repair sequence. After the current segment is judged, the system automatically advances to achieve full coverage detection of electromagnetic cables.

[0025] 5. Simultaneously, this autonomous underwater robot system is equipped with an autonomous charging function. By monitoring the battery level in real time, it activates the navigation system when the battery level drops below 20%. It utilizes low-frequency electromagnetic pulse transmitters and a ring LED array deployed around the charging platform for enhanced detection and optical-assisted positioning, combined with an adaptive noise suppression module to optimize the signal. Subsequently, it employs an extended Kalman filter framework that fuses multi-source information, integrating electromagnetic field strength, optical characteristics, Doppler frequency shift, and inertial sensor data to achieve a switch from global navigation to precise positioning and docking. First, it uses electromagnetic guidance to coarsely align with the platform entrance, and then uses optical visual servoing for fine correction at close range, finally completing docking and charging. After charging, the autonomous underwater robot constructs a topology map based on a path memory algorithm and returns along the original route, adjusting the path in real time to cope with environmental changes. It continues to detect the remaining electromagnetic cables until the mission is completed, and finally returns to the starting point to await recovery.

[0026] This invention discloses a lightweight tracking and detection system and method for submarine electromagnetic cables. In complex seabed environments, it enables a complete process for tracking and detecting electromagnetic cables. Unlike previous electromagnetic cable detection methods, the designed system can track every electromagnetic cable in the detected sea area in a shorter time and with lower power consumption. It also quantifies the degree of damage at the point of breakage. The new system further includes an autonomous underwater vehicle (AUV) for self-charging, effectively improving the repair efficiency of submarine electromagnetic cables and enabling long-term underwater operations for the AUV. The system includes the following steps:

[0027] (1) A lightweight system for tracking and detecting submarine electromagnetic cables is equipped with an electromagnetic sensor, a visual camera, a Doppler velocimeter and an inertial sensor; it is equipped with a lithium battery, a buoyancy drive system and a thruster;

[0028] (2) Submarine electromagnetic cable detection begins. The autonomous underwater vehicle (AUV) is placed in the required detection area and allowed to sink to the seabed before commencing detection. First, the AUV circles the area while simultaneously activating its electromagnetic sensors to detect electromagnetic activity and create a grid map. The area is divided into grids, each with an electromagnetic intensity value. Based on the intensity, high-intensity grids correspond to the location of electromagnetic cables, while low-intensity areas close to zero are eliminated. Then, optimal path planning is performed to calculate the optimal path for cable detection, quickly locating areas where electromagnetic cables are clustered and reducing redundant searches.

[0029] (3) This invention designs a novel autonomous hibernation system. Hibernation runs through the entire detection process. Its core is to integrate multi-sensor data to establish a confidence model of the existence of electromagnetic cables and judge based on dynamic thresholds to enable the autonomous underwater robot to automatically enter a low-power hibernation state when there are no electromagnetic cable targets. It needs to simultaneously meet the conditions of stable confidence, sufficient duration and static body. When specific signals such as sudden changes in water flow or electromagnetic anomalies are detected, a graded strategy is adopted to gradually wake up the sensors and processing modules, thereby improving the endurance and work efficiency of the autonomous underwater robot for long-term underwater operations.

[0030] (4) After successfully tracing the electromagnetic cable, the automated detection and quantitative assessment of electromagnetic cable damage is initiated. First, the damage point is initially identified and confirmed by fusing the local texture anomaly index of the visual camera image and the distortion factor of the electromagnetic sensor. Then, the three-dimensional area of ​​the damaged area is quantified using stereo vision and convex hull algorithm. Next, the confidence level is introduced to assess the reliability of the measurement results. If the reliability is insufficient, a remeasurement is triggered. Finally, the damage level is calculated by combining the damaged area and the electromagnetic distortion factor to provide a priority basis for maintenance. Finally, the location information of the confirmed damage location is synchronized, and after the current point assessment is completed, it automatically moves along the electromagnetic cable for continuous detection.

[0031] (5) During the detection process, if the autonomous underwater vehicle's battery level is below 20%, the autonomous underwater vehicle will automatically navigate to the charging platform for contact charging. After charging to 100%, the autonomous underwater vehicle will automatically return to the place where the last detection was interrupted and continue to detect the electromagnetic cable. This process continues until all the electromagnetic cables in the detected sea area have been scanned and detected. The autonomous underwater vehicle will then return to the location where it first entered the water and wait for retrieval.

[0032] Step (2) specifically includes the following sub-steps:

[0033] (2.1) First, the autonomous underwater vehicle (AUV) circled the boundary of the sea area, collected electromagnetic intensity data, and constructed a grid map. The AUV integrated visual sensors, Doppler velocimeters, and inertial sensors to track and detect submarine electromagnetic cables.

[0034] (2.2) Simultaneously, the electromagnetic intensity of the surveyed sea area is detected. The electromagnetic signals are scanned. The electromagnetic detector continuously samples at a fixed frequency, and the measurement points are recorded as follows: Its magnetic flux density modulus is denoted as And record its position coordinates. The autonomous underwater vehicle's position is determined by fusing Doppler velocimetry and inertial sensors. At this point, it is necessary to calculate the electromagnetic intensity value of each individual grid cell. The electromagnetic intensity value is calculated based on the spatial weight of the measurement point and the attenuation effect, as shown in equation (1): (1) in This represents the total number of measurement points. For measurement points Distance to the center of the grid This is the attenuation coefficient, set based on empirical values; it can be set here as follows: This is to ensure that neighboring measurement points contribute more.

[0035] Then output the electromagnetic grid map. Each element Represents a grid Standardized electromagnetic intensity;

[0036] (2.3) Next, the divided grids are classified to identify the strong electromagnetic grids corresponding to the electromagnetic cables, and irrelevant areas with electromagnetic intensity close to zero are removed. An adaptive threshold classification model is established for this purpose. The statistical characteristics of the global electromagnetic intensity are defined, including the average intensity. and strength standard deviation As shown in equation (2): (2) in, This represents the number of rows in the raster map. This represents the number of columns in the raster map.

[0037] A threshold formula is set to distinguish between high-intensity and low-intensity grids. The threshold is dynamically generated and adapts to the electromagnetic characteristics of the sea area to avoid misjudgment caused by a fixed threshold. The high-intensity and low-intensity thresholds are set as shown in equation (3): (3) in For high intensity threshold, The grid markings indicate areas of concentrated electromagnetic cables. These concentrated areas may correspond to electromagnetic cables with coefficients... Ensures coverage of significantly strong signals, with adjustable coefficients; Low intensity threshold, The raster is removed, and the coefficient is... Filters noise, and the coefficient is adjustable;

[0038] (2.4) Next, form a set of the raster regions that meet the conditions, and extract all those that satisfy the conditions. The center point of the grid forms a concentrated area of ​​electromagnetic cables. ,in It is the first The coordinates of the concentrated area of ​​electromagnetic cables are identified, and those that meet the criteria are excluded. The grid does not participate in path planning;

[0039] After completing the above operations, the output electromagnetic cable concentration area is collected. ,size Represents the potential number of electromagnetic cables;

[0040] (2.5) Finally, shortest path planning is performed on the retained grid areas to calculate the shortest path for the autonomous underwater robot to access all areas with concentrated electromagnetic cables, minimizing the travel time. First, the path cost function is designed. Defined as the sequence of arrangements of concentrated areas of electromagnetic cables: ,in It is 1 to The arrangement.

[0041] Path cost Combining path length and electromagnetic intensity weights, strong signal areas are traversed first. The cost function here is not simply geographical distance, but rather guides the path to cover dense electromagnetic cable areas through electromagnetic weights, thereby improving efficiency. The path cost formula is shown in equation (4): (4) in It is the Euclidean distance between areas where electromagnetic cables are concentrated. It is an electromagnetic weighting factor. It is a concentrated area of ​​electromagnetic cables Electromagnetic intensity value, coefficient Used to enhance priority in high-intensity areas; the coefficient is adjustable.

[0042] Finally, path optimization is performed, transforming the problem into finding the minimum solution. Arrangement Initialize random path Iterate the disturbance path, swap the two electromagnetic cable concentration areas, and calculate... As shown in equation (5): (5)

[0043] like Then accept the new path; if The old path is retained; the final output is the optimal path sequence. Then, visit all areas where electromagnetic cables are concentrated in sequence.

[0044] Step (3) includes the following steps:

[0045] (3.1) First, establish a confidence model for the existence of electromagnetic cables, with visual features represented as follows: , In color space, motion is represented as , where is the acceleration and angular velocity, as shown in equation (6): (6)

[0046] The existence characteristic vector of the electromagnetic cable is shown in equation (7): (7) in , , These are dynamic weighting coefficients, adjusted in real time using an environment-adaptive algorithm. Indicates electromagnetic intensity;

[0047] (3.2) Next, define the spatiotemporal correlation function within the time window. The confidence level of the existence of electromagnetic cables within. Defined as: (8) in In order to be in The feature vectors of the electromagnetic cable are collected and extracted at all times. It serves as a historical electromagnetic cable feature template, which is updated in real time through online incremental learning.

[0048] (3.3) Next is the adjustment of the dynamic threshold, as shown in equation (9): (9) in For dynamic thresholds, This is the baseline threshold when environmental interference is zero. Environmental noise attenuation factor; The visual contribution coefficient is adjustable. Variance of visual features;

[0049] (3.4) The triggering conditions for hibernation must be met simultaneously for the system to enter a low-power hibernation state, retaining only the basic wake-up circuit. The first condition is confidence level. Limited to a dynamic threshold, followed by duration. Greater than the adaptive idle time; finally, the motion state satisfies... Less than the motion threshold, that is: (10) in To enable adaptive idle time, which is the minimum idle time required before the system enters hibernation, a default value is set and adjusted adaptively; if the system is frequently woken up briefly, then... Automatically extend the sleep / wake cycle to avoid excessively frequent sleep / wake cycles. The motion threshold is used to determine at the physical level whether an autonomous underwater robot is being moved or used.

[0050] (3.5) When a condition requiring a response occurs, the sensor and processing module are woken up in stages. The multi-level wake-up strategy is divided into two levels:

[0051] First, the Doppler velocimeter measures the speed of the autonomous underwater robot relative to the bottom of the water by transmitting / receiving the Doppler frequency shift of sound waves. When the Doppler velocimeter detects a sudden change in flow velocity Δv that reaches a set threshold, it is initially awakened. At this time, only the low-power Doppler velocimeter is used to monitor the flow velocity, which is suitable for long-term standby monitoring. When the initial wake-up is triggered, the device switches from deep sleep to shallow sleep.

[0052] If an electromagnetic anomaly is detected, the autonomous underwater robot will be deep-wake-up. When an electromagnetic anomaly is detected, i.e., the magnitude of the electromagnetic field... If the value exceeds the set threshold, the autonomous underwater robot will be deeply awakened.

[0053] Step (4) includes the following steps:

[0054] (4.1) First, the damaged area is detected. A multimodal damage feature extraction strategy is adopted to process the image data collected by the camera and the electromagnetic anomalies detected by the electromagnetic sensor respectively, and then they are fused to determine the probability of damage.

[0055] Visual cameras capture images of electromagnetic cables Damage is detected by local texture anomaly index, and the anomaly index is used to detect damage. This is represented as shown in equation (11): (11) in It is the local image standard deviation. This represents the total number of pixels within a local area. A damage alarm is triggered when the set threshold is exceeded; These represent the row and column coordinates when traversing each pixel in the electromagnetic cable image, respectively. That is, the first The row and column coordinates of each pixel;

[0056] Secondly, electromagnetic anomalies are detected. Damage to the electromagnetic cable causes electromagnetic field distortion, and the electromagnetic distortion factor is defined as follows: As shown in equation (12): (12) in This is the electromagnetic field reference value for a healthy electromagnetic cable. These are actual electromagnetic measurement values. Represents the testing area. A value greater than the set threshold indicates significant damage;

[0057] Next, the damage is repeatedly confirmed and fused, and the visual and electromagnetic detection results are fused. The basic probability allocation function is shown in Equation (13): (13) in This is a normalized metric based on the visual detection channel. This is a normalized metric based on the electromagnetic detection channel. The threshold value for the visual detection channel. The threshold for electromagnetic detection;

[0058] The fusion rule is shown in equation (14), when the fusion metric value Greater than the set threshold Upon confirmation of damage, the fusion metric value is calculated as shown in equation (14): (14)

[0059] (4.2) Next, the damaged area is quantized, the three-dimensional damaged area is calculated, and stereoscopic vision measurement is performed on the damaged area. The depth information of the damaged area is obtained through a camera, and the pixel coordinates are set as follows: actual coordinates Calculated using the following transformation formula (15): (15) in and It is the camera extrinsic parameter matrix. For rotation matrix, It is a translation vector. The distance data is provided by a Doppler velocimeter;

[0060] Then, the damaged area is quantified, and the damaged region is defined as a set of points. The area of ​​damage is calculated using the convex hull algorithm. The calculation is shown in equation (16): (16) in This indicates the coordinates of the damaged boundary point in the local coordinate system of the electromagnetic cable;

[0061] (4.3) Uncertainty assessment: An area measurement confidence level is introduced, and a threshold is set for the area confidence level. When the confidence level is lower than the threshold, it indicates that the reliability of the damage quantification is not high, and the damaged area needs to be remeasured. The damaged area is calculated as shown in Equation (17): (17) in It is the standard deviation of depth measurement. It is the average distance. It is pixel coordinate error. This is the average area of ​​the pixel region. If the confidence level is below a set threshold, a remeasurement is required.

[0062] (4.4) Calculate the damage level of the damaged area based on the calculated damaged area and electromagnetic distortion factor, and simultaneously mark the latitude and longitude of the damaged location to facilitate repair personnel to repair in order of damage severity, as shown in formula (18): (18) in, Indicates the level of damage;

[0063] (4.5) After completing the current damage quantification, automatically switch to the next section of electromagnetic cable for detection until the entire electromagnetic cable is covered.

[0064] Step (5) includes the following steps:

[0065] (5.1) Design a low battery trigger mechanism. First, determine whether the autonomous underwater vehicle's battery level is below 20%. The current battery level needs to be checked. Real-time calculation is performed, specifically the difference between the battery consumption at the previous moment and the battery consumption during the current time period. The navigation system will be activated at that time. As shown in equation (19): (19) in For motor power, Battery charging and discharging efficiency (discharging efficiency is used in the discharging scenario). This refers to the battery's nominal capacity.

[0066] (5.2) To facilitate the detection of the charging platform, an enhanced detection module for the charging platform was designed. Six sets of low-frequency electromagnetic pulse transmitters were deployed around the charging platform. The frequency setting was required to be greater than the electromagnetic intensity of the electromagnetic cable being detected. The transmitters were evenly distributed around the platform, and each transmitter had a built-in adjustable frequency pulse generation circuit to generate a square wave signal and a power amplifier was placed there.

[0067] After deploying the pulse transmitter, optical-assisted positioning needs to be further enhanced. A ring-shaped LED array is installed on top of the charging platform, using blue light with a longer wavelength. Time-division multiplexing modulation technology is employed to set the reference light pulse period and azimuth-encoded light.

[0068] Next, an adaptive noise suppression module is integrated, which adjusts the transmission parameters in real time using the LMS algorithm, while updating the parameters in real time and continuously tracking the changes in the input signal, as shown in equation (20): (20)

[0069] in For the emission parameter vector, Step size factor For positioning error, The environmental noise covariance matrix is... Forgetting factor, Indicates time;

[0070] (5.3) Next, a positioning and navigation system for the autonomous underwater robot is designed, a multi-source information fusion positioning algorithm is designed, an extended Kalman filter framework is constructed, and data from four types of sensors are fused to predict and update the sensor data, as shown in Equation (21): (twenty one) in Indicates based on all The optimal estimate of the state vector obtained from the observation data at and before time step. This represents the nonlinear state transition function of the system. Indicates in Optimal estimation of the state vector at time 1. express The control input vector of the time system, express The Kalman gain matrix at time t. Indicates in The actual observation vector obtained at any time is the result of the fusion of data from multiple sensors. Represents a nonlinear observation function. Indicates in The posterior covariance matrix of the time-state estimation error. Represents the identity matrix. Represents the observation function The Jacobian matrix calculated at the latest state estimate Indicates in The prior state estimation error covariance matrix at time t represents the uncertainty of the prediction, and the state vector... Includes location attitude and speed Observation vector Fusion of data from four types of sensors: electromagnetic field intensity gradient Optical feature matching error DVL Doppler frequency shift and IMU attitude angular velocity ;

[0071] Guided by the extended Kalman filter framework, the autonomous underwater vehicle moves toward the approximate location of the charging platform. When the distance between the autonomous underwater vehicle and the charging platform enters a medium-to-close range (e.g., 5-10 meters), the system switches from global navigation mode to precise positioning and docking mode.

[0072] The first consideration is electromagnetic guidance and coarse alignment. At this stage, the six sets of low-frequency electromagnetic pulse transmitters deployed around the charging platform change their function from long-range beacons to short-range guide beacons, and the electromagnetic sensor array on the autonomous underwater vehicle (AUV) begins operation. The first step is orientation calculation. By comparing the phase difference and intensity gradient of the electromagnetic pulse signals received by sensors at different locations, the AUV can calculate its lateral offset and orientation deviation relative to the charging platform entrance—that is, the Y-axis deviation and yaw angle deviation—and then perform heading correction. The EKF fuses the electromagnetic orientation information with the velocity vector provided by the Doppler velocimeter and the attitude angles provided by the inertial sensors. The generated control commands are no longer simply forward-moving but focus on eliminating lateral and yaw angle deviations, enabling the AUV to accurately align with the centerline of the charging platform entrance.

[0073] Secondly, at closer distances, optical vision servoing and fine correction are performed. Once the autonomous underwater vehicle (AUV) enters the effective line-of-sight of the optical beacon via electromagnetic guidance, the positioning control is transferred to the optical-assisted positioning system. The high-definition camera on the AUV's head begins to capture the coded light signals emitted by the ring LED array on top of the charging platform. Through image processing algorithms, the system first identifies the LED light spots, then analyzes their coded information using time-division multiplexing modulation technology, and then performs pose calculation. Based on the imaging position, shape distortion, and coded information of the LED light spots in the camera, the vision algorithm can calculate the relative pose of the AUV and the charging platform entrance in six degrees of freedom in real time. The observed values ​​of this pose information are input into the EKF (Optical Keyframe). Finally, the control system issues fine-tuning commands based on the calculated fine pose, driving the AUV's thrusters to perform final heading and attitude corrections to ensure alignment and a smooth approach to the charging interface.

[0074] Finally, guided by the optical system, the autonomous underwater vehicle slowly drives into the guide structure of the charging platform at a low speed. The docking mechanism of the charging platform is usually designed with a funnel-shaped guide cover. When the charging contact of the autonomous underwater vehicle touches the guide cover, the mechanical structure will passively correct the error of the last few centimeters and guide the contact to slide into the correct position. When the charging contact establishes a stable physical and electrical connection with the socket on the platform, a clear connection success signal will be generated. After receiving the connection success signal, the autonomous underwater vehicle immediately shuts down the main thruster, the positioning and navigation task is completed, and the system then starts the charging process.

[0075] (5.4) After the autonomous underwater vehicle is fully charged, it needs to autonomously return to the position where the last interrupted detection was to continue detection until all electromagnetic cables in the detected sea area have been detected. The autonomous underwater vehicle then autonomously returns to the entry position and waits for the staff to retrieve it. The autonomous return realizes the path memory algorithm to construct a topology map, as shown in Equation (22): (twenty two) in For waypoint assembly, The edges connecting waypoints To account for the cost, energy consumption and time are combined in the calculation;

[0076] The route is returned along the same path based on the generated topology map. However, the seabed environment is constantly changing, so the environment is monitored in real time. The path is adjusted in real time according to the detected changes, as shown in equation (23). (twenty three) in Indicates in The AUV system state vector at time t. Indicates in The AUV system state vector at time t. This represents the rotation matrix, which transforms the AUV from its own coordinate system to the global coordinate system. Indicates the time increment. R represents the angular displacement measured by the IMU, and R is the rotation matrix.

[0077] (5.5) After returning along the original route, continue to inspect the remaining electromagnetic cables until all electromagnetic cables in the surveyed sea area have been inspected. Then, the autonomous underwater vehicle returns to the starting point and waits for the staff to retrieve it.

[0078] The present invention provides a lightweight tracking and detection system for submarine electromagnetic cables. The system is used to implement the above-mentioned method and includes an electromagnetic sensor, a visual camera, a Doppler velocimeter, and an inertial sensor; it is equipped with a lithium battery, a buoyancy drive system, and a propulsion unit.

[0079] Simulation Experiment

[0080] The autonomous underwater vehicle integrates an NVIDIA Jetson Orin Nano development board as the main control unit. Sensors include a ZED2i binocular camera (30Hz, acquiring 640×360 resolution images, with a built-in IMU outputting 200Hz six-axis data), an electromagnetic sensor, a Doppler velocimeter, and an inertial measurement unit (IMU). The positioning system is a GNSS-RTK integrated navigation system for surface positioning reference. The Jetson Orin Nano runs Ubuntu and ROS, receiving real-time data from the camera, IMU, and RTK via serial port, relying on ROS system timestamps to ensure time synchronization of all sensors. The charging platform is equipped with six low-frequency electromagnetic pulse transmitters and a ring LED array.

[0081] A simulated underwater environment was created in a pool, where five miniature electromagnetic cables were deployed. Low-voltage current was passed through them to generate an electromagnetic field, and six artificial damage points were created on the cables. An autonomous underwater vehicle then began its inspection from the starting point.

[0082] The method of this invention prioritizes access to strong signal regions, reducing redundant detection. Experimental results are shown in Table 1. Compared with traditional methods, the task time is reduced by 15.6%. The detection path of the autonomous underwater robot is as follows: Figure 3 As shown, the autonomous underwater vehicle successfully tracked the five deployed cables in a shorter detection time.

[0083] Table 1 Comparison of detection times for autonomous underwater robots method Task time (s) Traditional traversal scanning method 352 Patented method 297

[0084] Secondly, the method of this invention integrates the visual texture anomaly index and the electromagnetic distortion factor, improving the accuracy of damage detection by 4% and reducing the quantification error of the damaged area to 5.3%. Furthermore, the multi-source fusion positioning method of this invention achieves a 100% docking success rate during experiments. Figure 4 The trajectory map of the autonomous underwater robot docking with the charging platform was displayed. Figure 5 The exhibition showcases a 3D trajectory map when connected to a charging platform, demonstrating the ability of an autonomous underwater robot using a patented method to smoothly switch from global navigation to precise positioning.

[0085] Experiments have verified the superiority of the method of this invention in the inspection of submarine electromagnetic cables. It effectively reduces the detection time of electromagnetic cables, improving detection efficiency; simultaneously, it enhances the accuracy of damage detection and reduces the quantification error at the point of damage; and it can autonomously dock with a charging platform, ensuring long-term operation. Compared with traditional methods, this invention can achieve a more efficient, accurate, and energy-saving complete inspection process.

Claims

1. A lightweight tracking and detection method for submarine electromagnetic cables, characterized in that, Includes the following steps: (1) Build a hardware system. The autonomous underwater robot is equipped with an electromagnetic sensor, a vision camera, a Doppler velocimeter and an inertial sensor; it is equipped with a lithium battery, a buoyancy drive system and a thruster. (2) The detection of submarine electromagnetic cables begins. The autonomous underwater robot is placed in the sea area to be detected. After it sinks to the seabed, the autonomous underwater robot officially starts the detection work. First, the autonomous underwater robot circles the sea area to be detected and simultaneously turns on the electromagnetic sensor to conduct electromagnetic induction on the sea area, establishes a grid map, divides the sea area into grids, and each grid has an electromagnetic intensity value. According to the electromagnetic intensity classification, high intensity grids correspond to the location of electromagnetic cables, and low intensity areas close to zero are eliminated. Then, optimal path planning is performed to calculate the optimal path for submarine cable detection, quickly locate the area where electromagnetic cables are concentrated, and reduce redundant searches; (3) A confidence model for the existence of electromagnetic cables is established by integrating data from multiple sensors. Based on the dynamic threshold, the autonomous underwater robot can automatically enter a low-power sleep state when there is no electromagnetic cable target. The confidence level must be stable, the duration must meet the standard, and the body must be stationary. When specific signals such as sudden changes in water flow or electromagnetic anomalies are detected, a graded strategy is adopted to gradually wake up the sensors and processing modules, thereby improving the endurance and work efficiency of the autonomous underwater robot during long-term underwater operations. (4) After successfully tracing the electromagnetic cable, the automated detection and quantitative assessment of electromagnetic cable damage is initiated. First, the damage point is initially identified and confirmed by fusing the local texture anomaly index of the visual camera image and the distortion factor of the electromagnetic sensor. Then, the three-dimensional area of ​​the damaged area is quantified using stereo vision and convex hull algorithm. Next, the confidence level is introduced to assess the credibility of the measurement results. If the credibility is insufficient, a remeasurement is triggered. Finally, the damage level is calculated by combining the damaged area and the electromagnetic distortion factor to provide a priority basis for maintenance. Finally, the location information of the confirmed damage location is synchronized, and after the current point assessment is completed, it automatically moves along the electromagnetic cable for continuous detection. (5) During the detection process, if the autonomous underwater vehicle's battery level is below 20%, the autonomous underwater vehicle will automatically navigate to the charging platform for contact charging. After charging to 100%, the autonomous underwater vehicle will automatically return to the place where the last detection was interrupted and continue to detect the electromagnetic cable. This process continues until all the electromagnetic cables in the detected sea area have been scanned and detected. The autonomous underwater vehicle will then return to the location where it first entered the water and wait for retrieval.

2. The lightweight tracking and detection method for submarine electromagnetic cables according to claim 1, characterized in that, Step (2) specifically includes the following sub-steps: (2.1) First, the autonomous underwater vehicle (AUV) circled the boundary of the sea area, collected electromagnetic intensity data, and constructed a grid map. The AUV integrated visual sensors, Doppler velocimeters, and inertial sensors to track and detect submarine electromagnetic cables. (2.2) Simultaneously, the electromagnetic intensity of the surveyed sea area is detected, and the electromagnetic signals are scanned. The electromagnetic detector continuously samples at a fixed frequency, and the measurement points are recorded as follows: Its magnetic flux density modulus is denoted as And record its position coordinates. The autonomous underwater vehicle's position is determined by fusing Doppler velocimetry and inertial sensors. At this point, it is necessary to calculate the electromagnetic intensity value of each individual grid cell. The electromagnetic intensity value is calculated based on the spatial weight of the measurement point and the attenuation effect, as shown in equation (1): （1） in This represents the total number of measurement points. For measurement points Distance to the center of the grid The attenuation coefficient is set based on empirical values, and then an electromagnetic grid map is output. Each element Represents grid Standardized electromagnetic intensity; (2.3) Next, the divided grids are classified to identify the strong electromagnetic grids corresponding to the electromagnetic cables, and irrelevant areas with electromagnetic intensity close to zero are removed. For this purpose, an adaptive threshold classification model is established, and the statistical characteristics of the global electromagnetic intensity and the average intensity are set. and strength standard deviation As shown in equation (2): (2) Where M is the number of rows in the raster map and N is the number of columns in the raster map; A threshold formula is set to distinguish between high-intensity and low-intensity grids. The threshold is dynamically generated and adapts to the electromagnetic characteristics of the sea area to avoid misjudgment caused by a fixed threshold. The high-intensity and low-intensity thresholds are set as shown in equation (3): (3) in For high intensity threshold, The grid markings indicate areas of concentrated electromagnetic cables. These concentrated areas may correspond to electromagnetic cables with coefficients... Ensures coverage of significantly strong signals, with adjustable coefficients; Low intensity threshold, The raster is removed, and the coefficient is... Filters noise, and the coefficient is adjustable; (2.4) Next, form a set of the raster regions that meet the conditions, and extract all those that satisfy the conditions. The center point of the grid forms a concentrated area of ​​electromagnetic cables. ,in It is the first The coordinates of the concentrated area of ​​electromagnetic cables are identified, and those that meet the criteria are excluded. The grid does not participate in path planning; Output electromagnetic cable concentrated area collection ,size Represents the potential number of electromagnetic cables; (2.5) Finally, shortest path planning is performed on the retained grid areas to calculate the shortest path for the autonomous underwater vehicle to access all areas with concentrated electromagnetic cables, minimizing the travel time. First, the path cost function is designed. Defined as the sequence of arrangements of concentrated areas of electromagnetic cables: ; Path cost Combining path length and electromagnetic intensity weights, strong signal areas are traversed first. The cost function here is not simply geographical distance, but rather guides the path to cover dense electromagnetic cable areas through electromagnetic weights, thereby improving efficiency. The path cost formula is shown in equation (4): (4) in It is the Euclidean distance between areas where electromagnetic cables are concentrated. It is an electromagnetic weighting factor. It is a concentrated area of ​​electromagnetic cables Electromagnetic intensity value, coefficient Used to enhance priority in high-intensity areas; the coefficient is adjustable. Finally, path optimization is performed, transforming the problem into finding the minimum solution. Arrangement Initialize random path Iterate the disturbance path, swap the two electromagnetic cable concentration areas, and calculate... As shown in equation (5): (5) like Then accept the new path; if The old path is retained; the final output is the optimal path sequence. Then, visit all areas where electromagnetic cables are concentrated in sequence.

3. The lightweight tracking and detection method for submarine electromagnetic cables according to claim 2, characterized in that, Step (3) includes the following steps: (3.1) First, establish a confidence model for the existence of electromagnetic cables, with visual features represented as follows: , In color space, motion is represented as , where is the acceleration and angular velocity, as shown in equation (6): (6) Then the eigenvector of the electromagnetic cable existence As shown in equation (7): (7) in , , These are dynamic weighting coefficients, adjusted in real time using an environment-adaptive algorithm. Indicates electromagnetic intensity; (3.2) Next, define the spatiotemporal correlation function within the time window. The confidence level of the existence of electromagnetic cables within. Defined as: (8) in In order to be in The feature vectors of the electromagnetic cable are collected and extracted at all times. It serves as a historical electromagnetic cable feature template, which is updated in real time through online incremental learning. (3.3) Next is the adjustment of the dynamic threshold, as shown in equation (9): (9) in For dynamic thresholds, This is the baseline threshold when environmental interference is zero. Environmental noise attenuation factor; The visual contribution coefficient is adjustable. Variance of visual features; (3.4) The triggering conditions for hibernation must be met simultaneously for the system to enter a low-power hibernation state, retaining only the basic wake-up circuit. The first condition is confidence level. Limited to a dynamic threshold, followed by duration. Greater than the adaptive idle time; finally, the motion state satisfies... Less than the motion threshold, that is: (10) in To enable adaptive idle time, which is the minimum idle time required before the system enters hibernation, a default value is set and adjusted adaptively; if the system is frequently woken up briefly, then... Automatically extend the sleep / wake cycle to avoid excessively frequent sleep / wake cycles. The motion threshold is used to determine at the physical level whether an autonomous underwater robot is being moved or used. (3.5) When a condition requiring a response occurs, the sensor and processing module are woken up in stages. The multi-level wake-up strategy is divided into two levels: First, the Doppler velocimeter measures the speed of the autonomous underwater robot relative to the bottom of the water by transmitting / receiving the Doppler frequency shift of sound waves. When the Doppler velocimeter detects a sudden change in flow velocity Δv that reaches a set threshold, it is initially awakened. At this time, only the low-power Doppler velocimeter is used to monitor the flow velocity, which is suitable for long-term standby monitoring. When the initial wake-up is triggered, the device switches from deep sleep to shallow sleep. If an electromagnetic anomaly is detected, the autonomous underwater robot will be deep-wake-up. When an electromagnetic anomaly is detected, i.e., the magnitude of the electromagnetic field... If the value exceeds the set threshold, the autonomous underwater robot will be deeply awakened.

4. The lightweight tracking and detection system and method for submarine electromagnetic cables according to claim 3, characterized in that, Step (4) includes the following steps: (4.1) First, the damaged area is detected. A multimodal damage feature extraction strategy is adopted to process the image data collected by the visual camera and the electromagnetic anomaly detected by the electromagnetic sensor respectively, and then they are fused to determine the probability of damage. Visual cameras capture images of electromagnetic cables Damage is detected by local texture anomaly index, and the anomaly index is used to detect damage. This is represented as shown in equation (11): (11) in It is the local image standard deviation. This represents the total number of pixels within a local area. A damage alarm is triggered when the set threshold is exceeded; m and n represent the row and column coordinates when traversing each pixel in the electromagnetic cable image, respectively. , That is, the row and column coordinates of the i-th pixel; Secondly, electromagnetic anomalies are detected. Damage to the electromagnetic cable causes electromagnetic field distortion, and the electromagnetic distortion factor is defined as follows: As shown in equation (12): (12) in This is the electromagnetic field reference value for a healthy electromagnetic cable. These are actual electromagnetic measurement values. Represents the testing area. A value greater than the set threshold indicates significant damage; Next, the damage is repeatedly confirmed and fused, and the visual and electromagnetic detection results are fused. The basic probability allocation function is shown in Equation (13): (13) in This is a normalized metric based on the visual detection channel. This is a normalized metric based on the electromagnetic detection channel. The threshold value for the visual detection channel. The threshold for electromagnetic detection; The fusion rule is shown in equation (14), when the fusion metric value When the value exceeds the set threshold, the damage is confirmed, and the fusion metric is calculated as shown in equation (14): (14) (4.2) Next, the damaged area is quantized, the three-dimensional damaged area is calculated, and stereoscopic vision measurement is performed on the damaged area. The depth information of the damaged area is obtained through a camera, and the pixel coordinates are set as follows: Then the actual coordinates Calculated using the following transformation formula (15): (15) in and It is the camera extrinsic parameter matrix. For rotation matrix, It is a translation vector. The distance data is provided by a Doppler velocimeter; Then, the damaged area is quantified, and the damaged region is defined as a set of points. Damaged area The calculation is shown in equation (16): (16) in This indicates the coordinates of the damaged boundary point in the local coordinate system of the electromagnetic cable; (4.3) Uncertainty assessment: An area measurement confidence level is introduced, and a threshold is set for the area confidence level. When the confidence level is lower than the threshold, it indicates that the reliability of the damage quantification is not high, and the damaged area needs to be remeasured. The damaged area is calculated as shown in Equation (17): (17) in It is the standard deviation of depth measurement. It is the average distance. It is pixel coordinate error. It is the average area of ​​the pixel region. If the confidence level is lower than the set threshold, it needs to be measured again. (4.4) Calculate the damage level of the damaged area based on the calculated damaged area and electromagnetic distortion factor, and simultaneously mark the latitude and longitude of the damaged location to facilitate repair personnel to repair in order of damage severity, as shown in formula (18): (18) in, Indicates the level of damage; (4.5) After completing the current damage quantification, automatically switch to the next section of electromagnetic cable for detection until the entire electromagnetic cable is covered.

5. A lightweight tracking and detection system and method for submarine electromagnetic cables according to claim 4, characterized in that, Step (5) includes the following steps: (5.1) Design a low battery trigger mechanism. First, determine whether the autonomous underwater vehicle's battery level is below 20%. The current battery level needs to be checked. Real-time calculation, that is, the difference between the power consumption at the previous moment and the power consumption in the current time period. The navigation system will be activated at that time. As shown in equation (19): (19) in Motor power, Battery charging and discharging efficiency (discharging efficiency is used in the discharging scenario). This refers to the battery's nominal capacity. (5.2) Deploy 6 sets of low-frequency electromagnetic pulse transmitters around the charging platform. The frequency setting must be greater than the electromagnetic intensity of the electromagnetic cable being tested. The transmitters are evenly distributed around the platform, and each transmitter has a built-in adjustable frequency pulse generation circuit to generate a square wave signal, and a power amplifier is placed in it. After deploying the pulse transmitter, it is necessary to further enhance optical-assisted positioning by installing a ring-shaped LED array on the top of the charging platform, selecting blue light with a longer wavelength, using time-division multiplexing modulation technology, setting the reference light pulse period, and azimuth encoding light. Next, an adaptive noise suppression module is integrated, which adjusts the transmission parameters in real time using the LMS algorithm, while updating the parameters in real time and continuously tracking the changes in the input signal, as shown in equation (20): (20) in For the emission parameter vector, Step size factor For positioning error, The environmental noise covariance matrix is... Forgetting factor, Indicates time; (5.3) Next, a positioning and navigation system for the autonomous underwater robot is designed, a multi-source information fusion positioning algorithm is designed, an extended Kalman filter framework is constructed, and data from four types of sensors are fused to predict and update the sensor data, as shown in Equation (21): (21) in, Indicates based on all The optimal estimate of the state vector obtained from the observation data at and before time step. This represents the nonlinear state transition function of the system. Indicates in Optimal estimation of the state vector at time 1. express The control input vector of the time system, express The Kalman gain matrix at time 10:

00. Indicates in The actual observation vector obtained by fusing data from multiple sensors at any given moment. Represents a nonlinear observation function. Indicates in The posterior covariance matrix of the time-state estimation error. Represents the identity matrix. Represents the observation function The Jacobian matrix calculated at the latest state estimate Indicates in The prior state estimation error covariance matrix at time t represents the uncertainty of the prediction, and the state vector... Includes location attitude and speed Observation vector Fusion of data from four types of sensors: electromagnetic field intensity gradient Optical feature matching error DVL Doppler frequency shift and IMU attitude angular velocity ; Guided by the extended Kalman filter framework, the autonomous underwater vehicle (AUV) moves towards the approximate location of the charging platform. When the distance between the AUV and the charging platform approaches mid-to-close range, the system switches from global navigation mode to precise positioning and docking mode. First, coarse alignment is performed using electromagnetic guidance. The low-frequency electromagnetic pulses from the charging platform and the onboard electromagnetic sensor array are used to calculate the orientation and fuse multi-source data to eliminate lateral and yaw deviations, aligning it with the entrance centerline. Then, at close range, the system switches to optical vision servoing, using a camera to capture the LED-encoded light signals from the platform to calculate the fine pose and make fine adjustments. Finally, the AUV slowly enters under optical guidance, relying on a mechanical guide shield to correct the final error and achieve a stable connection of the charging contacts. The thrusters are then turned off and charging is initiated. (5.4) After the autonomous underwater vehicle is fully charged, it needs to autonomously return to the position where the last interrupted detection was to continue detection until all electromagnetic cables in the detected sea area have been detected. The autonomous underwater vehicle then autonomously returns to the entry position and waits for the staff to retrieve it. The autonomous return realizes the path memory algorithm to construct a topology map, as shown in Equation (22): (22) in For waypoint assembly, The edges connecting waypoints To account for the cost, energy consumption and time are combined in the calculation; The route is returned along the same path based on the generated topology map. However, the seabed environment is constantly changing, so the environment is monitored in real time. The path is adjusted in real time according to the detected changes, as shown in equation (23). (23) in Indicates in The state vector of the AUV system at time t. Indicates in The state vector of the AUV system at time t. This represents the rotation matrix, which transforms the AUV from its own coordinate system to the global coordinate system. Indicates the time increment. R represents the angular displacement measured by the IMU, and R is the rotation matrix. (5.5) After returning along the original route, continue to inspect the remaining electromagnetic cables until all electromagnetic cables in the surveyed sea area have been inspected. Then, the autonomous underwater vehicle returns to the starting point and waits for the staff to retrieve it.

6. A lightweight tracking and detection system for submarine electromagnetic cables, characterized in that, The system is used to implement the method described in any one of claims 1-5. The system includes an electromagnetic sensor, a vision camera, a Doppler velocimeter, and an inertial sensor; and is equipped with a lithium battery, a buoyancy drive system, and a propulsion unit.

Images