Data acquisition method and system for visual management of underwater voltage-withstanding battery pack

Abstract

The invention provides a data acquisition method and system for visual management of an underwater pressure-resistant battery pack, and relates to the technical field of battery management, and the method comprises the steps: 1, enabling a remote control submersible to dive to the position of a pressure-resistant cabin of the battery pack, and positioning a wet-plugging electric connector; acquiring a real-time image of the wet-plugging electric connector through a visual identification part of the remote control submersible, and identifying the blocking state of the connector interface according to the real-time image; 2, a cleaning device of the remote control submersible is controlled to conduct high-pressure water jet cleaning on the connector interface, and the cleaned connector interface is obtained; and acquiring a cleaned image through a visual identification part, and extracting a plurality of space coordinate points on the surface of the connector interface from the cleaned image. Reliable connection of the connector is achieved through positioning cleaning, deformation evaluation and compensation butt joint, it is guaranteed that historical operation data of the battery pack are completely collected and transmitted to a shore base, and accurate data support is provided for visual diagnosis and evaluation of the health state of the underwater voltage-withstanding battery pack.

Description

Technical Field

[0001] This invention relates to the field of battery management technology, and in particular to a data acquisition method and system for the visual management of underwater pressure-resistant battery packs. Background Technology

[0002] Currently, underwater high-pressure battery packs typically communicate with external devices via wet-plug electrical connectors. These connectors are generally equipped with protective covers or guiding structures to reduce the impact of the underwater environment on the interface. However, under the long-term effects of high pressure and corrosive underwater environments, the connector interface may undergo a certain degree of structural deformation, which may affect the stability of subsequent docking operations. Existing underwater connector maintenance methods mostly employ remotely operated vehicles (ROVs) for visual inspection and cleaning.

[0003] For example, during the maintenance of the battery pack of a certain type of underwater observation equipment, after the operator identifies and cleans the connector interface using a remotely operated vehicle, it may be difficult to effectively identify the micro-geometric changes of the interface due to the reliance on visual judgment. In this case, the reliability of the physical connection between the docking device and the connector may be affected by insufficient fitting accuracy, thus posing a certain challenge to the stability of data acquisition. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a data acquisition method and system for the visual management of underwater pressure-resistant battery packs, thereby realizing the visualization and intelligent management of underwater battery packs.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows: Firstly, a data acquisition method for visual management of underwater pressure-resistant battery packs, the method comprising: Step 1: The remotely operated vehicle (ROV) descends to the pressure chamber of the battery pack and locates the wet-plug electrical connector; real-time images of the wet-plug electrical connector are obtained through the visual recognition component of the ROV, and the blockage status of the connector interface is identified based on the real-time images; Step 2: Control the cleaning device of the remotely operated vehicle to clean the connector interface with high-pressure water jet to obtain the cleaned connector interface; acquire the cleaned image through the visual recognition component, and extract multiple spatial coordinate points on the surface of the connector interface from the cleaned image; Step 3: Based on multiple spatial coordinate points, determine the central conductive contact area, the left edge sealing area, and the right edge sealing area on the connector interface, and determine the planar spatial equation through the three areas; Step 4: Based on the plane space equation, calculate the average distance between multiple spatial coordinate points and the plane space equation as a surface deviation metric, and calculate the rate of change of the normal vector of multiple spatial coordinate points; Step 5: Combine surface deviation measurement and normal vector change rate to calculate the overall deformation index of the connector interface, and generate deformation compensation coefficient based on the overall deformation index. Step 6: Using the deformation compensation coefficient and docking guide structure, control the docking device to establish a physical connection with the wet plug-in electrical connector to read the complete historical operating data of the battery pack from the battery management unit and transmit the historical operating data to the shore-based control center for visual diagnosis and evaluation of battery health status.

[0006] Secondly, the data acquisition system for the visual management of underwater pressure-resistant battery packs includes: The positioning and visual recognition module is used for the remotely operated vehicle to descend to the battery pack pressure chamber and locate the wet-plug electrical connector; it acquires real-time images of the wet-plug electrical connector through the visual recognition component of the remotely operated vehicle and identifies the blockage status of the connector interface based on the real-time images; The adaptive cleaning module controls the cleaning device of the remotely operated vehicle to perform high-pressure water jet cleaning on the connector interface to obtain a cleaned connector interface; it acquires a cleaned image through a vision recognition component and extracts multiple spatial coordinate points on the surface of the connector interface from the cleaned image. The geometric partitioning and plane fitting module is used to determine the central conductive contact area, left edge sealing area and right edge sealing area on the connector interface based on multiple spatial coordinate points, and to determine the plane space equation through the three areas; The deformation analysis module is used to calculate the average distance between multiple spatial coordinate points and the plane spatial equation as a surface deviation metric, and to calculate the rate of change of the normal vector of multiple spatial coordinate points, based on the plane spatial equation. The deformation compensation coefficient generation module is used to calculate the overall deformation index of the connector interface by combining the surface deviation metric and the normal vector change rate, and to generate the deformation compensation coefficient based on the overall deformation index. The adaptive docking and data feedback module is used to control the docking device to establish a physical connection with the wet plug-in electrical connector by utilizing the deformation compensation coefficient and docking guide structure. This allows the module to read complete historical operating data of the battery pack from the battery management unit and transmit the historical operating data to the shore-based control center for visual diagnosis and evaluation of battery health status.

[0007] Thirdly, a computer-readable storage medium storing a program that, when executed by a processor, implements the method.

[0008] The above-described solution of the present invention has at least the following beneficial effects: By acquiring real-time images through the visual recognition components of a remotely operated vehicle (ROV) and combining this with blockage status identification, such as a comprehensive judgment of the area covered by marine organisms and the thickness of sediment, the blockage status of the connector interface can be determined more accurately, avoiding the omission of potential blockage risks due to relying solely on surface visual observation. After high-pressure water jet cleaning, the spatial coordinates of the connector interface surface are extracted using multi-view image fusion, overcoming the limitations of two-dimensional image analysis and enabling a comprehensive capture of the interface surface's geometric features. By dividing the interface into a central conductive contact area, a left edge sealing area, and a right edge sealing area, and determining a planar spatial equation based on the key coordinates of these three areas as a reference plane, combined with surface deviation measurement and normal vector change rate, a comprehensive assessment of the interface's macroscopic flatness and microscopic deformation is achieved, avoiding the reliance on a single indicator. The limitations of deformation state judgment are addressed by generating a deformation compensation coefficient based on the overall deformation index, dynamically adjusting the pose parameters of the docking device, and ensuring precise matching between the docking guide structure and the connector interface. This effectively offsets the interface deformation caused by underwater high pressure and corrosive environments, improves the stability of the physical connection, and solves the connection reliability problem caused by insufficient mating precision. A stable physical connection activates a high-speed data communication link, enabling complete reading of historical operating data from the battery management unit, such as cell voltage curves, temperature curves, and charge / discharge cycle records. This data is then transmitted to the shore-based control center via the underwater communication module, providing comprehensive and reliable data support for the visual diagnosis of battery health status and the safety risk assessment of performance degradation. This achieves visualization and intelligent management of underwater battery packs. Attached Figure Description

[0009] Figure 1 This is a flowchart illustrating the data acquisition method for visual management of underwater pressure-resistant battery packs provided in an embodiment of the present invention.

[0010] Figure 2 This is a schematic diagram of a data acquisition system for visual management of underwater pressure-resistant battery packs provided in an embodiment of the present invention. Detailed Implementation

[0011] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0012] like Figure 1 As shown, embodiments of the present invention propose a data acquisition method for the visual management of underwater pressure-resistant battery packs, the method comprising the following steps: Step 1: The remotely operated vehicle (ROV) descends to the pressure chamber of the battery pack and locates the wet-plug electrical connector; real-time images of the wet-plug electrical connector are obtained through the visual recognition component of the ROV, and the blockage status of the connector interface is identified based on the real-time images; Step 2: Control the cleaning device of the remotely operated vehicle to clean the connector interface with high-pressure water jet to obtain the cleaned connector interface; acquire the cleaned image through the visual recognition component, and extract multiple spatial coordinate points on the surface of the connector interface from the cleaned image; Step 3: Based on multiple spatial coordinate points, determine the central conductive contact area, the left edge sealing area, and the right edge sealing area on the connector interface, and determine the planar spatial equation through the three areas; Step 4: Based on the plane space equation, calculate the average distance between multiple spatial coordinate points and the plane space equation as a surface deviation metric, and calculate the rate of change of the normal vector of multiple spatial coordinate points; Step 5: Combine surface deviation measurement and normal vector change rate to calculate the overall deformation index of the connector interface, and generate deformation compensation coefficient based on the overall deformation index. Step 6: Using the deformation compensation coefficient and docking guide structure, control the docking device to establish a physical connection with the wet plug-in electrical connector to read the complete historical operating data of the battery pack from the battery management unit and transmit the historical operating data to the shore-based control center for visual diagnosis and evaluation of battery health status.

[0013] In this embodiment of the invention, real-time images are acquired through the visual recognition component of a remotely operated vehicle (ROV). Combined with blockage status identification, such as a comprehensive judgment of the marine organism coverage area and sediment thickness, the blockage status of the connector interface can be determined more accurately, avoiding the omission of potential blockage risks due to reliance solely on surface visual observation. After high-pressure water jet cleaning, spatial coordinate points on the connector interface surface are extracted using multi-view image fusion, overcoming the limitations of two-dimensional image analysis and comprehensively capturing the geometric features of the interface surface. By dividing the interface into a central conductive contact area, a left edge sealing area, and a right edge sealing area, and determining a planar spatial equation based on the key coordinates of these three areas as a reference plane, combined with surface deviation measurement and normal vector change rate, a comprehensive assessment of the macroscopic flatness and microscopic deformation of the interface is achieved, avoiding the omission of potential blockage risks due to reliance solely on surface visual observation. The limitations of using a single indicator to judge deformation status are addressed by generating a deformation compensation coefficient based on the overall deformation index. This dynamically adjusts the pose parameters of the docking device, ensuring precise matching between the docking guide structure and the connector interface. This effectively offsets the impact of interface deformation caused by underwater high pressure and corrosive environments, improving the stability of the physical connection and resolving connection reliability issues due to insufficient mating precision. Furthermore, the stable physical connection activates a high-speed data communication link, enabling the complete reading of historical operating data from the battery management unit, such as cell voltage curves, temperature curves, and charge / discharge cycle records. This data is then transmitted to the shore-based control center via the underwater communication module, providing comprehensive and reliable data support for the visual diagnosis of battery health status and the safety risk assessment of performance degradation. This achieves visualization and intelligent management of underwater battery packs.

[0014] In another preferred embodiment of the present invention, before step 1, a wet-plug electrical connector is provided on the outer shell of the battery pack pressure chamber. The wet-plug electrical connector integrates an openable and closable protective cover and a docking guide structure for sealing the connector interface in a non-docked state.

[0015] In this embodiment of the invention, the integrated installation of the wet-plug electrical connector must be based on the sealing requirements of the underwater pressure-resistant environment, combined with the stability requirements of battery pack data transmission. The process is completed in stages: positioning preprocessing, main assembly, functional structure integration, and performance verification. The specific process is as follows: First, based on the structural design drawings of the battery pack pressure-resistant chamber, the location of the signal output interface of the battery management unit inside the chamber is determined. Using this as a reference, the installation area of ​​the connector is marked on the outer shell of the chamber. This area must avoid key structures such as load-bearing welds and pressure-resistant reinforcing ribs, and a non-load-bearing area with a gentle surface curvature is selected to ensure that the overall pressure resistance performance of the chamber is not affected after installation. After determining the area, a CNC milling machine is used to perform planar milling on this part to form a rectangular mounting slot that perfectly matches the size of the connector base. The groove depth is controlled at 1 / 2 the thickness of the connector base to ensure that the base is flush with the surface of the tank after installation, reducing water flow resistance and debris adhesion. To enhance the sealing effect, annular sealing grooves with a width of 5mm and a depth of 3mm are milled around the installation groove. The inner wall of the groove is polished to remove burrs. Then, a pre-customized seawater corrosion-resistant fluororubber seal is embedded in the groove. The seal needs to be 1mm higher than the groove surface to ensure a compression seal when it is fitted with the connector base. After the groove is processed, two layers of epoxy resin anti-corrosion coating are sprayed on the groove surface and surrounding area. The coating thickness is controlled between 80 and 100μm. After the coating is fully cured, the surface roughness is tested with a roughness meter to ensure that the Ra value is not greater than 1.6μm to avoid sealing failure due to uneven surface.

[0016] The second step involves selecting a connector body that meets the pressure resistance rating required for the corresponding underwater depth. Its outer shell is made of a single-piece forged high-strength titanium alloy, with solution treatment to enhance mechanical strength. The shell surface undergoes anodizing to form a wear-resistant and corrosion-resistant layer. During assembly, a small amount of anaerobic adhesive is first applied to the mounting holes of the connector base. The base is then placed into the mounting slot and adjusted to a horizontal position. A torque wrench is used to tighten the high-strength stainless steel bolts through the mounting holes of the base and into the pre-drilled internal threaded holes in the hull. The bolt tightening torque is performed according to design requirements, ensuring a tight fit without over-tightening and causing deformation of the base. After tightening the bolts, waterproof sealant is applied to the bolt heads and wrapped with PTFE tape to form a secondary seal. Following this, in-hull wiring operations are performed, installing the connectors into the pre-drilled through holes in the corresponding locations within the hull. A high-pressure sealed penetration component is installed, employing a metal-ceramic sealing structure to ensure no leakage under deep water pressure. The signal and power pins of the connector are connected to the internal interface of the penetration component via silver-plated shielded wires. The wires are covered with polyamide protective tubing to prevent damage from friction with other components inside the chamber. The other end of the wire is soldered to the corresponding interface of the battery management unit using tin-lead solder. During the soldering process, a hot air gun is used to control the soldering temperature to prevent high temperature damage to the interface components. After soldering, the solder joint is wrapped with insulating and waterproof heat shrink tubing, covering 10mm on both ends of the solder joint. After heat shrinking, ensure there are no bubbles or wrinkles, effectively isolating moisture and electromagnetic interference. After wiring is completed, the wires inside the chamber are organized and fixed to the chamber support with nylon cable ties to prevent the wires from shifting due to chamber vibration.

[0017] The third step involves advancing the core integrated design of the protective cover and the docking guide structure. This step is crucial for ensuring the protection of the connector in its non-docking state and the accuracy of docking. It requires seamless collaboration between the two structures and the connector body. The protective cover adopts a flip-top design, and the main body is made of aging-resistant and impact-resistant polyetheretherketone (PEEK) engineering plastic, manufactured through injection molding. Its dimensions perfectly match the connector interface. An annular boss is integrally formed on the inner edge, and an oil- and water-resistant silicone rubber sealing strip is attached to the boss. The sealing strip has a wedge-shaped cross-section, forming a mutually supportive structure with the chamfered sealing surface of the connector interface edge to ensure a surface seal when closed. The opening and closing mechanism of the protective cover consists of a torque spring and a miniature electromagnetic lock, which are symmetrically installed on both sides of the connector body. The torque spring is made of stainless steel and, after pre-compression installation, provides stable flipping force, allowing the protective cover to open naturally. The electromagnetic lock is a low-power, waterproof model, with its latch precisely engaging with the latch on the protective cover. In the non-docking state, the electromagnetic lock is powered by the internal power supply, and the latch extends to lock the protective cover, ensuring a tight seal between the sealing strip and the interface, forming the first line of defense. When the remotely operated vehicle arrives at the work area, it sends an unlocking signal through the underwater acoustic communication module. Upon receiving the signal, the internal control unit de-energizes the electromagnetic lock, causing the latch to retract. The torque spring's force then causes the protective cover to automatically flip open to a 90° position around the hinge, fully exposing the interface for docking operations. Simultaneously, the opening of the protective cover does not obstruct the docking guide structure, preventing interference with subsequent operations.

[0018] The docking guide structure and connector body are integrated into the outer periphery of the interface, forming a ring shape. The main structure is a conical guide sleeve with a gradually transitioning slope design on the inner side. The slope gradually transitions from a large diameter at the entrance end to a small diameter at the interface end, and the surface roughness is controlled to a profile arithmetic mean deviation of no more than 0.8 micrometers to reduce frictional resistance during docking. The entrance end of the guide sleeve has a 15° chamfer to facilitate quick insertion and positioning of the docking device of the remotely operated vehicle. Inside the guide sleeve, three cylindrical positioning pins are evenly distributed along the circumference. The positioning pins are made of high-strength alloy steel with a nitrided surface to increase hardness. The positioning pins are fixed to the guide sleeve with an interference fit, and their axes are aligned with the connector body. The central axes of the connector interfaces remain parallel, and the diameter of the positioning pin and the size tolerance of the positioning hole on the remotely operated vehicle docking device are controlled within 0.02mm to ensure a precise fit during docking, effectively compensating for positioning errors during underwater operations and ensuring complete alignment of the central axes of the interfaces. To further extend service life, a low-friction polytetrafluoroethylene coating with a thickness of 5 to 8μm is sprayed on the inner surface of the guide sleeve and the outer surface of the positioning pin. This coating does not affect the fitting accuracy and reduces mechanical wear during docking. At the same time, the material of the guide structure is consistent with that of the connector body, both being titanium alloy, ensuring the same corrosion rate in the seawater environment and avoiding structural loosening or sealing failure due to material differences.

[0019] The fourth step, after completing all structural installations, involves a sealing performance test. The battery pack pressure chamber with connectors installed is placed entirely into a large high-pressure tank. A simulated seawater salt solution is injected into the tank, and the pressure is gradually increased to the maximum designed working pressure using a pressure control system. This pressure is maintained for 30 minutes, during which a high-precision leak detector is used to check key areas such as the connector-to-chamber joint, penetrations, and the sealing surface of the protective cover to ensure there are no signs of leakage. A depressurization process is then conducted, with the test repeated after each pressure reduction to verify the sealing stability during pressure changes. After the sealing test is passed, functional debugging is performed. An unlocking signal is sent using a simulated remotely operated vehicle to test the opening and closing response speed and reliability of the protective cover. The process involved 50 consecutive opening and closing cycles to ensure the electromagnetic lock operated sensitively and the protective cover flipped smoothly without jamming. Next, a docking device simulator for the remotely operated underwater vehicle (ROV) was docked using a robotic arm to test the positioning accuracy of the guide structure. During docking, a visual recognition system was used to observe the fit between the positioning pin and the positioning hole, ensuring precise insertion without deviation or collision. Finally, in the docked state, test data was sent via the battery management unit to verify the connector's signal transmission stability, and to check the data transmission rate and bit error rate, ensuring that the battery's operational data transmission requirements were met. After all tests were passed, the connector's external structure was cleaned and tidied, completing the entire installation process.

[0020] In this embodiment, physical sealing prevents marine organisms from attaching, silting up, and direct seawater erosion in the non-docked state, thus avoiding a decrease in the conductivity of the interface or structural damage. The docking guide structure significantly improves the success rate of remotely operated vehicle (ROV) docking operations. Through mechanical guidance and positioning functions, it compensates for positioning errors during underwater operations, reduces interface collision and wear during docking, and ensures the stability and reliability of the physical connection. The integrated design of the two components allows the connector to have both protective and docking guidance functions, eliminating the need for additional independent protective devices and guide components. This simplifies the structural design of the pressure tank shell and reduces the overall size and weight of the equipment. At the same time, the synergistic effect of sealing and guidance functions provides a guarantee for stable data interaction between the battery pack and external devices, ensuring that the operating data of the battery management unit can be accurately collected.

[0021] In a preferred embodiment of the present invention, step 1 includes: Step 100: The remotely operated vehicle (ROV) moves along the surface of the battery pack pressure chamber to the wet-plug electrical connector installation area. It acquires front and side images of the connector interface using a visual recognition component. Specifically, the shore-based control center pre-configures core parameters and transmits two types of key data to the ROV: first, the three-dimensional structural parameters of the battery pack pressure chamber, including precise dimensions such as a chamber diameter of 1.2 meters and a shell curvature radius of 0.6 meters; second, the preset installation coordinates of the wet-plug electrical connector. These coordinates are defined using a combination of geodetic and local coordinate systems, simultaneously linked to a local coordinate system with the center point at the bottom of the pressure chamber as the origin. Assuming the local coordinates are (X=0.5 meters, Y=0 meters, Z=0.3 meters), where the X-axis points upwards along the chamber's axis, the Y-axis points outwards radially, and the Z-axis is perpendicular to the chamber surface and points outwards, with coordinate accuracy controlled within ±1 centimeter. After the ROV starts, it uses its onboard acoustic positioning system... The remotely operated vehicle (ROV) is equipped with a positioning system, an inertial navigation module, and a high-definition side-scan sonar (scanning range 0 to 50 meters) to capture the contour features of the hull and its own position information in real time. During navigation, the ROV uses the preset installation coordinates as the target point, strictly maintaining a safe distance of 1 meter from the hull, and moves along the surface of the hull at a uniform speed of 0.2 meters per second. During movement, the side-scan sonar continuously scans at a frequency of 10 Hz, comparing the acquired hull contour data with preset three-dimensional structural parameters. When a ring-shaped protrusion structure (characteristic of the connector base with a diameter of 0.2 meters) is detected within the local coordinate range (X=0.5 m ± 0.1 m, Y=0 m ± 0.1 m, Z=0.3 m ± 0.1 m), the hovering procedure is immediately initiated. By fine-tuning the thrusters, the ROV stabilizes itself 1.5 meters directly in front of the connector. At this time, the local coordinates of the ROV are (X=0.5 m, Y=1.5 m, Z=0.3 m), ensuring that the attitude level error does not exceed ±1°.

[0022] Subsequently, the remotely operated vehicle adjusted the angle of its robotic arm, aligning the high-definition camera and the visual recognition component of the adjustable LED fill light module with the connector. For frontal acquisition, the camera lens remained perpendicular to the center of the connector interface, with a preset shooting distance of 0.8 meters (corresponding to local camera coordinates X=0.5 meters, Y=0.7 meters, Z=0.3 meters). This distance could clearly display interface details down to 0.1 millimeters. The fill light module switched to an 800-lumen diffused light mode, with the light beam angled at 45° to the interface surface to avoid glare interference. For side acquisition, the robotic arm moved the visual component laterally along the Z-axis to a position 0.6 meters to the side of the interface. The camera's local coordinates are changed to (X=0.5m, Y=1.5m, Z=0.9m), the lens is kept parallel to the edge of the interface, and the shooting focus is concentrated on the 2mm preset gap between the interface and the protective cover; the supplementary lighting module is switched to 1200-lumen focusing mode to enhance the contrast between the silt and the metal interface in the gap; after the angle and parameter adjustment is completed, the visual recognition component acquires 3 front images and 3 side images in the order of front, side, and front (the resolution of a single image is 4096×3072 pixels). After lossless compression, the image data is transmitted back to the shore-based data processing terminal in real time through an anti-interference underwater acoustic communication link.

[0023] Step 101, based on the marine organism coverage area on the connector interface surface of the front image recognition and the sediment thickness in the connector interface gap of the side image recognition, specifically includes: After receiving the image, the shore-based data processing terminal immediately executes a preset preprocessing procedure, using a 3×3 kernel Gaussian filter algorithm to filter out noise, and using histogram equalization to increase the image grayscale level from 64 levels to 256 levels, clearly outlining the boundary contour; for the front image, the recognition algorithm trained on 50,000 sets of samples is called, the pixel blocks are divided into 16×16 pixels, and the coverage area is marked according to the grayscale value range (marine organisms 50 to 150, interface body 180 to 220), generating a binary image (white 255 represents marine organisms, black 0 represents...). (Table interface); The front image is divided into several independent analysis units with a fixed specification of 16×16 pixels. For each unit, pixels undergo dual feature judgment: first, grayscale value range matching, comparing pixel grayscale values ​​with a preset range. Pixel grayscale values ​​in areas covered by marine life are typically between 50 and 150, while the metal surface of the connector interface body is highly reflective, with pixel grayscale values ​​concentrated between 180 and 220; second, texture feature comparison, distinguishing the irregular texture of marine life from the uniform metallic texture of the interface body by calculating the variance of the grayscale distribution of pixels within the unit. After successful dual judgment, the system marks the pixel range covered by marine life and finally generates a black and white binary image, where the pixel values ​​are... The white area with a pixel value of 255 represents the marine life coverage area, and the black area with a pixel value of 0 represents the connector interface body area. For the side images, the system generates three-dimensional morphological data of the interface gap through a three-step process of feature point extraction, coordinate matching, and 3D point cloud construction. The specific operation is as follows: First, the light and dark boundary line of the interface edge and the texture change points on the surface of the sediment are used as the core identification targets. Through the SIFT feature extraction algorithm, no less than 50 points with stable features are selected from each of the three side images from different perspectives. Each feature point is marked as (u, v) pixel coordinates on the image. At the same time, two key pieces of information are recorded: one is the gray level gradient, that is, the gray level of the pixel and its neighboring pixels. The first step involves the variation range and direction of the value; the second step involves neighborhood features, i.e., the distribution pattern of light and dark within a 3×3 pixel range around the point, forming a unique feature fingerprint for that feature point to avoid matching confusion; the second step is feature point matching and coordinate transformation. A side image with the shooting angle closest to the front is selected as the reference image. Feature points in the other two images are paired with the reference image. Accurate matching is completed based on the feature fingerprint and gray-scale gradient change pattern of the feature points to ensure that the paired feature points correspond to the same physical location; then, coordinate transformation is completed by combining preset camera parameters, i.e., using camera intrinsic parameters (focal length 10 mm, pixel physical size 2 micrometers, optical center pixel coordinates 2048×1536) and extrinsic parameters (shooting distance 0).The camera uses a 6-meter distance and a 0° horizontal lens angle to convert the pixel coordinates (u, v) of the feature points into 3D coordinates (Xc, Yc, Zc) in the camera coordinate system. During the conversion, the origin is the optical center pixel coordinates (u0, v0), and the coordinates are calculated using the formulas Xc = (u - u0) × shooting distance ÷ focal length and Yc = (v - v0) × shooting distance ÷ focal length. The Zc value is fixed at a shooting distance of 0.6 meters, achieving a precise mapping from pixel coordinates to spatial coordinates. The third step is 3D morphology construction. For successfully matched feature points, triangulation is used to obtain depth information. A triangle is constructed using the camera optical centers of the two images as two vertices and the paired feature point as the third vertex. Using the known distance between the two cameras and the shooting angle, the actual vertical distance from the feature point to the camera plane is calculated using trigonometric functions, and the Zc value is corrected to obtain the true depth. The corrected 3D coordinates of all feature points are then aggregated and stitched together to form dense 3D point cloud data. The coordinates of these points are all based on the pressure tank local coordinate system defined earlier, ensuring a one-to-one correspondence with the actual physical location.

[0024] When locating the reference plane for the interface gap based on 3D point cloud data, the points in the interface body area are first selected from the point cloud data. The Z coordinates of these points are relatively concentrated and conform to the surface curvature characteristics of the connector design. Fifty evenly distributed points are selected, and the least squares method is used for plane calculation. That is, through iterative calculation, a plane that minimizes the sum of the squares of the distances from all points to this plane is found, which is used as the lower reference plane for the gap. Combined with the 2 mm gap size marked in the connector design drawings, the upper reference plane is obtained by offsetting it upwards by 2 mm. The two planes together constitute the standard gap space in the unaccumulated state. Then, the points in the accumulation area are delineated in the 3D point cloud data, and the vertical distance from each point to the upper reference plane is measured. The average of these distances is taken as the accumulation thickness of a single image. The measurement results of 3 side images are then arithmetically averaged again. By averaging multiple times, random errors are reduced, and finally, accumulation thickness data with an accuracy controlled within ±0.01 mm is obtained.

[0025] Step 102: The ratio of marine organism coverage area to the total area of ​​the connector interface is used as the first blockage indicator, and the ratio of sediment thickness to the standard gap of the interface is used as the second blockage indicator. When either the first or second blockage indicator exceeds a preset threshold, the connector interface is determined to be blocked. This includes: blockage indicator calculation and status determination. First, basic parameters are retrieved: the total interface area is 0.01 square meters (100 square centimeters), and the standard gap is 2 millimeters, both of which are pre-entered into the system. When calculating the first blockage indicator, the number of marine organism coverage pixels in the frontal image is multiplied by the actual area of ​​a single pixel (0.128 square millimeters, calculated from camera parameters) to obtain the actual coverage area. Then, it is divided by the total interface area (10,000 square millimeters) to obtain the first blockage indicator in percentage form. The second blockage indicator is the ratio of the average thickness of the sediment to the 2-millimeter standard gap. The system calls preset thresholds (30% for the first indicator and 50% for the second indicator). When either indicator exceeds the threshold, the interface is determined to be blocked and a cleaning command is triggered. If both indicators meet the thresholds, the interface is determined to be normal, and the docking preparation process begins.

[0026] This embodiment ensures the comprehensiveness of the detection data through the precise positioning and multi-view image acquisition of the remotely operated vehicle, avoiding the omission of contaminants due to a single shooting angle; the quantitative index calculation method upgrades the blockage judgment from qualitative to quantitative, eliminating subjective errors and ensuring that only blockages that truly affect docking will trigger cleaning operations, avoiding interface damage caused by over-cleaning, and also preventing docking failure due to missed judgments.

[0027] In a preferred embodiment of the present invention, step 2 includes: Step 200: Control the high-pressure water jet cleaning device to perform multiple cycles of cleaning on the connector interface at a preset pressure and preset spray angle to obtain a cleaned connector interface. Specifically, this includes: First, the shore-based control center sends cleaning parameter instructions to the remotely operated vehicle (ROV) based on the previously detected blockage type (marine organism coverage or siltation), presetting the core parameters of the high-pressure water jet. For marine organism coverage, the water jet pressure is set to a safe value matching the connector's pressure resistance level, and the spray angle is based on 45 degrees. For siltation, the pressure is slightly increased, and the spray angle is adjusted to 30 degrees to ensure that the impact force is concentrated inside the gap. After receiving the instruction, the ROV adjusts the posture of its robotic arm, aligning the nozzle of the high-pressure water jet cleaning device with the connector interface. The initial position of the nozzle is 0.5 meters from the interface surface and aligned with the center of the interface. Horizontal alignment; the cleaning process adopts a zoned cyclic mode. The first cycle focuses on the central conductive contact area. The nozzle moves in a clockwise spiral motion with the center of the interface as the origin, gradually increasing in radius, while maintaining a constant speed. Each spraying time is controlled at 2 seconds, completing 3 spraying cycles. The second cycle targets the left and right edge sealing areas. The nozzle moves in a reciprocating straight line along the left and right edges of the interface, covering the entire sealing area. Each edge is sprayed 4 times. The third cycle returns to the entire interface area. The nozzle performs a full-area scanning spray to ensure no blind spots. After each cycle, the visual recognition component quickly takes an image of the interface and transmits it back to the shore terminal to judge the cleaning progress. Cleaning stops and the high-pressure water jet device is retracted when there are no obvious contaminant residues in the image, resulting in a cleaned connector interface.

[0028] Step 201: Acquire images of the cleaned connector interface using a visual recognition component, and extract the two-dimensional image coordinates of each pixel on the connector interface surface from these images. Based on the two-dimensional image coordinates of corresponding pixels in multiple connector interface images acquired from different viewpoints, calculate the depth information of each pixel. Specifically, after cleaning, the remotely operated vehicle immediately activates the positioning program of the visual recognition component, adjusting it to a preset acquisition position that is completely consistent with the previous blockage detection. This position is precisely defined based on the local coordinate system of the pressure chamber, with the center of the connector interface as the reference point. The parameters of the three acquisition viewpoints are as follows: The camera optical center coordinates of the front acquisition position are (X=0.5 m, Y=0.7 m, Z=0.3 m), where the X-axis points to the top along the chamber axis, the Y-axis points outward along the chamber radially, and the Z-axis is perpendicular to the chamber surface; the camera optical center coordinates of the left front acquisition position are (X=0.5 m, Y=0.7 m). The camera's optical center coordinates at the right front acquisition position are (X=0.5m, Y=0.7m, Z=0.1m). The lines connecting the optical centers of the three cameras form an equilateral triangle with a side length of 0.2m. Each camera maintains its lens optical axis pointing towards the center of the interface, and the angle between the lens and the interface surface is 90 degrees. This ensures complete coverage of the entire 0.2m diameter area of ​​the interface surface, while maintaining continuity with the previous detection perspective for easy data comparison. After positioning, the high-definition camera of the visual recognition component acquires images in the order of front, left front, and right front. When acquiring images from the front, the center of the interface is used as the target point to ensure that the edge of the interface is completely captured in the image, acquiring one high-definition image with a resolution of 4096×3072 pixels. When acquiring images from the left front and right front, the viewing angle is shifted 30 degrees to the left and right sides of the interface, respectively. Each image contains complete details of the center of the interface and the corresponding side edge, avoiding blind spots.

[0029] After receiving three images, the shore-based data processing terminal immediately executes a standardized preprocessing procedure. First, a median filter algorithm is used to eliminate image noise caused by residual underwater water mist. Then, edge enhancement is performed using the Laplacian operator to highlight the pixel feature differences between the metallic luster and minor scratches on the interface surface. For each image, the system extracts the two-dimensional image coordinates of the effective pixels on the interface surface in a row-by-row, column-by-column scanning order. The coordinates are centered at the top left corner of the image, with the horizontal axis (range 0 to 4095) and the vertical axis (range 0 to 3071). The coordinates of each pixel are precisely recorded in the form of (horizontal axis value, vertical axis value). For example, the coordinates of the first pixel in the top left corner of the image are (0, 0), its right neighbor is (1, 0), and the lower neighbor is (1, 0). The pixel is (0, 1), and so on to complete the coordinate extraction of all interface pixels. The depth information calculation adopts the triangulation method, with the front image and the left front image as the first set of calculation benchmarks. The specific process is as follows: First, by comparing the pixel gray values ​​and matching the neighborhood texture features, the pixel pair representing the same physical location in the two images is determined. For example, the pixel with coordinates (2048, 1536) in the front image has a texture feature that is completely matched with the pixel with coordinates (2060, 1536) in the left front image, which constitutes a valid pixel pair. It is known that the actual distance (baseline distance) between the optical centers of the two cameras is 0.2 meters, and the camera attitude angle is fixed at 0 degrees (horizontal shooting). A triangle model is constructed with the optical centers of the two cameras as vertices and the matched pixels as the third vertex.

[0030] First, calculate the disparity of the pixel pair, which is the difference in the horizontal axis coordinates of the same pixel in the two images. The formula is: disparity = horizontal axis coordinate of the pixel in the front image - horizontal axis coordinate of the corresponding pixel in the left front image. In this example, the disparity is the absolute value of 2048 - 2060, which is 12 pixels. Then, calculate the depth information using the formula: depth = baseline distance × camera focal length ÷ disparity. The camera focal length is preset to 10 mm. After substituting the value, the depth is 0.2 m × 10 mm ÷ 12 pixels. After unifying the units, the specific depth value is obtained. Using the same method, combine the pixel matching results of the front image and the right front image to complete the second calculation of the depth information. Take the arithmetic mean of the two calculation results as the final depth information of the pixel, which effectively reduces the error of a single measurement.

[0031] Step 202 involves fusing the two-dimensional image coordinates and depth information of each pixel to generate multiple spatial coordinate points on the connector interface surface. Some of these spatial coordinate points are located in the central conductive contact area, some in the left edge sealing area, and some in the right edge sealing area. This process includes coordinate fusion and spatial point generation, as well as partitioning. First, coordinate fusion is performed. Based on the two-dimensional coordinates of the front image, the conversion is completed using preset camera intrinsic parameters (pixel physical size 2 micrometers, optical center pixel coordinates 2048×1536) to obtain the three-dimensional coordinates of the pixel in the camera coordinate system. The conversion formulas are as follows: Horizontal coordinate = (two-dimensional image horizontal axis coordinate - optical center horizontal coordinate) × pixel physical size × depth ÷ camera focal length; Vertical coordinate = (two-dimensional image vertical axis coordinate - optical center vertical coordinate) × pixel physical size × depth ÷ camera focal length; Depth coordinate = depth information. This calculation transforms the two-dimensional image coordinates of each pixel into spatial coordinate points containing three dimensions: horizontal, vertical, and depth, forming a dense set of spatial coordinate points covering the interface surface.

[0032] Subsequently, the area was divided. Based on the connector design drawings, the overall diameter of the interface was determined to be 0.2 meters. Using this as a benchmark, the geometric ranges of the three functional areas were defined and specific values ​​were given. The central conductive contact area is a circular area with a fixed radius of 0.05 meters centered on the interface center; the left edge sealing area is an annular strip-shaped area from the left edge of the interface (0.1 meters from the center) to the central area (0.05 meters from the center), with an annular width of 0.05 meters; the right edge sealing area is an annular strip-shaped area from the right edge of the interface (0.1 meters from the center) to the central area (0.05 meters from the center), with an annular width of... Also 0.05 meters; the system calculates the straight-line distance from each spatial coordinate point to the center of the interface using a distance formula, and compares it with the range of each area to complete the partitioning. That is, coordinate points with a straight-line distance ≤ 0.05 meters are assigned to the central conductive contact area; coordinate points with a straight-line distance between 0.05 meters and 0.1 meters and located on the left side of the interface (horizontal coordinate < horizontal coordinate of the interface center) are assigned to the left edge sealing area; coordinate points with a straight-line distance between 0.05 meters and 0.1 meters and located on the right side of the interface (horizontal coordinate > horizontal coordinate of the interface center) are assigned to the right edge sealing area. Finally, multiple sets of spatial coordinate point data belonging to three functional areas are formed.

[0033] This embodiment first uses a zoned, circulating high-pressure water jet cleaning method to specifically remove contaminants from different areas, avoiding the problems of incomplete cleaning or over-cleaning that could damage the interface and ensuring that the connector interface is restored to a suitable docking state. In the coordinate acquisition and processing stage, multi-view image acquisition combined with triangulation to calculate depth improves the accuracy of spatial coordinate points. Compared with single-view acquisition, it effectively avoids coordinate deviations caused by surface depressions or protrusions on the interface. Dividing coordinate points according to functional areas makes the subsequent construction of plane equations more targeted and can reflect the geometry of key areas of the interface. The entire process achieves seamless integration of cleaning and data acquisition without the need for additional adjustments to the remotely operated vehicle's attitude, thus improving underwater operation efficiency.

[0034] In a preferred embodiment of the present invention, step 3 includes: Step 300: Based on the three-dimensional coordinate distribution of multiple spatial coordinate points, the multiple spatial coordinate points located in the geometric center region of the connector interface are divided into a central conductive contact area, the multiple spatial coordinate points located in the left edge region of the connector interface are divided into a left edge sealing area, and the multiple spatial coordinate points located in the right edge region of the connector interface are divided into a right edge sealing area. Specifically, this includes: region division based on spatial coordinates. First, the geometric reference of the connector interface is defined. Taking the projection point of the interface center on the horizontal plane as the origin (coordinates set to (0, 0, 0)), a local coordinate system for region division is established. This coordinate system is consistent with the coordinates collected in the previous stage. The local coordinate system used in the pressure chamber is kept consistent to ensure the continuity of coordinate data. Based on the connector design specifications (overall interface diameter is 0.2 meters), a regional division standard is set. The central conductive contact area is defined by a circular area with a fixed radius of 0.05 meters centered on the overall geometric center of the interface as the origin (the boundary threshold of the central conductive contact area is 0.05 meters). Using the left edge contour line of the interface (at a horizontal coordinate of -0.1 meters from the origin) as a reference, a fixed width of 0.03 meters is extended inward to form the left edge sealing area (the horizontal coordinate range of this area is -0.1 meters to -0.07 meters). The right edge contour line of the interface is used as the sealing area. Using the side edge contour line (0.1 meters from the origin at horizontal coordinates) as a reference, a fixed width of 0.03 meters is extended inward to form the right edge sealing area, with a horizontal coordinate range of 0.07 meters to 0.1 meters. During the area division operation, the system first extracts the horizontal and vertical coordinates of all spatial coordinate points, and calculates the straight-line distance from each point to the interface geometric center (0, 0, 0) using the Pythagorean theorem. For a single spatial coordinate point, if its straight-line distance to the geometric center is less than or equal to 0.05 meters, then the point is determined to belong to the central conductive contact area; if its straight-line distance to the geometric center is greater than 0.05 meters, and the horizontal coordinate is greater than or equal to 0.05 meters, then the point belongs to the central conductive contact area. If the coordinate of the point is less than the horizontal coordinate (0) of the geometric center and is located within a preset range of -0.1 meters to -0.07 meters on the left edge of the interface, it is determined to be the left edge sealed area; if the straight-line distance from the point to the geometric center is greater than 0.05 meters and the horizontal coordinate is greater than the horizontal coordinate (0) of the geometric center, and is located within a preset range of 0.07 meters to 0.1 meters on the right edge of the interface, it is determined to be the right edge sealed area; after the division is completed, the system classifies and stores the coordinate points of the three areas separately, forming three independent coordinate point datasets. Each dataset contains the horizontal, vertical, and depth coordinate information of all points in the corresponding area.

[0035] Step 301: Calculate the center coordinates of the central conductive contact area based on multiple spatial coordinate points of the central conductive contact area; calculate the left edge coordinates of the left edge sealing area based on multiple spatial coordinate points of the left edge sealing area; and calculate the right edge coordinates of the right edge sealing area based on multiple spatial coordinate points of the right edge sealing area. Specifically, this includes: calculating the coordinates of the core feature positions for each of the three regions' coordinate point datasets, with the calculation process based on the arithmetic mean of the three-dimensional coordinates to ensure the results reflect the overall positional characteristics of the regions; for the center position coordinate calculation, for the coordinate point dataset of the central conductive contact area, first extract the horizontal coordinates of all points in the dataset, and then add these coordinate values ​​to obtain the horizontal coordinates. The sum of these coordinates is divided by the number of coordinate points in the area to obtain the average horizontal coordinates. Using the same method, the average vertical coordinates and average depth coordinates are calculated separately. These three average coordinates together constitute the center position coordinates of the central conductive contact area. That is, the horizontal value of the center position coordinates = the sum of the horizontal coordinates of all points in the center area ÷ the number of points in the center area; the vertical value = the sum of the vertical coordinates of all points in the center area ÷ the number of points in the center area; and the depth value = the sum of the depth coordinates of all points in the center area ÷ the number of points in the center area. For example, if the central conductive contact area contains 120 coordinate points, and the sum of the horizontal coordinates of all points is 0.6 meters, then the average horizontal coordinates = 0.6 meters ÷ 120 = 0.005 meters. The calculation of the vertical and depth coordinates follows the same principle.

[0036] The left edge position coordinate calculation is performed on the coordinate point dataset of the left edge sealing area. The core feature of this area is its proximity to the left edge of the interface. Therefore, the points with the smallest horizontal coordinates in the dataset are first selected (these points are closest to the left edge). Then, the arithmetic mean of the three-dimensional coordinates of these points is calculated. Specifically, the horizontal coordinates of all points in the left edge area are extracted, and the minimum value is found (e.g., -0.099 meters, close to the design value of -0.1 meters for the left edge). All points with horizontal coordinates equal to this minimum value are selected to form an edge subset. If the number of points in this subset is less than 20 (e.g., only 12), the selection range is expanded to select points with horizontal coordinates in the range of the minimum value to the minimum value + 0.003 meters (i.e., -0.099 meters to -0.096 meters) to form a subset. Then, the horizontal, vertical, and depth average coordinates of this subset are calculated to obtain the left edge position coordinates of the left edge sealing area.

[0037] The calculation of the right edge position coordinates is similar to that of the left edge. Extract the horizontal coordinates of all points in the right edge sealing area, find the maximum value, such as 0.098 meters, which is close to the design value of 0.1 meters for the right edge, and select all points with horizontal coordinates equal to this maximum value to form the right edge subset. If the subset has a small number of points, expand it to points with horizontal coordinates in the range of -0.003 meters to the maximum value. Calculate the arithmetic mean of the three-dimensional coordinates of the subset, which gives the right edge position coordinates of the right edge sealing area. During the calculation, the selection range is expanded to ensure that the number of points in the subset is sufficient and to ensure the stability of the calculation results.

[0038] Step 302: Based on the coordinates of the center position, the left edge position, and the right edge position, determine the plane space equation. Specifically, this includes: the core of the plane space equation is to solve for the general expression of the plane using the coordinates of three non-collinear feature points; it is known that three non-collinear points in space can uniquely determine a plane. The center position coordinates, left edge position coordinates, and right edge position coordinates calculated in step 301 precisely satisfy the non-collinearity condition, and the plane equation is derived based on this. First, the general form of the plane space equation is defined as Ax + By + Cz + D = 0, where A, B, and C are the three components of the plane normal vector, D is a constant term, and x, y, and z represent the horizontal, vertical, and depth coordinates of the spatial point, respectively. Substituting the coordinates of the three feature points into this general form yields a system of three linear equations. For example, let the center position coordinates be (0.005 m, 0.002 m, 0.05 m), the left edge position coordinates be (-0.098 m, 0.001 m, 0.051 m), and the right edge position coordinates be (0.099 m, 0.003 m, 0.049 m). Substituting these values, we get: A×0.005 + B×0.002 + C×0.05 + D = 0; A×(-0.098) + B×0.001 + C×0.051 + D = 0; A×0.099 + B×0.003 + C×0.049 + D = 0; To solve this system of equations, first eliminate D using the first two equations to obtain linear equations about A, B, and C; then eliminate D using the first and third equations to obtain another linear equation about A, B, and C; combine these two equations, assuming A is 1 (the coefficients of the plane equations can be proportional, setting A to 1 does not affect the plane position), substitute them to solve for the specific values ​​of B and C, substitute A=1 to solve for B and C, then substitute the values ​​of A, B, and C into the first equation to calculate the value of D, and substitute the solved A, B, C, and D into the general form of the plane to determine the planar space equation of the connector interface, which reflects the ideal reference plane of the interface surface.

[0039] This embodiment achieves the division of interface functional areas through coordinate analysis, avoiding errors caused by subjective human judgment in the division method, and ensuring that the definition of key areas such as the central conductive contact area and the edge sealing area conforms to the actual working requirements of the connector. The calculation of feature position coordinates adopts a multi-coordinate point averaging method, which effectively offsets the measurement error of a single coordinate point, making the obtained center position and edge position more reflective of the true geometric characteristics of the area, and providing reliable basic data for the construction of the plane equation. The plane space equation is determined based on three key feature points, ensuring that the equation can accurately fit the ideal geometric shape of the connector interface. Compared with the plane fitted by a single area, this equation covers the core functional area of ​​the interface and is more representative.

[0040] In a preferred embodiment of the present invention, step 4 includes: Step 400: Based on the planar spatial equation, calculate the vertical distance from each spatial coordinate point to the planar spatial equation, and take the arithmetic mean of the vertical distances of all spatial coordinate points as the surface deviation metric. Specifically, this includes: Surface deviation metric calculation. First, clarify the calculation basis. Using the planar spatial equation determined in step 302 as the benchmark, calculate the vertical distance from all used spatial coordinate points (covering the central conductive contact area, left and right edge sealing areas) to the plane. This distance directly reflects the degree of deviation of the interface surface position represented by a single coordinate point from the ideal plane. The vertical distance calculation follows the formula for the distance from a spatial point to a plane. The specific steps are as follows: For a single spatial coordinate point (x0, y0, z0), the first step is to calculate the absolute value of the point substituted into the left side of the planar equation, i.e., calculate Ax0 + By0 + Cz0 + ... The absolute value of the result D is recorded as the numerator. The second step is to calculate the modulus of the plane equation coefficients, that is, to sum the squares of A, B, and C and take the square root, which is recorded as the denominator. The third step is to divide the numerator by the denominator to obtain the perpendicular distance from the point to the reference plane. After completing the perpendicular distance calculation for all spatial coordinate points, add all the distance values ​​to get the total distance, and then divide by the total number of coordinate points. The resulting arithmetic mean is the surface deviation metric. The larger the metric value, the more serious the degree of deviation of the connector interface surface from the ideal plane.

[0041] Step 401: Construct a triangular mesh surface based on multiple spatial coordinate points, calculate the normal vector direction of each triangular face in the triangular mesh surface, and calculate the standard deviation of the angle change between the normal vector directions of all adjacent triangular faces as the normal vector change rate. Specifically, this includes: calculating the normal vector change rate. By constructing a triangular mesh model of the interface surface, analyzing the change characteristics of the local surface morphology, and quantifying the degree of local deformation, the triangular mesh surface is first constructed. Using the horizontal and vertical coordinates of all spatial coordinate points as plane projection references, adjacent coordinate points in space are connected sequentially to form multiple non-overlapping and gap-free triangular units. Each triangular unit corresponds to a small region of the interface surface. All triangular units are spliced ​​together to form a complete triangle. The mesh surface is designed to ensure that the mesh shape matches the actual surface contour of the connector interface. Next, the normal vector direction of each triangular face is calculated. For a single triangular unit, the complete spatial coordinates of its three vertices are selected and denoted as Vertex 1 (Horizontal 1, Vertical 1, Depth 1), Vertex 2 (Horizontal 2, Vertical 2, Depth 2), and Vertex 3 (Horizontal 3, Vertical 3, Depth 3). The first step constructs two edge vectors for each triangular face. Edge vector 1 is the coordinate difference between Vertex 2 and Vertex 1, calculated as follows: horizontal difference equals horizontal 2 minus horizontal 1; vertical difference equals vertical 2 minus vertical 1; depth difference equals depth 2 minus depth 1. Edge vector 2 is the coordinate difference between Vertex 3 and Vertex 1, calculated as follows: horizontal difference equals horizontal 2 minus horizontal 1; vertical difference equals vertical 2 minus vertical 1; depth difference equals depth 2 minus depth 1. The first step is to subtract the horizontal direction difference from the horizontal direction difference of the first vector. The vertical difference equals the vertical direction difference of the third vector minus the vertical direction difference of the first vector. The depth difference equals the depth difference of the third vector minus the depth difference of the first vector. The second step is to calculate the normal vector using the cross product of the edge vectors. The three components of the cross product are calculated as follows: the first component equals the vertical direction difference of edge vector one multiplied by the depth direction difference of edge vector two, minus the depth direction difference of edge vector one multiplied by the vertical direction difference of edge vector two; the second component equals the depth direction difference of edge vector one multiplied by the horizontal direction difference of edge vector two, minus the horizontal direction difference of edge vector one multiplied by the depth direction difference of edge vector two; the third component equals the horizontal direction difference of edge vector one multiplied by the vertical direction difference of edge vector two, minus the vertical direction difference of edge vector one multiplied by the depth difference of edge vector two. The third step is to normalize the cross product result by calculating the magnitude of the cross product (taking the square root of the sum of the squares of the three components), and then dividing each of the three components by the magnitude to obtain the unit normal vector of the triangle face, whose direction represents the spatial orientation of the triangle face; then, the angle changes of the normal vectors of adjacent triangle faces are statistically analyzed, all triangle faces are traversed, and the adjacent triangle faces of each triangle face (i.e., triangle faces sharing a side) are determined, and the included angle between the normal vectors of each pair of adjacent triangle faces is calculated; the included angle is calculated by multiplying the corresponding components of the two normal vectors and then summing them to obtain the dot product result. Since it is a unit normal vector, its magnitude is 1, so the dot product result is directly equal to the cosine value of the included angle, and the angle value can be obtained by using the inverse cosine function.

[0042] Collect the included angle values ​​of all adjacent triangular faces, and calculate the standard deviation of these angle values ​​as the rate of change of the normal vector. The first step is to calculate the arithmetic mean of all angle values, that is, the sum of angles divided by the total number of angles. The second step is to calculate the difference between each angle value and the mean, square each difference and sum them to get the sum of squares. The third step is to divide the sum of squares by (the total number of angles minus 1) to get the variance. The fourth step is to take the square root of the variance. The result is the rate of change of the normal vector. The larger this value is, the more severe the local undulations on the interface surface.

[0043] In this embodiment, the surface deviation measurement uses an ideal plane as a reference to reflect the flatness of the interface from an overall perspective, avoiding the one-sidedness of judging deformation based solely on local observations. This ensures that issues affecting the docking contact area, such as overall interface tilting and denting, are captured. The normal vector change rate focuses on the local morphology of the interface surface. Through triangular mesh and normal vector analysis, minute local protrusions, wrinkles, and other detailed deformations are identified. The calculation of both indicators is based on prior spatial coordinate data and plane equations. The calculation ensures the objectivity of the results and eliminates the subjective errors of manual evaluation. At the same time, the process of triangular mesh construction and normal vector analysis transforms discrete coordinate points into continuous surface morphology data, elevating deformation assessment from the point level to the surface level, providing a basis for the formulation of subsequent compensation schemes.

[0044] In a preferred embodiment of the present invention, step 5 includes: Step 500: Linearly weighted combination of surface deviation measurement and normal vector change rate according to preset weight coefficients to calculate the overall deformation index of the connector interface. Specifically, this includes: calculating the overall deformation index by first addressing the dimensional difference between surface deviation measurement (reflecting spatial distance characteristics) and normal vector change rate (reflecting angular change characteristics), and then combining them into a single index reflecting the overall deformation state of the interface. The core processing logic is to normalize the two indices to a dimensionless range of 0 to 1, eliminating the influence of units before weighted combination. First, index normalization is completed. Based on historical monitoring data of connector interface deformation and design allowable thresholds, the effective value range and normalization benchmark of the two indices are determined. The historical maximum effective value of surface deviation measurement is 0.1 mm, and the historical maximum effective value of normal vector change rate is 1 degree. The normalization calculation method is: normalized value of a single index = actual measured value of the index ÷ corresponding... The historical maximum effective value of the indicator is used, and the calculation result is retained to three decimal places to ensure accuracy and achieve dimensionlessness. Then, preset weight coefficients are determined. The weight allocation is based on the degree of influence of the two indicators on the docking performance. The normalized result of the surface deviation measure reflects the overall flatness and is directly related to the stability of conductive contact, so it has a higher weight and is set to 0.6. The normalized result of the normal vector change rate reflects the local morphological fluctuations and is closely related to the sealing performance, so it is set to 0.4. The sum of the two coefficients is 1 to ensure the balance of the evaluation dimensions. The calculation process is divided into three steps. The first step is to calculate the weighted contribution value of the surface deviation measure, that is, its normalized value multiplied by the corresponding weight coefficient 0.6. The second step is to calculate the weighted contribution value of the normal vector change rate, that is, its normalized value multiplied by the corresponding weight coefficient 0.4. The third step is to add the two weighted contribution values. The sum is the dimensionless overall deformation index of the connector interface. The index value range is simultaneously constrained to between 0 and 1.

[0045] Step 501: Obtain the corresponding deformation compensation coefficient based on the overall deformation index through a predefined mapping table. Specifically, this involves: The core of obtaining the deformation compensation coefficient is constructing a predefined mapping table between the overall deformation index interval and the deformation compensation coefficient. This table is determined based on a large amount of underwater docking data and simulation results. Specifically, it records the compensation parameters required to achieve optimal contact and sealing effects by simulating the matching process between the interface and docking device with different normalized deformation degrees. The interval division and coefficient values ​​have been repeatedly verified to ensure they are suitable for actual engineering needs and completely match the dimensionless index from step 500. In practice, the system first extracts the dimensionless overall deformation index calculated in step 500. The index range and corresponding compensation coefficient are determined by comparing it with the index range in the mapping table. The compensation methods corresponding to different coefficients are based on the position adjustment of the docking device and are all performed based on the local coordinate system of the pressure chamber established in the early stage to ensure the adjustment accuracy and coordinate continuity. The specific compensation logic is as follows: The mapping table clearly stipulates that when the overall deformation index is between 0 and 0.1, the corresponding compensation coefficient is 1.0, which means that the interface deformation is minimal and no additional compensation is required. At this time, the docking device maintains the preset reference posture, that is, it is parallel and aligned with the ideal plane of the interface, and the central axis coincides with the center of the interface. It directly enters the subsequent docking process, and the docking effect can be guaranteed by the manufacturing accuracy of the device itself.

[0046] When the overall deformation index is between 0.1 and 0.3, the corresponding compensation coefficient is 1.05, indicating that there is slight deformation at the interface, requiring slight compensation. The compensation operation is based on the reference adjustment amount of the docking device (the preset reference adjustment amount is 0.2 mm translation on the X-axis, 0.2 mm translation on the Y-axis, and 0.1 degrees rotation around the Z-axis). The specific adjustment value is calculated by "reference adjustment amount × compensation coefficient", that is, 0.2 × 1.05 translation on the X-axis, 0.2 × 1.05 translation on the Y-axis, and 0.1 × 1.05 rotation around the Z-axis. The adjustment direction is determined by the deviation analysis of the spatial coordinate points in the early stage. For example, if the interface shifts in the positive X-axis direction, the docking device will simultaneously translate in the positive X-axis direction to ensure that the conductive contact end of the device is accurately aligned with the central conductive contact area of ​​the interface and that the sealing structure fits snugly with the edge sealing area.

[0047] When the overall deformation index is between 0.3 and 0.5, the corresponding compensation coefficient is 1.1, indicating that the interface has moderate deformation and requires moderate compensation. At this time, the compensation dimension adds a fine adjustment in the depth direction (Z-axis) on the basis of translation and rotation. The baseline adjustment amount is upgraded to: X-axis translation 0.5 mm, Y-axis translation 0.5 mm, Z-axis translation 0.3 mm, and rotation around the Z-axis 0.3 degrees. The specific adjustment value is calculated by multiplying the baseline adjustment amount by the compensation coefficient, that is, X-axis 0.5 × 1.1, Y-axis 0.5 × 1.1, Z-axis 0.3 × 1.1, and rotation around the Z-axis 0.3 × 1.1. Before adjustment, the system will combine the deviation trend of the spatial coordinate points of each area to determine the adjustment direction of each dimension. For example, if the normal vector of the left edge sealing area is biased towards the negative Y-axis direction, the docking device will rotate counterclockwise around the Z-axis by the corresponding angle to make the sealing surface of the device completely fit with the sealing area of ​​the interface, avoiding sealing failure due to local deformation.

[0048] After all compensation parameters are calculated, the system will transmit the adjustment command to the docking device control module of the remotely operated vehicle via the underwater acoustic communication link. The control module drives the thruster and fine-tuning motor to perform attitude adjustment. During the adjustment process, the position feedback data of the device is collected in real time to ensure that the deviation between the actual adjustment amount and the calculated value is controlled within ±0.01 mm (translation) and ±0.01 degrees (rotation).

[0049] This embodiment employs a linear weighted combination method to integrate deformation indicators from both overall and local dimensions into a unified overall deformation index, resolving the issues of ambiguous weights and difficulty in synthesizing results in multi-indicator evaluation. The weighting coefficients are set in accordance with the actual needs of interface docking, ensuring that the index prioritizes deformation characteristics that have a greater impact on docking performance, thus guaranteeing a close correlation between the evaluation results and engineering practice. A predefined mapping table transforms the abstract deformation index into concrete, executable compensation coefficients, achieving seamless integration of deformation evaluation and docking adjustment. The system can quickly output compensation schemes based on the deformation state, improving the automation and success rate of underwater docking while reducing problems such as poor contact and sealing failure caused by deformation, ensuring the stability of battery pack data transmission and power supply.

[0050] In a preferred embodiment of the present invention, step 6 includes: Step 600: Adjust the pose parameters of the docking device according to the deformation compensation coefficient to match and dock with the docking guide structure of the wet-plug electrical connector, establishing a stable physical connection. Specifically, this includes: using the acquired deformation compensation coefficient as the core basis, combined with the reference pose parameters of the docking device, to complete precise pose adjustment and achieve physical connection. First, it is clarified that the pose parameters of the docking device include three translational dimensions (along the X, Y, and Z axes of the local coordinate system of the pressure chamber) and three rotational dimensions (rotation angles around the X, Y, and Z axes). The reference pose parameters are the pre-set ideal docking position, i.e., all translations are 0 and all rotation angles are 0. The pose adjustment amount is calculated according to the reference… The logic of parameter × deformation compensation coefficient is calculated separately for different dimensions. If the deformation compensation coefficient is 1.05, the X-axis reference translation is 0.5 mm (for aligning the interface center), then the actual X-axis translation is 0.5 × 1.05; the Y-axis reference translation is 0.3 mm (for fitting the cabin surface), so the actual translation is 0.3 × 1.05; the Z-axis reference translation is 0.2 mm (for controlling the docking depth), so the actual translation is 0.2 × 1.05; in the rotation dimension, the rotation angle around the Z-axis reference is 0.1 degrees (to correct the horizontal offset of the interface), so the actual rotation angle is 0.1 × 1.05. The other rotation dimensions are not involved due to interface deformation, so the reference value is kept at 0.

[0051] The adjustment command is transmitted from the shore-based control center to the docking control module of the remotely operated vehicle (ROV) via an underwater acoustic communication link. The module drives the six-degree-of-freedom thrusters and fine-tuning motors to perform actions, first completing the translational adjustment to align the central axis of the docking device with the center of the connector interface; then performing the rotational adjustment to ensure that the sealing surface of the docking device is parallel to the sealing area of ​​the interface. During the adjustment process, the visual recognition component acquires images of the relative positions of the docking device and the interface in real time, calculates the deviation value between them, and if the deviation is greater than 0.02, it is fed back to the control module for secondary fine-tuning until the deviation is less than 0.02; after the attitude adjustment is completed, the docking device... The device slowly advances along the positive Z-axis at a controlled speed of 0.1 mm per second. When the guide pin of the docking device is inserted into the guide hole of the connector interface (the gap between the guide pin and the guide hole is 0.05 mm), the position sensor signal is triggered, and the advancing speed is reduced to 0.05 mm per second. The device continues to advance until the conductive contact of the docking device is fully engaged with the central conductive contact area of ​​the interface, and the sealing ring is in tight contact with the edge sealing area. At this time, the locking mechanism of the docking device is automatically activated, and the locking claw is driven by the motor to engage the interface flange, completing a stable physical connection. The locking force is monitored in real time by a pressure sensor to ensure that the preset sealing pressure requirement is met.

[0052] Step 601: Activate the high-speed data communication link through the established physical connection, and read the complete historical operating data of the battery pack from the battery management unit, including cell voltage curves, temperature curves, and charge / discharge cycle records. Specifically, after the physical connection is established, the power supply pin of the docking device first forms a path with the power supply contact of the interface, providing working power to the communication module built into the wet-plug connector. After the communication module is powered on, it automatically sends a handshake signal. The communication pin of the docking device receives the signal and sends back an acknowledgment signal, completing the activation of the high-speed data communication link. The link communication rate is preset to 100 megabits per second, and the bit error rate is controlled below one in a million. After the link is activated, the docking device sends a data read command to the battery management unit through the communication protocol. The command includes parameters such as data type and read time range. The battery management unit responds. Following the command, data is transmitted in the order of cell voltage curve, temperature curve, and charge / discharge cycle records. The cell voltage curve contains voltage change data for each cell over the last 300 charge / discharge cycles, with the voltage value recorded every ten seconds. The temperature curve contains temperature data for the core area of ​​the battery pack and the surface of each cell, with the recording interval consistent with the voltage curve. The charge / discharge cycle records contain key parameters such as the charging start voltage, discharging end voltage, charging duration, and discharging duration for each cycle. During data reading, the docking device performs real-time verification of the received data, using an additive checksum method. All bytes of each data segment are summed to obtain the cumulative sum, which is then compared with the checksum sent by the battery management unit. If they do not match, a retransmission command is sent; if they match, the data is marked as valid and stored in a temporary buffer to ensure data integrity and error-free operation.

[0053] Step 602: The complete historical operating data is transmitted to the shore-based control center via the underwater communication module of the remotely operated vehicle (ROV) for visualized diagnosis of battery health status and safety risk assessment of performance degradation. Specifically, this includes: battery data in the temporary buffer is first processed by a compression algorithm to reduce data volume to match the underwater communication bandwidth; the compressed data is then encrypted using the AES encryption algorithm to ensure data transmission security; the encrypted data is then transmitted to the ROV's underwater communication module, which employs a dual-mode transmission method using primary acoustic communication and secondary optical communication. In shallow water (depth less than 200 meters), optical communication is prioritized with a transmission rate of up to 50 megabits per second; in deep water, it switches to acoustic communication with a transmission rate of 1 megabit per second, with automatic and seamless switching between the two modes. Upon receiving the data transmitted from underwater, the receiving module at the shore-based control center immediately initiates the data restoration process. First, the data encryption protection is removed using the AES decryption algorithm, and then the compressed data packet is restored using a decompression algorithm to completely recover the original battery operating data. The restored data is then transmitted to the battery health diagnosis system, which simultaneously performs visualized diagnosis. The system performs two core tasks: data visualization and health assessment. Specifically, in the data visualization stage, the system uses differentiated presentation methods for different types of data. The cell voltage curve generates an independent line graph for each cell, with time as the horizontal axis and voltage as the vertical axis. The standard voltage range for this type of lithium battery is also overlaid on the graph. The normal operating voltage of a single cell is 2.75V to 4.2V, the charging cut-off voltage does not exceed 4.25V, and the discharging cut-off voltage is not lower than 2.7V. If the voltage exceeds this range at a certain time, it is automatically marked with a red line segment. The temperature curve is presented in the form of a heat map, using the physical layout of the battery pack as a template. Different color gradients visually display the temperature distribution on the surface of each cell and in the core area of ​​the battery pack. Red represents the over-temperature risk area, and blue represents the normal temperature area. The charge / discharge cycle records are summarized in a structured table. The table columns include key parameters such as cycle number, charging start time, charging end time, charging start voltage, charging end voltage, discharging start time, discharging end time, discharging start voltage, discharging end voltage, and single charge / discharge capacity, facilitating rapid location of abnormal cycle data.

[0054] In the health status assessment phase, the system completes risk judgment and performance degradation assessment in two steps. The first step is abnormal risk identification. The system scans the voltage curve of each cell one by one. If the voltage of a cell exceeds 4.25 volts (overcharge) or falls below 2.7 volts (over-discharge) for more than 10 seconds, it is immediately marked as a voltage anomaly. At the same time, the voltage difference of each cell at the same moment is calculated. If the maximum difference exceeds 0.1 volts, it is marked as a voltage consistency anomaly. Both anomalies are included in the risk item. The second step is battery performance degradation assessment. This assessment uses capacity degradation rate as the core indicator. The standard, combined with the effects of cycle number and temperature, is calculated as follows: First, the single discharge capacity of the most recent 10 complete charge-discharge cycles is extracted from the charge-discharge cycle records, and its average value is calculated as the current actual usable capacity. Then, the rated capacity parameter of the battery pack is retrieved, and the base decay rate is calculated using the formula: Capacity decay rate = (Rated capacity - Current actual usable capacity) ÷ Rated capacity × 100%. Next, the base decay rate is corrected based on the number of cycles, i.e., the battery cycle life curve is consulted. This curve is provided by the manufacturer, specifying that the normal decay rate is 10% after 500 cycles. The normal degradation rate after 1000 cycles is 20%. If the current battery cycle count is 600, and the base degradation rate is 12%, then the corrected degradation rate = 12% - (12% - 10%) × 0.5 = 11%, where 0.5 is the cycle count correction factor, calculated from the difference between the actual cycle count and the standard cycle node. Finally, further correction is made based on the effect of temperature. The proportion of high-temperature operating time exceeding 45℃ in the temperature curve is statistically analyzed. If the proportion is 15%, then the degradation rate increases by 1% for every 10% increase in high-temperature time. The final degradation rate = 11% + (15% ÷ 10%) × 1% = 12.5%; The system determines the battery health level based on the final degradation rate and abnormal risk items. That is, a degradation rate of less than 10% and no abnormal risk is considered healthy; a degradation rate of 10% to 20% or the presence of slight voltage consistency abnormalities is considered slight degradation; a degradation rate of 20% to 30% or the presence of overcharge and over-discharge records is considered moderate degradation; and a degradation rate of more than 30% is considered severe degradation. Finally, the system integrates the visualization charts, abnormal risk descriptions, degradation rate calculation process, health level, and other content to generate a complete battery health diagnosis conclusion and risk level assessment report.

[0055] In this embodiment, the pose adjustment process achieves precise adaptation through deformation compensation coefficients, avoiding docking deviations caused by interface deformation and improving the docking success rate of wet-plug electrical connectors. Simultaneously, it ensures effective contact between the conductive contact area and the sealing area, guaranteeing the stability of power transmission and preventing seawater from seeping into the interface and damaging the equipment. The activation and data reading process of the high-speed data communication link enables complete collection of historical operating data of the battery pack. Compared with sampling methods, this provides a more comprehensive reflection of the battery's long-term operating status, offering rich data support for health diagnosis. The shore-based visualization diagnosis and risk assessment transform abstract battery data into intuitive charts and clear conclusions, reducing the data analysis difficulty for maintenance personnel. It makes the battery's health status and performance degradation risks readily apparent, helping to predict potential faults in advance, formulate targeted maintenance strategies, extend the battery pack's lifespan, and ensure the continuous and stable operation of underwater equipment.

[0056] like Figure 2 As shown, embodiments of the present invention also provide a data acquisition system for the visual management of underwater pressure-resistant battery packs, including: The positioning and visual recognition module is used for the remotely operated vehicle to descend to the battery pack pressure chamber and locate the wet-plug electrical connector; it acquires real-time images of the wet-plug electrical connector through the visual recognition component of the remotely operated vehicle and identifies the blockage status of the connector interface based on the real-time images; The adaptive cleaning module controls the cleaning device of the remotely operated vehicle to perform high-pressure water jet cleaning on the connector interface to obtain a cleaned connector interface; it acquires a cleaned image through a vision recognition component and extracts multiple spatial coordinate points on the surface of the connector interface from the cleaned image. The geometric partitioning and plane fitting module is used to determine the central conductive contact area, left edge sealing area and right edge sealing area on the connector interface based on multiple spatial coordinate points, and to determine the plane space equation through the three areas; The deformation analysis module is used to calculate the average distance between multiple spatial coordinate points and the plane spatial equation as a surface deviation metric, and to calculate the rate of change of the normal vector of multiple spatial coordinate points, based on the plane spatial equation. The deformation compensation coefficient generation module is used to calculate the overall deformation index of the connector interface by combining the surface deviation metric and the normal vector change rate, and to generate the deformation compensation coefficient based on the overall deformation index. The adaptive docking and data feedback module is used to control the docking device to establish a physical connection with the wet plug-in electrical connector by utilizing the deformation compensation coefficient and docking guide structure. This allows the module to read complete historical operating data of the battery pack from the battery management unit and transmit the historical operating data to the shore-based control center for visual diagnosis and evaluation of battery health status.

[0057] It should be noted that this system is a system corresponding to the above method. All implementation methods in the above method embodiments are applicable to this embodiment and can achieve the same technical effect.

[0058] Embodiments of the present invention also provide a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the method described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0059] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A data acquisition method for visual management of underwater pressure-resistant battery packs, characterized in that, The method includes: Step 1: The remotely operated vehicle (ROV) descends to the pressure chamber of the battery pack and locates the wet-plug electrical connector; real-time images of the wet-plug electrical connector are obtained through the visual recognition component of the ROV, and the blockage status of the connector interface is identified based on the real-time images; Step 2: Control the cleaning device of the remotely operated vehicle to clean the connector interface with high-pressure water jet to obtain the cleaned connector interface; acquire the cleaned image through the visual recognition component, and extract multiple spatial coordinate points on the surface of the connector interface from the cleaned image; Step 3: Based on multiple spatial coordinate points, determine the central conductive contact area, the left edge sealing area, and the right edge sealing area on the connector interface, and determine the planar spatial equation through the three areas; Step 4: Based on the plane space equation, calculate the average distance between multiple spatial coordinate points and the plane space equation as a surface deviation metric, and calculate the rate of change of the normal vector of multiple spatial coordinate points; Step 5: Combine surface deviation measurement and normal vector change rate to calculate the overall deformation index of the connector interface, and generate deformation compensation coefficient based on the overall deformation index. Step 6: Using the deformation compensation coefficient and docking guide structure, control the docking device to establish a physical connection with the wet plug-in electrical connector to read the complete historical operating data of the battery pack from the battery management unit and transmit the historical operating data to the shore-based control center for visual diagnosis and evaluation of battery health status.

2. The data acquisition method for visual management of underwater pressure-resistant battery packs according to claim 1, characterized in that, Before step 1, a wet-plug electrical connector is installed on the outer shell of the battery pack pressure chamber. The wet-plug electrical connector integrates an openable and closable protective cover and a docking guide structure to close the connector interface in the non-docked state.

3. The data acquisition method and system for visual management of underwater pressure-resistant battery packs according to claim 2, characterized in that, Step 1 includes: The remotely operated vehicle moves along the surface of the battery pack pressure chamber shell to the wet plug-in electrical connector installation area, and acquires front and side images of the connector interface through a visual recognition component; Based on the marine organism coverage area on the front image recognition connector interface surface, and based on the sediment thickness in the gap of the side image recognition connector interface; The ratio of marine organism coverage area to the total area of ​​the connector interface is used as the first blockage indicator, and the ratio of sediment thickness to the standard gap of the interface is used as the second blockage indicator. When the first blockage indicator exceeds the preset threshold or the second blockage indicator exceeds the preset threshold, the connector interface is determined to be in a blocked state.

4. The data acquisition method for visual management of underwater pressure-resistant battery packs according to claim 3, characterized in that, Step 2 includes: The high-pressure water jet cleaning device is controlled to perform multiple cycles of cleaning on the connector interface at a preset pressure and preset spray angle to obtain a cleaned connector interface. The system acquires cleaned connector interface images using a visual recognition component, extracts two-dimensional image coordinates of each pixel on the connector interface surface from the cleaned connector interface images, and calculates the depth information of each pixel based on the two-dimensional image coordinates of the corresponding pixels in multiple connector interface images acquired from different viewpoints. By fusing the two-dimensional image coordinates and depth information of each pixel, multiple spatial coordinate points are generated on the surface of the connector interface. Some of these spatial coordinate points are located in the central conductive contact area, some in the left edge sealing area, and some in the right edge sealing area.

5. The data acquisition method for visual management of underwater pressure-resistant battery packs according to claim 4, characterized in that, Step 3 includes: Based on the three-dimensional coordinate distribution of multiple spatial coordinate points, the multiple spatial coordinate points located in the geometric center area of ​​the connector interface are divided into the central conductive contact area, the multiple spatial coordinate points located in the left edge area of ​​the connector interface are divided into the left edge sealing area, and the multiple spatial coordinate points located in the right edge area of ​​the connector interface are divided into the right edge sealing area. The center position coordinates of the central conductive contact area are calculated based on multiple spatial coordinate points of the central conductive contact area; the left edge position coordinates of the left edge sealing area are calculated based on multiple spatial coordinate points of the left edge sealing area; and the right edge position coordinates of the right edge sealing area are calculated based on multiple spatial coordinate points of the right edge sealing area. The planar spatial equation is determined based on the coordinates of the center position, the left edge position, and the right edge position.

6. The data acquisition method for visual management of underwater pressure-resistant battery packs according to claim 5, characterized in that, Step 4 includes: Based on the planar space equation, the perpendicular distance from each spatial coordinate point to the planar space equation is calculated, and the arithmetic mean of the perpendicular distances of all spatial coordinate points is used as the surface deviation measure. A triangular mesh surface is constructed based on multiple spatial coordinate points. The direction of the normal vector of each triangular face in the triangular mesh surface is calculated, and the standard deviation of the angle change between the directions of the normal vectors of all adjacent triangular faces is used as the rate of change of the normal vector.

7. The data acquisition method for visual management of underwater pressure-resistant battery packs according to claim 6, characterized in that, Step 5 includes: The surface deviation measure and the rate of change of the normal vector are linearly weighted and combined according to preset weighting coefficients to calculate the overall deformation index of the connector interface. The corresponding deformation compensation coefficient is obtained from the overall deformation index through a predefined mapping table.

8. The data acquisition method for visual management of underwater pressure-resistant battery packs according to claim 7, characterized in that, Step 6 includes: Adjust the position parameters of the docking device according to the deformation compensation coefficient so that the docking device and the docking guide structure of the wet plug-in electrical connector can be matched and docked to establish a stable physical connection. By activating a high-speed data communication link through the established physical connection, the complete historical operating data of the battery pack can be read from the battery management unit, including cell voltage curves, temperature curves, and charge-discharge cycle records. Complete historical operational data is transmitted to the shore-based control center via the underwater communication module of the remotely operated vehicle, enabling visualized diagnosis of battery health status and safety risk assessment of performance degradation.

9. A data acquisition system for visual management of underwater pressure-resistant battery packs, wherein the system implements the method as described in any one of claims 1 to 8, characterized in that, include: The positioning and visual recognition module is used to locate the wet-plug electrical connector when the remotely operated vehicle descends to the pressure chamber of the battery pack. The visual recognition component of the remotely operated vehicle acquires real-time images of wet-plug electrical connectors and identifies the blockage status of the connector interface based on the real-time images. The adaptive cleaning module is used to control the cleaning device of the remotely operated vehicle to clean the connector interface with high-pressure water jets, resulting in a cleaned connector interface. The cleaning image is obtained through a visual recognition component, and multiple spatial coordinate points of the connector interface surface are extracted from the cleaning image. The geometric partitioning and plane fitting module is used to determine the central conductive contact area, left edge sealing area and right edge sealing area on the connector interface based on multiple spatial coordinate points, and to determine the plane space equation through the three areas; The deformation analysis module is used to calculate the average distance between multiple spatial coordinate points and the plane spatial equation as a surface deviation metric, and to calculate the rate of change of the normal vector of multiple spatial coordinate points, based on the plane spatial equation. The deformation compensation coefficient generation module is used to calculate the overall deformation index of the connector interface by combining the surface deviation metric and the normal vector change rate, and to generate the deformation compensation coefficient based on the overall deformation index. The adaptive docking and data feedback module is used to control the docking device to establish a physical connection with the wet plug-in electrical connector by utilizing the deformation compensation coefficient and docking guide structure. This allows the module to read complete historical operating data of the battery pack from the battery management unit and transmit the historical operating data to the shore-based control center for visual diagnosis and evaluation of battery health status.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program that, when executed by a processor, implements the method as described in any one of claims 1 to 8.

Images