Ship bottom dirt removing system and removing process thereof

Abstract

The invention relates to a ship bottom dirt removal system and a ship bottom dirt removal process, and belongs to the technical field of ship bottom dirt removal. The ship bottom dirt removal system comprises an underwater robot platform, four multi-degree-of-freedom mechanical arms, a high-pressure water jet nozzle, a sprinkler type coating nozzle, an underwater motor and an ultrasonic generator; the four multi-degree-of-freedom mechanical arms are all installed on the underwater robot platform, the underwater robot platform is provided with an underwater holder, the underwater holder is provided with a high-definition underwater camera, and the underwater robot platform is provided with a water jet pump, a liquid storage water tank and a conveying pump. According to the system, the underwater robot with the path designed through an algorithm can achieve dirt recognition, multi-mode collaborative removal (high-pressure water jet, bio-enzyme and ultrasonic wave) and self-repairing coating spraying, the system has high efficiency, environment friendliness and long-acting antifouling capacity, and the ship maintenance cost and the marine pollution risk can be remarkably reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of ship bottom fouling removal, in particular to a ship bottom fouling removal system and a removal process thereof. Background Art

[0002] Existing ship bottom fouling removal systems mainly include mechanical removal, chemical removal, biological control and dry dock cleaning methods. Among them, mechanical removal methods mainly include: 1. Manual scraping: manual cleaning using tools such as scrapers and wire brushes, suitable for small areas or localized dirt; 2. High-pressure water jetting: using high-pressure water guns (pressure is usually 2000-3000psi) to flush the bottom of the ship, which can quickly remove soft dirt (such as algae) and some hard organisms; 3. Robotic / automated equipment: underwater robots or dock cleaning equipment, suitable for large ships or hard-to-reach areas. Failure to remove ship bottom dirt will increase fuel consumption (up to 40%), accelerate hull corrosion, and affect navigation stability.

[0003] Existing ship bottom fouling removal systems have the following main problems: 1. Mechanical scraping can easily damage the anti-rust layer, and using only high-pressure water jets has limited effectiveness against hard organisms (such as barnacles); 2. Chemical cleaning agents may contain toxic ingredients and do not comply with international environmental regulations; 3. Traditional antifouling coatings require regular recoating, and fouling can still easily accumulate on ships moored for long periods of time; Furthermore, existing robotic cleaning equipment lacks the ability to adapt to fouling types and hull curvature. To address this, a ship bottom fouling removal system and process are proposed. This system achieves multimodal collaborative cleaning by combining high-pressure water jets, bio-enzymatic decomposition, and ultrasonic loosening. The system dynamically adjusts the removal strategy for different fouling types, optimizes the path and cleaning parameters in real time through an algorithm, adapts to complex hull curvature, and reduces manual intervention. The bio-enzymatic cleaning agent is biodegradable and complies with IMO environmental standards. A self-healing coating fills scratches with nanomaterials, extending the antifouling effectiveness to 5-7 years. The system is suitable not only for ocean-going cargo ships and yachts, but also for offshore drilling platforms. It eliminates the need for dry docking and supports underwater in-situ operations. Summary of the Invention

[0004] The present invention provides a ship bottom fouling removal system and a removal process thereof, which solve the problems raised by the above-mentioned existing background technology, integrate dynamic monitoring, multi-modal cleaning and long-term anti-fouling functions, reduce maintenance costs and extend the life of the ship.

[0005] The present invention solves the above-mentioned technical problems with the following solution: a ship bottom fouling removal system, comprising an underwater robot platform, a multi-degree-of-freedom robotic arm, a high-pressure water jet nozzle, a shower-type coating nozzle, an underwater motor, and an ultrasonic generator. The multi-degree-of-freedom robotic arms are provided with four, each of which is mounted on the underwater robot platform. The underwater robot platform is equipped with an underwater pan / tilt platform, which is equipped with a high-definition underwater camera. The underwater robot platform is equipped with a water jet pump, a liquid storage tank, and a delivery pump. A cleaning scraper is mounted on the driving end of the underwater motor via bolts. The high-pressure water jet nozzle, the shower-type coating nozzle, the underwater motor, and the ultrasonic generator are respectively mounted on the driving end of the multi-degree-of-freedom robotic arm.

[0006] The cleaning process includes the following steps:

[0007] S1: The underwater robot platform scans the bottom of the ship using a high-definition underwater camera equipped with an underwater gimbal, collecting high-definition images of bottom fouling, including algae, barnacles, and shellfish. These images are annotated with their categories and location bounding boxes to construct a dataset. This dataset is then fed into the fouling classification model. The model uses a lightweight convolutional neural network in its backbone network to balance computational efficiency and recognition accuracy. The model then performs a classification branch, outputs the fouling category, and then performs a semantic segmentation branch to generate a pixel-level fouling distribution mask. Finally, transfer learning is performed, fine-tuning the pre-trained ImageNet model using the bottom fouling dataset.

[0008] S2: Path planning algorithm design: 1. Dirt distribution modeling: converting semantic segmentation results into a two-dimensional grid map, with each grid labeled with pollution level and dirt type. Pollution level: based on dirt density, such as algae coverage, is divided into low, medium, and high. Dirt type: corresponds to the removal strategy, such as the area that needs to be covered by high-pressure water jets and the need for precise positioning of barnacles. 2. Optimization objectives and constraints: objective function: minimizing total operation time (movement time + removal time); constraints: robot kinematic limitations (turning radius, speed), removal efficiency differences (e.g., single-point barnacle treatment time > algae area coverage), energy consumption upper limit (battery capacity limit); 3. Hybrid path planning strategy: global planning, generating preliminary paths, and prioritizing traversal of high-pollution areas.

[0009] S3: Algorithm implementation process: 1. Image input and real-time processing: the camera on the robot captures 10 frames of images per second, which are input into the classification model after preprocessing, and the inference time is less than 50ms / frame; 2. Dirt map update: through SLAM technology, the recognition results are mapped to the 3D surface model of the hull, and the pollution grid is dynamically updated; 3. Path generation and execution: Decision tree logic: If a barnacle gathering area is detected (single area area > 0.5m 2 ), trigger the "high-frequency water jet + ultrasonic" mode;

[0010] S4: The water jet pump can draw seawater in and spray it out through the high-pressure water jet nozzle of the multi-degree-of-freedom robotic arm for high-pressure water jet. First, the barnacle gathering area on the bottom of the ship is cleaned through the high-pressure water jet nozzle of the multi-degree-of-freedom robotic arm and the ultrasonic generator. The delivery pump can spray the chitinase in the liquid storage tank through the shower-type coating nozzle to the bottom of the ship, apply the chitinase to decompose the barnacle mucus, start the underwater motor to rotate the cleaning scraper, and use the shower-type coating nozzle and the cleaning scraper combination to perform a secondary cleaning on the cleaned area and spray a self-repairing coating. Finally, the nano-silicon-based material and zinc oxide particles are replaced in the liquid storage tank and then sprayed through the shower-type coating nozzle to the bottom of the ship. The self-repairing coating releases zinc oxide particles when it meets seawater to inhibit the attachment of algae spores.

[0011] On the basis of the above technical solution, the present invention can also be improved as follows.

[0012] Furthermore, in step S1, the underwater image is enhanced by using adaptive histogram equalization to compensate for underwater light attenuation, a deep learning denoising model is used to eliminate image blur caused by turbid water, and a color correction algorithm is used to restore color distortion caused by water absorption.

[0013] Furthermore, in step S2, local optimization: reinforcement learning (RL) dynamic adjustment: training the DQN (Deep Q-Network) model, adjusting the path weight in real time according to the remaining power and dirt distribution, bionic algorithm assistance: introducing the ant colony algorithm to optimize the access order of multiple target points and reduce repeated paths.

[0014] Furthermore, in step S3, the path is covered in a spiral manner, and the algae dispersion area is traversed in a "Z-shaped" manner, combined with wide-area cleaning using a high-pressure water jet, and real-time obstacle avoidance: protrusions (such as rivets) are detected through lidar, and the posture of the robotic arm is adjusted to avoid collisions.

[0015] The beneficial effects of the present invention are as follows: the present invention provides a ship bottom fouling removal system and a removal process thereof, which have the following advantages:

[0016] 1. Achieve multi-modal collaborative cleaning, combining high-pressure water jets, bio-enzyme decomposition, and ultrasonic loosening, dynamically adjust the cleaning strategy for different fouling types, optimize the path and cleaning parameters in real time through algorithms, adapt to complex hull surfaces, and reduce manual intervention;

[0017] 2. The bio-enzyme cleaning agent is biodegradable and meets IMO environmental standards. The self-healing coating uses nanomaterials to fill scratches, extending the anti-fouling period to 5-7 years. It is suitable not only for ocean-going cargo ships and yachts but also for offshore drilling platforms. It does not require dry docking and supports underwater in-situ operations.

[0018] 3. Underwater robots with algorithm-designed paths can realize dirt identification, multimodal collaborative removal (high-pressure water jet + biological enzyme + ultrasound), and self-healing coating spraying. The system is highly efficient, environmentally friendly, and has long-term anti-fouling capabilities, which can significantly reduce ship maintenance costs and marine pollution risks.

[0019] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention and to implement it according to the contents of the description, the following preferred embodiments of the present invention are described in detail with reference to the accompanying drawings. The specific implementation methods of the present invention are given in detail by the following embodiments and the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0021] Figure 1 A process flow chart of a ship bottom fouling removal system and its removal process provided in one embodiment of the present invention;

[0022] Figure 2 A schematic structural diagram of a ship bottom fouling removal system and its removal process provided by one embodiment of the present invention;

[0023] Figure 3 A front view of a ship bottom fouling removal system and its removal process provided in one embodiment of the present invention.

[0024] In the accompanying drawings, the components represented by the reference numerals are as follows:

[0025] 1. Underwater robot platform; 2. Multi-degree-of-freedom robotic arm; 3. Underwater pan / tilt; 4. High-definition underwater camera; 5. Water jet pump; 6. Liquid storage tank; 7. Delivery pump; 8. High-pressure water jet nozzle; 9. Shower-type coating nozzle; 10. Underwater motor; 11. Ultrasonic generator; 12. Cleaning scraper. DETAILED DESCRIPTION

[0026] The following is combined with Figure 1-3 The principles and features of the present invention are described, and the examples given are only for the purpose of explaining the present invention and are not intended to limit the scope of the present invention. The following paragraphs describe the present invention in more detail by way of example with reference to the accompanying drawings. The advantages and features of the present invention will become more apparent from the following description and claims. It should be noted that the drawings are in a very simplified form and are not in exact proportions, and are only used for the purpose of conveniently and clearly assisting in illustrating the embodiments of the present invention.

[0027] It should be noted that when a component is referred to as being "fixed to" another component, it may be directly on the other component or there may also be a central component. When a component is considered to be "connected to" another component, it may be directly connected to the other component or there may also be a central component. When a component is considered to be "set on" another component, it may be directly set on the other component or there may also be a central component. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only.

[0028] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this invention pertains. The terms used in this specification of the present invention are for the purpose of describing specific embodiments only and are not intended to limit the present invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0029] like Figure 1-3 As shown, the present invention provides a ship bottom dirt removal system, including an underwater robot platform 1, a multi-degree-of-freedom robotic arm 2, a high-pressure water jet nozzle 8, a shower-type coating nozzle 9, an underwater motor 10, and an ultrasonic generator 11. Four multi-degree-of-freedom robotic arms 2 are provided, and the four multi-degree-of-freedom robotic arms 2 are all installed on the underwater robot platform 1. The underwater robot platform 1 is equipped with an underwater pan-tilt platform 3, and the underwater pan-tilt platform 3 is equipped with a high-definition underwater camera 4. The underwater robot platform 1 is equipped with a water jet pump 5, a liquid storage tank 6, and a delivery pump 7. The driving end of the underwater motor 10 is equipped with a cleaning scraper 12 through bolts, and the high-pressure water jet nozzle 8, the shower-type coating nozzle 9, the underwater motor 10, and the ultrasonic generator 11 are respectively installed on the driving end of the multi-degree-of-freedom robotic arm 2.

[0030] The specific working principle and method of use of the present invention are as follows:

[0031] S1: The underwater robot platform 1 scans the bottom of the ship using the high-definition underwater camera 4 of the underwater gimbal 3, collecting high-definition images of bottom fouling, including algae, barnacles, and shellfish. The images are annotated with their categories and location bounding boxes to construct a dataset. The dataset is then passed to the fouling classification model. The fouling classification model uses a lightweight convolutional neural network in the backbone network to balance computational efficiency and recognition accuracy. The model then performs a classification branch, outputs the fouling category, performs a semantic segmentation branch, and generates a pixel-level fouling distribution mask. Finally, transfer learning is performed. Based on the pre-trained ImageNet model, the model is fine-tuned using the bottom fouling dataset, and underwater images are enhanced. Adaptive histogram equalization is used to compensate for underwater light attenuation. A deep learning denoising model is used to eliminate image blur caused by turbid water. A color correction algorithm is used to restore color distortion caused by water absorption.

[0032] S2: Path planning algorithm design: 1. Dirt distribution modeling, converting the semantic segmentation results into a two-dimensional grid map, where each grid is marked with pollution level and dirt type. Pollution level: divided into low, medium and high according to dirt density, such as algae coverage rate. Dirt type: corresponding to the removal strategy, such as the area that high-pressure water jets need to cover, and barnacles need to be accurately positioned; 2. Optimization objectives and constraints: objective function, minimizing the total operation time, moving time + removal time, constraints, robot kinematics restrictions on turning radius and speed, removal efficiency differences such as single-point barnacle processing time > algae area coverage, energy consumption upper limit battery capacity limit; 3. Hybrid path planning strategy: global planning, generating preliminary paths, prioritizing traversal of high-pollution areas, local optimization: reinforcement learning RL dynamic adjustment: training DQN Deep Q-Network model, adjusting path weights in real time according to remaining power and dirt distribution, bionic algorithm assistance: introducing ant colony algorithm to optimize the access order of multiple target points and reduce repeated paths;

[0033] S3: Algorithm implementation process: 1. Image input and real-time processing: The camera on the robot captures 10 frames of images per second, which are input into the classification model after preprocessing, with an inference time of less than 50ms / frame; 2. Dirt map update: Using SLAM technology, the recognition results are mapped to the 3D surface model of the hull, and the pollution grid is dynamically updated; 3. Path generation and execution: Decision tree logic: If the area of ​​a single barnacle gathering area is detected to be greater than 0.5m 2 , triggering the "high-frequency water jet + ultrasonic" mode, the path adopts spiral coverage, the algae dispersion area adopts "Z-shaped" traversal, combined with high-pressure water jet wide-area cleaning, real-time obstacle avoidance: through the laser radar to detect protrusions such as rivets, adjust the robot arm posture to avoid collision;

[0034] S4: The water jet pump 5 can draw seawater into the high-pressure water jet nozzle 8 of the multi-degree-of-freedom robotic arm 2 and spray it out for high-pressure water jet. First, the barnacle gathering area on the bottom of the ship is cleaned through the high-pressure water jet nozzle 8 of the multi-degree-of-freedom robotic arm 2 and the ultrasonic generator 11. The delivery pump 7 can spray the chitinase in the liquid storage tank 6 through the shower-type coating nozzle 9 to the bottom of the ship, and apply the chitinase to decompose the barnacle mucus. The underwater motor 10 can be started to rotate the cleaning scraper 12. The combination of the shower-type coating nozzle 9 and the cleaning scraper 12 can be used to perform secondary cleaning on the cleaned area and spray a self-repairing coating. Finally, the nano-silicon-based material and zinc oxide particles are replaced in the liquid storage tank 6 and then sprayed through the shower-type coating nozzle 9 to the bottom of the ship. The self-repairing coating releases zinc oxide particles when it meets seawater to inhibit the attachment of algae spores.

[0035] Example 1: Identification and removal of mixed algae and barnacle pollution

[0036] 1. Input parameters and conditions:

[0037] Image input: Resolution: 1920 × 1080 pixels, underwater RGB-NIR multispectral image (near-infrared band range: 700-1000nm);

[0038] Lighting conditions: simulated water depth of 5 meters, visibility of 2 meters, red channel attenuation coefficient of 0.8;

[0039] Dirt distribution: Algae coverage: 40% (evenly distributed), barnacle aggregation area: 0.3m 2 (3 locations in total);

[0040] Robot parameters: moving speed: 0.5m / s, high-pressure water jet coverage width: 0.2m, robotic arm positioning accuracy ±5mm, battery capacity: 10kWh, operating power consumption: 500W (moving) + 1500W (clearing);

[0041] 2. Algorithm processing flow:

[0042] Image preprocessing:

[0043] CLAHE was applied to enhance the contrast, with a red channel gain compensation factor of 1.5;

[0044] Using the U-Net denoising model (training data: 100,000 turbid underwater images), PSNR increased to 28dB;

[0045] Dirt classification and segmentation:

[0046] Input EfficientNet-B4 model (pre-trained weights + fine-tuning dataset: 50,000 annotated images);

[0047] Output:

[0048] Algae classification confidence: 96.2%, segmentation IoU: 89.5%.

[0049] Barnacle classification confidence: 88.7%, segmentation IoU: 83.1%.

[0050] Multispectral fusion: NIR band enhances barnacle edge detection, reducing the false positive rate by 12%;

[0051] 3. Path planning:

[0052] Global path:

[0053] Priority: barnacle aggregation area (weight coefficient 1.5) > high-density algae area (weight 1.2) > medium-low density area (weight 1.0);

[0054] Initial path length: 18.7m, estimated time: 37.4 minutes;

[0055] Reinforcement learning dynamic adjustment (DQN model):

[0056] Real-time detection of remaining battery power (8.2kWh) reduces dwell time in low-density areas (from 30 seconds to 15 seconds in a single area);

[0057] Optimized path length: 14.2m, time taken: 28.4 minutes, energy consumption saved: 24%;

[0058] 4. Verification of cleaning effect:

[0059] Algae removal rate: 94.3% (high-pressure water jet pressure 2500psi, coverage rate 98%);

[0060] Barnacle removal rate: 91.5% (pulsed water jet 3000psi + 80kHz ultrasonic wave, robotic arm scraping residual rate <5%)

[0061] Example 2: Path Planning Optimization for a Complex Curved Hull (Yacht)

[0062] 1. Input parameters and conditions:

[0063] Hull model:

[0064] Surface curvature radius: 0.5-3m (local concave and convex areas);

[0065] Fouling type: Dense attachment of shellfish (size 5-15 cm), coverage 25%.

[0066] Sensor configuration:

[0067] 3D laser radar (accuracy ±2mm), SLAM positioning error <1cm.

[0068] Algorithm parameters:

[0069] Path safety distance: 10cm (to avoid collision with protrusions on the hull).

[0070] Maximum extension length of the robotic arm: 1.2m;

[0071] 2. Algorithm processing flow:

[0072] 3D Dirt Map Construction:

[0073] SLAM generates hull surface point cloud (resolution 1cm 3 ), mapping the dirt distribution to a 3D grid;

[0074] Analysis of dirt density: the area with curvature greater than 1.5 m-1 in the shellfish gathering area accounts for 70%;

[0075] Path planning:

[0076] Global planning:

[0077] The ant colony algorithm optimizes the access sequence and reduces the number of robot arm posture adjustments (from 32 times to 18 times);

[0078] Initial path length: 22.5m, time taken: 45 minutes;

[0079] Local obstacle avoidance:

[0080] Detect rivets (height 2cm) and trigger an avoidance path (detour radius 15cm);

[0081] Dynamically adjust the water jet angle (angle with the surface normal <30°) to ensure uniform cleaning pressure;

[0082] Performance comparison:

[0083] Traditional U-shaped path: took 58 minutes, and the robot arm collided twice;

[0084] This algorithm path: took 41 minutes, had zero collisions, and the surface area clearance rate increased to 88.6%;

[0085] Example 3: Extreme Environment (Low Visibility + Dynamic Addition of Dirt)

[0086] 1. Input parameters and conditions:

[0087] Environmental simulation:

[0088] Visibility: 0.5m, suspended particle concentration: 200mg / L.

[0089] Dynamically added dirt: 5% more algae coverage is added every 10 minutes of simulated navigation.

[0090] Algorithm robustness test:

[0091] Initial fouling distribution: algae 30%, barnacles 10%.

[0092] Operating time: 60 minutes, battery capacity limited to 8kWh.

[0093] 2. Algorithm processing flow:

[0094] Image enhancement and recognition:

[0095] The PSNR of turbid images increases from 18dB to 25dB (U-Net denoising + multi-spectral compensation).

[0096] Classification accuracy remained at 92% for algae and 85% for barnacles (lower than in clear water environments but meeting operational requirements).

[0097] Dynamic path replanning;

[0098] Update the dirt map every 5 minutes to detect new areas;

[0099] Reinforcement learning model response:

[0100] The weight of the newly added algae area is automatically increased to 1.3 and inserted into the current path queue;

[0101] Replanning time: 1.8 seconds, path change <15%;

[0102] 3. Result statistics:

[0103] Final removal rate: 91% of initial dirt + 87% of new dirt.

[0104] Total energy consumption: 7.8kWh (battery utilization 97.5%), no power protection interruption triggered;

[0105] It should be noted that, in this document, relational terms such as first and second are used solely to distinguish one entity or operation from another, and do not necessarily require or imply any actual relationship or order between these entities or operations. Matters not described in detail in this specification are well known to those skilled in the art.

[0106] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Any ordinary technician in this industry can smoothly implement the present invention as shown in the drawings and described above. However, any equivalent changes, modifications and evolutions made by technicians familiar with this profession without departing from the scope of the technical solution of the present invention using the technical content disclosed above are all equivalent embodiments of the present invention. At the same time, any equivalent changes, modifications and evolutions made to the above embodiments based on the essential technology of the present invention are still within the scope of protection of the technical solution of the present invention.

Claims

1. A ship bottom fouling removal system, comprising an underwater robot platform (1), a multi-degree-of-freedom robotic arm (2), a high-pressure water jet nozzle (8), a shower-type coating nozzle (9), an underwater motor (10), and an ultrasonic generator (11), characterized in that: Four multi-degree-of-freedom robotic arms (2) are provided, and the four multi-degree-of-freedom robotic arms (2) are all installed at the underwater robot platform (1); the underwater robot platform (1) is installed with an underwater pan / tilt platform (3); the underwater pan / tilt platform (3) is installed with a high-definition underwater camera (4); the underwater robot platform (1) is installed with a water jet pump (5), a liquid storage tank (6), and a delivery pump (7); a cleaning scraper (12) is installed at the driving end of the underwater motor (10) through bolts; the high-pressure water jet nozzle (8), the shower-type coating nozzle (9), the underwater motor (10), and the ultrasonic generator (11) are respectively installed at the driving end of the multi-degree-of-freedom robotic arm (2); The cleaning process includes the following steps: S1: The underwater robot platform (1) scans the bottom of the ship through the high-definition underwater camera (4) of the underwater gimbal (3), collects high-definition images of the ship bottom dirt including algae, barnacles, shellfish, etc., annotates their categories and position bounding boxes to build a dataset, and transmits the dataset to the dirt classification model. The dirt classification model uses a lightweight convolutional neural network through the backbone network to balance computational efficiency and recognition accuracy, performs classification branch, outputs dirt category, performs semantic segmentation branch, generates pixel-level dirt distribution mask, and finally transfers learning. Based on the pre-trained ImageNet model, it uses the ship bottom dirt dataset for fine-tuning; S2: Path planning algorithm design:

1. Dirt distribution modeling: Converting semantic segmentation results into a two-dimensional grid map, with each grid labeled with pollution level and dirt type. Pollution level: Based on dirt density, such as algae coverage, it is divided into low, medium, and high. Dirt type: Corresponding removal strategy, such as the area required for high-pressure water jet coverage and the need for precise positioning of barnacles.

2. Optimization objectives and constraints: Objective function: Minimize total operation time (movement time + removal time). Constraints: Robot kinematic limitations (turning radius, speed), removal efficiency differences (e.g., single-point barnacle treatment time > algae area coverage), and energy consumption upper limit (battery capacity limit).

3. Hybrid path planning strategy: global planning, generating preliminary paths, and prioritizing traversal of high-pollution areas; S3: Algorithm implementation process:

1. Image input and real-time processing. The camera on the robot captures 10 frames of images per second, which are input into the classification model after preprocessing. The inference time is less than 50ms / frame.

2. Update the dirt map. Through SLAM technology, the recognition results are mapped to the 3D surface model of the hull, and the pollution grid is dynamically updated.

3. Path generation and execution: Decision tree logic: If a barnacle gathering area is detected (single area area > 0.5m 2 ), trigger the "high-frequency water jet + ultrasonic" mode; S4: The water jet pump (5) can draw seawater into the high-pressure water jet nozzle (8) of the multi-degree-of-freedom manipulator (2) and spray it out to perform high-pressure water jet. First, the barnacle gathering area on the bottom of the ship is cleaned by the high-pressure water jet nozzle (8) of the multi-degree-of-freedom manipulator (2) and the ultrasonic generator (11). The delivery pump (7) can spray the chitinase in the liquid storage tank (6) through the shower-type coating nozzle (9) to the bottom of the ship, and apply the chitinase to decompose the barnacle mucus. The underwater motor (10) can be started to rotate the cleaning scraper (12). The cleaning area can be cleaned for the second time and a self-repairing coating can be sprayed by the combination of the shower-type coating nozzle (9) and the cleaning scraper (12). Finally, the nano-silicon-based material and zinc oxide particles are replaced in the liquid storage tank (6) and then sprayed out through the shower-type coating nozzle (9) to the bottom of the ship. After the self-repairing coating meets seawater, the zinc oxide particles are released to inhibit the attachment of algae spores.

2. A ship bottom fouling removal system according to claim 1, characterized in that: In step S1, the underwater image is enhanced by using adaptive histogram equalization to compensate for underwater light attenuation, a deep learning denoising model is used to eliminate image blur caused by turbid water, and a color correction algorithm is used to restore color distortion caused by water absorption.

3. A ship bottom fouling removal system according to claim 1, characterized in that: In step S2, local optimization: reinforcement learning (RL) dynamic adjustment: training the DQN (Deep Q-Network) model, adjusting the path weight in real time according to the remaining power and dirt distribution, bionic algorithm assistance: introducing the ant colony algorithm to optimize the access order of multiple target points and reduce repeated paths.

4. A ship bottom fouling removal system according to claim 1, characterized in that: In step S3, the path is covered in a spiral pattern, and the algae dispersion area is traversed in a "Z-shaped" pattern, combined with wide-area cleaning using a high-pressure water jet, and real-time obstacle avoidance: protrusions (such as rivets) are detected through lidar, and the robot arm's posture is adjusted to avoid collisions.

Images