Deep-sea unmanned vehicle and deep-sea terrain adaptive cruise path planning method thereof

Abstract

This invention provides a deep-sea unmanned surface vessel (USV) and its deep-sea terrain adaptive cruise path planning method, comprising a hull, with a lighting component, a monitoring component, a main control component, and a photovoltaic component sequentially arranged on the upper surface of the hull. The main control component includes a working trough, a main processor located within the working trough, and a radar transceiver component. Both sides of the hull are provided with load-bearing bags, and the bottom of the hull is provided with a drive component located between two load-bearing bags. The bottom of the hull is also provided with a sonar component located between two load-bearing bags. The advantages of this invention are: by using lighting and sonar components, it facilitates the detection of the area around the deep-sea USV; the main processor analyzes and processes the data; real-time path planning is performed using sonar data and machine learning algorithms to select the optimal route to avoid collisions with obstacles; and then the deep-sea USV is easily controlled for automatic adjustment and operation, facilitating adaptive cruise path planning.

Description

Technical Field

[0001] This invention relates to the field of deep-sea unmanned surface vessel (USV) technology, and in particular to a deep-sea USV and its deep-sea terrain adaptive cruise path planning method. Background Technology

[0002] Deep-sea unmanned surface vessels (USVs) are core equipment for exploring the deep sea, and their cruising path planning directly determines operational efficiency and safety. However, the deep-sea environment is extremely complex, with varied terrain and other issues, making traditional path planning methods difficult to adapt. Traditional methods rely heavily on preset maps and fixed algorithms, requiring the acquisition of high-precision terrain data in advance. Faced with real-time changes such as seafloor landslides and sediment migration, they cannot dynamically adjust the path, which can easily lead to collisions or detours.

[0003] Meanwhile, deep-sea unmanned surface vessels (USVs) have limited computing power, and traditional deep learning models, with their large number of parameters and long inference time, cannot meet real-time planning requirements. Furthermore, in multi-vessel cooperative cruising, the conflict between data transmission and privacy protection becomes prominent, and traditional centralized planning struggles to balance efficiency and safety. Against this backdrop, there is an urgent need to leverage machine learning technology to overcome bottlenecks in terrain perception, real-time planning, and collaborative adaptation, enabling terrain-adaptive cruising for deep-sea USVs.

[0004] However, with the current level of technology, the terrain along the cruise path cannot be detected in real time during the use of deep-sea unmanned surface vessels (USVs), which prevents the USVs from making adaptive adjustments, which is not conducive to subsequent operations and reduces the operational efficiency of USVs. Summary of the Invention

[0005] In view of the shortcomings of the prior art described above, the purpose of this invention is to provide a deep-sea unmanned surface vessel and its deep-sea terrain adaptive cruise path planning method, which solves the problem that the existing cruise path planning methods cannot detect the terrain in the cruise path in real time, thus making the unmanned surface vessel unable to adaptively adjust and dynamically adjust the path, which easily leads to collisions or detours.

[0006] To achieve the above and other related objectives, the present invention provides the following technical solution:

[0007] A deep-sea unmanned surface vessel (USV) includes a hull. Along the length of the hull, a lighting assembly, a monitoring assembly, a main control assembly, and a photovoltaic assembly are sequentially arranged on the upper surface of the hull. The main control assembly includes a working slot disposed on the upper surface of the hull between the monitoring assembly and the photovoltaic assembly, a main processor located within the working slot, and a radar transceiver assembly. Two symmetrically arranged weight bags are provided on each of the two sides of the hull. A drive assembly is located between the two weight bags at the bottom of the hull. A sonar assembly is located between the two weight bags at the bottom of the side of the hull furthest from the drive assembly.

[0008] In one embodiment of the present invention, the lighting assembly includes a rectangular plate mounted on the upper surface of the housing, a lighting lamp located on the side of the rectangular plate, and a plurality of work lamps. The lighting lamp and the plurality of work lamps are mounted sequentially and evenly spaced along the circumference of the rectangular plate on the upper surface of the housing.

[0009] In one embodiment of the present invention, the monitoring component includes a rotatable support base (13) mounted on the upper surface of the housing and a monitor mounted on the support base. The upper surface of the housing is also provided with a first processor located on the side of the support base and connected to the main processor. The housing is also provided with a sensor component for collecting full-dimensional hull status data about the deep-sea unmanned surface vessel and connected to the first processor. The first processor is also connected to the lighting component, monitoring component, radar transceiver component, photovoltaic component and sonar component.

[0010] In one embodiment of the present invention, the radar transceiver assembly includes a lidar transmitter mounted in the working slot via a support column and located between the main processor and the photovoltaic module, and a plurality of lidar receivers mounted circumferentially along the working slot on the upper surface of the housing. The support column is provided with a second processor connected to the main processor.

[0011] In one embodiment of the present invention, the photovoltaic module includes a photovoltaic panel mounted on the upper surface of the housing and a voltage regulator located on the side of the photovoltaic panel, the voltage regulator being connected to the photovoltaic panel and a storage tank located inside the housing.

[0012] In one embodiment of the present invention, a pump is provided inside the housing corresponding to the load bag and connected to the load bag through an internal pipe. The pump is also connected to a transport pipe. The end of the transport pipe away from the pump extends to the outside of the housing, and a connection port is provided on the end of the transport pipe away from the pump. A filter is threadedly connected to the connection port.

[0013] A deep-sea terrain adaptive cruise path planning method, based on the aforementioned deep-sea unmanned surface vessel and applied to a main processor, the method includes the following steps: receiving multi-source environmental data collected by a monitoring component, a radar transceiver component, and a sonar component and processed by data cleaning and preprocessing sent by a first processor; and determining a standardized underwater three-dimensional terrain map and obstacle coordinate model based on the multi-source environmental data after data cleaning and preprocessing.

[0014] The cleaned and preprocessed multi-source environmental data is input into the obstacle recognition and risk classification model. The obstacle recognition and risk classification model, combined with the standardized underwater 3D topographic map and obstacle coordinate model, outputs a real-time path plan for the deep-sea unmanned surface vessel. The real-time path plan for the deep-sea unmanned surface vessel is then sent to the second processor for verification.

[0015] In one embodiment of the present invention, the step of outputting real-time path planning for the deep-sea unmanned surface vessel by combining the obstacle identification and risk classification model with the standardized underwater environment three-dimensional terrain map and obstacle coordinate model includes: extracting features from the standardized sonar data in the multi-source environmental data, and identifying obstacle types, determining obstacle threat levels, and calculating obstacle coverage and no-passage zones based on the extracted features; and obtaining real-time path planning for the deep-sea unmanned surface vessel based on the identified obstacle types, determined obstacle threat levels, and calculated obstacle coverage and no-passage zones.

[0016] In one embodiment of the present invention, the real-time path planning for the deep-sea unmanned surface vessel (USV) based on the identified obstacle type, the determined obstacle threat level, and the calculated obstacle coverage area and restricted area includes: generating an initial baseline route based on a preset cruise endpoint, combined with the identified obstacle type, the determined obstacle threat level, and the calculated obstacle coverage area and restricted area; dynamically generating multiple alternative detour routes based on the real-time detected obstacle distribution and the vessel's navigation status; simultaneously calculating the core indicators of each alternative detour route; and comprehensively evaluating all alternative detour routes through a weighted scoring mechanism to obtain the optimal alternative detour route based on the evaluation results.

[0017] In one embodiment of the present invention, the method further includes the following steps: receiving preprocessed, real-time, multi-dimensional hull state data collected by sensors from a first processor; inputting the preprocessed multi-dimensional hull state data into an adaptive buoyancy adjustment model, wherein the adaptive buoyancy adjustment model performs intelligent algorithm iterative calculations on the multi-dimensional hull state data to obtain the optimal stable draft under the current sea state, and simultaneously determines the flow rate, duration, and speed of the load bag water supply; and sending the optimal stable draft under the current sea state and the flow rate, duration, and speed of the load bag water supply to a second processor for verification.

[0018] As described above, the deep-sea unmanned surface vessel and its deep-sea terrain adaptive cruise path planning method of the present invention have the following beneficial effects:

[0019] 1. This invention facilitates the detection of the surroundings of the deep-sea unmanned surface vessel by using lighting and sonar components, analyzes and processes the data through a main processor, and uses sonar data and machine learning algorithms for real-time path planning to select the best route to avoid collisions with obstacles. Then, it facilitates the automatic adjustment and operation of the deep-sea unmanned surface vessel, enabling the deep-sea unmanned surface vessel to adaptively cruise along a path.

[0020] 2. The machine learning model in this invention can analyze the current depth, buoyancy, acceleration and other data of the unmanned surface vessel in real time, and automatically adjust the draft. The machine learning model can adaptively adjust the draft to avoid unstable buoyancy of the hull or rapid rise or sink. The draft of the deep-sea unmanned surface vessel can be adjusted by transporting water into the load bag or draining water.

[0021] 3. The present invention facilitates the analysis of data detected by the deep-sea unmanned surface vessel by setting up a first processor, a main processor and a second processor, thereby enabling judgment and facilitating the setting of a cruise path for the deep-sea unmanned surface vessel, so that it can complete its mission in the complex deep-sea environment.

[0022] 4. The present invention uses photovoltaic modules to convert light energy for use when the deep-sea unmanned surface vessel is operating on the sea surface, thereby providing power resources for the equipment. When the power inside the deep-sea unmanned surface vessel is insufficient, the photovoltaic modules supply power to the energy storage box, which facilitates the subsequent operation and use of the deep-sea unmanned surface vessel. Attached Figure Description

[0023] Figure 1 The image shown is a left-side perspective view of the deep-sea unmanned surface vessel according to the first embodiment of the present invention.

[0024] Figure 2 The image shown is a right-side perspective view of the deep-sea unmanned surface vessel according to the first embodiment of the present invention.

[0025] Figure 3 The image shown is a frontal, bottom-view perspective view of the deep-sea unmanned surface vessel according to the first embodiment of the present invention.

[0026] Figure 4 This is a perspective view showing the connection between the pump and the load bag in the deep-sea unmanned surface vessel according to the first embodiment of the present invention.

[0027] Figure 5 This is a perspective view showing the connection between the connection port and the filter in the deep-sea unmanned surface vessel according to the first embodiment of the present invention;

[0028] Figure 6 The flowchart shown is a deep-sea terrain adaptive cruise path planning method according to the second embodiment of the present invention.

[0029] Figure 7 The flowchart shown is a deep-sea terrain adaptive cruise path planning method according to the third embodiment of the present invention.

[0030] Component designation explanation

[0031] 1. Housing; 2. Work light; 3. First processor; 4. Monitor; 5. Main processor; 6. LiDAR transmitter; 7. LiDAR receiver; 8. Photovoltaic panel; 9. Support column; 10. Load bag; 11. Rectangular plate; 12. Lighting lamp; 13. Support base; 14. Second processor; 15. Voltage regulator; 16. Drive assembly; 17. Connection port; 18. Filter; 19. Sonar assembly; 20. Internal pipe; 21. Pump; 22. Transport pipe; 23. Sealing gasket; 24. Threaded port; 25. Work trough. Detailed Implementation

[0032] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. It should be noted that, unless otherwise specified, the following embodiments and features described herein can be combined with each other.

[0033] The first embodiment of the present invention relates to a deep-sea unmanned surface vessel; please refer to [link / reference]. Figures 1 to 5 The upper surface of the shell 1 is provided with a lighting component, a monitoring component, a main control component, and a photovoltaic component along the length of the shell 1. The lighting component includes a rectangular plate 11 installed on the upper surface of the shell 1, a lighting lamp 12 located on the side of the rectangular plate 11, and several work lights 2. In this embodiment, there are two work lights 2. The lighting lamp 12 and the two work lights 2 are installed at equal intervals along the circumference of the rectangular plate 11 on the upper surface of the shell 1. It should be noted that the rectangular plate 11 is designed to ensure the sealing of the interior of the unmanned surface vessel and facilitate subsequent maintenance operations. The lighting lamp 12 is used to provide illumination for the front of the deep-sea unmanned surface vessel, which is convenient for subsequent operations and ensures a certain distance of visibility in front of the deep-sea unmanned surface vessel. The work lights 2 operate during operation to alert the operator.

[0034] exist Figure 1 and Figure 2In this system, the monitoring component includes a rotatable support base 13 mounted on the upper surface of the hull 1 and a monitor 4 mounted on the support base 13. The upper surface of the hull 1 also has a first processor 3 located on the side of the support base 13 and connected to the main processor 5 described below. It should be noted that the support base 13 facilitates the rotation of the monitor 4, allowing for appropriate adjustment of the monitor 4's operating angle to provide visual data such as images or videos, assisting the unmanned surface vessel in environmental perception, object recognition, and task execution. The first processor 3 facilitates data processing and use. The hull 1 also has a sensor component for collecting full-dimensional hull status data about the deep-sea unmanned surface vessel and connected to the first processor 3. The first processor 3 is also connected to a lighting component, a monitoring component, a radar transceiver component, a photovoltaic component, and a sonar component 19.

[0035] exist Figure 1 and Figure 2 The main control component includes a working trough 25 located on the upper surface of the hull 1 between the monitoring component and the photovoltaic component, a main processor 5 located within the working trough 25, and a radar transceiver component. The main processor 5 facilitates the analysis and processing of data collected by the monitoring component on the upper part of the deep-sea unmanned surface vessel (USV), and then controls the operation of the USV. The radar transceiver component includes a lidar transmitter 6 mounted on the working trough 25 via a support column 9, located between the main processor 5 and the photovoltaic component, and several lidar receivers 7 mounted circumferentially on the upper surface of the hull 1 along the working trough 25. A second processor 14 connected to the main processor 5 is mounted on the support column 9. The support column 9 facilitates the position of the lidar transmitter 6, enabling it to transmit signals. The lidar receivers 7 facilitate signal reception, and the multiple receivers 7 ensure normal signal transmission from the USV. The second processor 14 controls the transmitter, facilitating subsequent operations.

[0036] exist Figure 2 In this system, the photovoltaic module includes a photovoltaic panel 8 mounted on the upper surface of the housing 1 and a voltage regulator 15 located on the side of the photovoltaic panel 8. The voltage regulator 15 is connected to the photovoltaic panel 8 and the energy storage box located inside the housing 1. It should be noted that the photovoltaic module facilitates the auxiliary power supply to the energy storage box inside the equipment. When the unmanned surface vessel is operating in surface water or shallow water, the photovoltaic module can extend the working time of the unmanned surface vessel by charging the battery with solar energy, ensuring the long-term autonomy of the unmanned surface vessel. The voltage regulator 15 facilitates voltage stabilization, ensures the stability of power resources, and facilitates subsequent power storage and use.

[0037] exist Figures 1 to 4In the design, two symmetrically arranged weight bags 10 are provided on both sides of the hull 1. Inside the hull 1, corresponding to the weight bags 10, there is a pump 21 connected to the weight bags 10 via an inner pipe 20. The pump 21 is also connected to a transport pipe 22. The end of the transport pipe 22 away from the pump 21 extends to the outside of the hull 1, and a connection port 17 is provided on the end of the transport pipe 22 away from the pump 21. It should be noted that during the operation of the deep-sea unmanned surface vessel, the draft can be adaptively adjusted according to changes in the marine environment, such as water flow, buoyancy changes, and temperature. The pump 21 facilitates the extraction of seawater, and the transport pipe 22 connects the connection port 17 and the pump 21, facilitating the pump 21's water extraction operation. Water is transported to the inside of the weight bags 10 through the inner pipe 20, which increases the weight of the deep-sea unmanned surface vessel, thereby increasing the draft. When it is necessary to reduce the draft, the pump 21 extracts and discharges the water from the weight bags 10, completing the buoyancy and facilitating subsequent deep-sea unmanned surface vessel cruising.

[0038] exist Figure 5 In the design, a filter 18 is threadedly connected to the connection port 17. The filter 18 has a threaded port 24 that is threaded to the connection port 17. A sealing gasket 23 is placed between the connection port 17 and the threaded port 24. It should be noted that the connection port 17 and the threaded port 24 facilitate the installation of the filter 18, thereby ensuring the stability of the structure. The sealing gasket 23 helps to ensure a firm connection between the connection port 17 and the threaded port 24, ensuring the working position of the filter 18, and then filtering is performed by using the filter 18.

[0039] exist Figure 3 In the design, the bottom of the shell 1 is provided with a drive assembly 16 located between two load bags 10, which facilitates the movement of the unmanned surface vessel (USV). The bottom of the shell 1 on the side away from the drive assembly 16 is provided with a sonar assembly 19 located between the two load bags 10. It should be noted that the sonar assembly 19 detects underwater objects, terrain, obstacles and water depth, and detects the surrounding environment by emitting sound waves and receiving echo signals. At the same time, it can also identify and classify objects through signals, helping the USV to identify specific targets such as marine life, shipwrecks or potential resource collection areas.

[0040] Furthermore, during the operation of the deep-sea unmanned surface vessel (USV), the draft can be adaptively adjusted according to changes in the marine environment, such as water flow, buoyancy, and temperature. The pump 21 facilitates the extraction of seawater, and the connection port 17 is connected to the pump 21 via the transport pipe 22, which facilitates the pump 21's water extraction operation. The water is transported to the interior of the weight bag 10 through the inner pipe 20, which increases the weight of the USV and thus facilitates a deeper draft. When it is necessary to reduce the draft, the pump 21 extracts and discharges the water from the weight bag 10, allowing the USV to float and facilitating subsequent deep-sea USV cruise operations.

[0041] The main processor 5 facilitates the analysis and processing of data collected by the detection and monitoring components mounted on the deep-sea unmanned surface vessel (USV). After judgment, the USV is controlled for operation. The sonar component 19 detects underwater objects, terrain, obstacles, and water depth. It detects the surrounding environment by emitting sound waves and receiving echo signals. It can also identify and classify objects through signals, helping the USV identify specific targets such as marine life, shipwrecks, or potential resource collection areas. The photovoltaic components facilitate the power supply to the internal energy storage tank. When the USV is operating in surface water or shallow water, the photovoltaic components can charge the battery with solar energy, extending the USV's working time and ensuring its long-term autonomy. The voltage regulator 15 facilitates voltage stabilization, ensuring stable power resources and facilitating subsequent power storage.

[0042] The second embodiment of the present invention relates to a deep-sea terrain adaptive cruise path planning method, the process of which is as follows: Figure 6 As shown, the details are as follows:

[0043] Step 101: Receive multi-source environmental data collected by the monitoring component, radar transceiver component, and sonar component 19 and processed by the first processor 3, and determine a standardized underwater three-dimensional terrain map and obstacle coordinate model based on the multi-source environmental data after data cleaning and preprocessing.

[0044] Step 102: Input the cleaned and preprocessed multi-source environmental data into the obstacle identification and risk classification model. Through the obstacle identification and risk classification model and combined with the standardized underwater 3D topographic map and obstacle coordinate model, output the real-time path planning for the deep-sea unmanned surface vessel.

[0045] Step 103: Send the real-time path planning for the deep-sea unmanned surface vessel to the second processor 14 for verification.

[0046] In practical applications, the specific steps include: Step 1: Synchronous acquisition of multi-source environmental data: During the unmanned surface vessel's navigation, the sonar component continuously transmits and receives underwater acoustic signals, and collects deep-sea topographic data and obstacle data (reefs, shipwrecks, underwater debris, nets, undulating terrain) in real time from the front, sides, and below, and outputs raw data such as obstacle distance, orientation, size, outline, and relative speed; at the same time, the lighting is turned on to provide auxiliary lighting, and the camera + radar acquires high-definition images at close range, making up for the shortcomings of sonar detail recognition, forming dual detection data of "large-scale sonar scanning + close-range visual verification", and all detection data are uploaded to the first processor 3 in real time;

[0047] Step 2: The first processor 3 performs data cleaning and preprocessing: The first processor 3 filters and reduces noise and removes clutter from the received raw sonar data, filtering out invalid interference signals generated by deep-sea currents, water temperature, and plankton, and correcting sonar ranging errors; at the same time, it completes the spatiotemporal registration of image data and sonar data, unifies the data coordinate system, and removes invalid and abnormal data, and then sends it to the main processor 5.

[0048] Step 3: Machine learning model completes obstacle identification and risk classification: The main processor 5 outputs a standardized underwater 3D topographic map and obstacle coordinate model based on multi-source environmental data, and calls a pre-trained deep learning / reinforcement learning model to extract features from the standardized sonar data: identifying obstacle types (fixed terrain obstacles, dynamic floating obstacles, large hard obstacles, small risk-free debris), determining the obstacle threat level, and calculating the obstacle coverage area and restricted areas. The model is trained on massive amounts of deep-sea topographic and obstacle samples, enabling it to accurately distinguish between passable and restricted areas, avoiding the misjudgment and omission problems of traditional fixed algorithms.

[0049] Step 4: Real-time dual-layer path planning (global baseline + local dynamic replanning): 1. Global path: Based on the preset cruise endpoint, an initial baseline route is generated in combination with the overall deep-sea topography; 2. Local dynamic replanning: The model dynamically generates multiple alternative detour routes based on the real-time detected obstacle distribution and hull navigation status, and simultaneously calculates five core indicators for each route: collision risk, navigation energy consumption, route smoothness, navigation distance, and attitude stability.

[0050] Step 5: Optimal route selection decision: The machine learning algorithm comprehensively evaluates all candidate routes through a weighted scoring mechanism, and prioritizes the optimal route with "zero collision risk, lowest energy consumption, smooth route, and minimal deviation from the original cruise path", while discarding sharp turns, high-risk, and high-energy-consumption routes to ensure cruise stability and safety.

[0051] Step 6: Closed-loop execution and real-time iterative update: The main processor 5 sends the optimal route command to the unmanned surface vessel's power and steering mechanism to adjust the heading and speed in real time; during the navigation, the sonar continuously scans the environment, and the model iteratively updates the path in milliseconds. Once a new obstacle or terrain change occurs, the route is immediately replanned, and the entire process is adaptively obstacle avoidance to achieve intelligent cruising in complex deep-sea terrain.

[0052] The third embodiment of the present invention relates to a deep-sea terrain adaptive cruise path planning method, the process of which is as follows: Figure 7 As shown, the details are as follows:

[0053] Step 201: Receive the full-dimensional hull status data collected in real time by the sensors and preprocessed from the first processor 3.

[0054] Step 202: Input the preprocessed full-dimensional hull state data into the adaptive buoyancy adjustment model. The adaptive buoyancy adjustment model performs intelligent algorithm iterative calculation on the full-dimensional hull state data to obtain the optimal stable draft under the current sea state, and at the same time determines the flow rate, duration and speed of water supply and drainage of the load bag 10.

[0055] Step 203: Send the optimal stable draft under the current sea state and the flow rate, duration and rate of water supply and drainage of the load bag 10 to the second processor 14 for verification.

[0056] It should be noted that a dual-algorithm coupling architecture of a lightweight BP neural network regression algorithm and a DQN deep reinforcement learning algorithm is adopted to replace the traditional PID fixed parameter adjustment. Among them, the BP neural network is responsible for accurately predicting the optimal draft, while the DQN reinforcement learning is responsible for dynamic adaptive adjustment and working condition adaptation. With the hydrodynamic characteristics of the deep-sea unmanned surface vessel as a priori constraint, it takes into account both prediction accuracy and environmental adaptability, perfectly adapting to the complex working conditions of varying water depth, current, and seawater density in the deep sea, and solving the defects of traditional control algorithms such as fixed parameters, weak anti-interference, and poor dynamic adjustment accuracy.

[0057] In practical applications, the specific steps include: Step 1: Real-time acquisition of full-dimensional hull status data: Sensors collect the core operating parameters of the unmanned surface vessel in real time and upload them to the machine learning model. The core data includes: real-time diving depth, seawater buoyancy value, three-axis acceleration (lifting, swaying, and propulsion acceleration), hull tilt angle, current water volume in the load bag, deep-sea seawater density, water flow disturbance parameters, etc., to comprehensively reflect the real-time force and attitude status of the hull.

[0058] Step 2: Model Construction Deep-sea Buoyancy Dynamics Model: The machine learning model is based on the fundamental rules of fluid mechanics and combined with a large number of deep-sea navigation samples to train an adaptive buoyancy adjustment model. It can dynamically fit the buoyancy balance equation of the hull under different depths and sea conditions. Unlike fixed parameter adjustment, it can adapt to the complex variable density and variable current environment in the deep sea.

[0059] Step 3: Attitude Deviation and Balance Error Calculation: Model Comparison of Preset Stable Attitude Parameters and Real-Time Operating Parameters of Unmanned Surface Vessel: When sudden acceleration changes, hull swaying, buoyancy imbalance, or abnormal depth fluctuations are detected, the current buoyancy deviation, attitude deviation, and rise and fall trend error are automatically calculated to determine whether the hull is at risk of rising, sinking, or tilting.

[0060] Step 4: Accurately solve the optimal target draft: The model combines real-time depth, buoyancy, acceleration, and ocean current disturbance data, and iteratively calculates through intelligent algorithms to output the optimal stable draft that is uniquely matched under the current sea conditions. At the same time, it accurately calculates the amount of water that needs to be replenished or discharged, and determines the flow rate, duration, and rate of water supply and drainage for the load bag.

[0061] Step 5: Adaptive Execution Adjustment and Steady-State Closed-Loop: Based on the model calculation results, the main processor 5 controls the operation of the water supply and drainage mechanism: when the draft needs to be increased, water is injected into the weight bag to increase weight; when the draft needs to be decreased, water is discharged from the weight bag to reduce weight. During the adjustment process, the model continuously monitors the changes in the hull attitude and buoyancy, dynamically fine-tuning the water supply and drainage parameters until the unmanned surface vessel reaches the required draft, achieves buoyancy balance, and stabilizes its attitude, preventing rapid ascent and descent, swaying, and imbalance problems.

[0062] Furthermore, in conjunction with the second and third embodiments, it is also necessary to explain that:

[0063] 1. First Processor 3 (Front-end Sensing Preprocessing Layer): Data Acquisition, Filtering, and Noise Reduction: The First Processor 3 is the core of the front-end sensing, directly connecting to all peripherals such as sonar components, sensing components, attitude sensors, depth sensors, and photovoltaic power supply modules. It is responsible for the first round of processing of raw data: real-time acquisition of all detection data and equipment operation data, completion of data filtering, noise reduction, format standardization, and abnormal data marking, filtering out invalid data caused by deep-sea environmental interference, and synchronously transmitting the preprocessed, valid, and standardized data to the main processor. Simultaneously, it monitors equipment faults in real time, immediately reporting any sensor anomalies to ensure the accuracy and reliability of the data source.

[0064] 2. Main Processor 5 (Core Decision Layer): AI Computation, Path Planning, Attitude Control: The main processor 5 is the core hub of the system. It receives standardized data transmitted from the first processor and performs core intelligent calculations: calling machine learning models to complete obstacle recognition, risk assessment, and real-time planning of the optimal cruising path. Simultaneously, it performs adaptive draft calculation, hull attitude balance control, and energy consumption matching analysis. It generates preliminary cruising path plans, attitude adjustment commands, and equipment operation commands, and synchronously uploads the preliminary decision results to the second processor 14 for secondary verification.

[0065] 3. Second processor 14 (verification and execution control layer): logic verification, scene adaptation, instruction issuance: The second processor 14 is responsible for global logic verification and scene adaptation judgment, which is the key to ensuring operation in complex deep-sea environment: (1) Verify the rationality of the path planning scheme, attitude adjustment parameters, and obstacle avoidance instructions output by the main processor, and eliminate unreasonable and high-risk instructions in combination with deep-sea restricted areas, operation task requirements, and equipment operating limit parameters; (2) Dynamically adapt the cruise strategy according to the real-time sea conditions, remaining power, and equipment status, optimize the cruise path in complex terrain, strong current, and low power scenarios, and prioritize navigation safety and task completion; (3) After the verification is passed, finally confirm and lock the optimal cruise path, issue execution instructions, and control all operation actions such as unmanned surface vessel navigation, obstacle avoidance, attitude adjustment, and power supply switching.

[0066] 4. Three-tiered closed-loop system for complex deep-sea mission scenarios: The three-tiered processor forms a closed-loop system of "front-end preprocessing → core intelligent decision-making → back-end verification and execution": the first processor ensures data accuracy, the main processor ensures optimal intelligent decision-making, and the second processor ensures the safety and reliability of instructions; the three work together to effectively cope with extreme scenarios such as complex deep-sea terrain, random obstacles, variable sea conditions, and strong signal interference, avoiding problems such as single processor calculation errors, response delays, and data distortion, ensuring that the unmanned surface vessel can stably plan its cruise path throughout the entire process and accurately complete deep-sea exploration, cruise, and operation tasks.

[0067] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the invention. All equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in this invention should still be covered by the claims of this invention.

Claims

1. A deep-sea unmanned surface vessel, comprising a hull (1), characterized in that: The upper surface of the housing (1) is provided with a lighting component, a monitoring component, a main control component and a photovoltaic component in sequence along the length direction of the housing (1). The main control component includes a working slot (25) disposed on the upper surface of the housing (1) and located between the monitoring component and the photovoltaic component, a main processor (5) located in the working slot (25) and a radar transceiver component. The housing (1) has two symmetrically arranged weight bags (10) on both sides. The bottom of the housing (1) is provided with a drive assembly (16) located between the two weight bags (10). The bottom of the housing (1) on the side away from the drive assembly (16) is provided with a sonar assembly (19) located between the two weight bags (10).

2. The deep-sea unmanned surface vessel according to claim 1, characterized in that: The lighting assembly includes a rectangular plate (11) mounted on the upper surface of the housing (1), a lighting lamp (12) located on the side of the rectangular plate (11), and a number of work lamps (2). The lighting lamp (12) and the number of work lamps (2) are mounted on the upper surface of the housing (1) at equal intervals along the circumference of the rectangular plate (11).

3. The deep-sea unmanned surface vessel according to claim 1, characterized in that: The monitoring component includes a rotatable support base (13) mounted on the upper surface of the housing (1) and a monitor (4) mounted on the support base (13). The upper surface of the housing (1) is also provided with a first processor (3) located on the side of the support base (13) and connected to the main processor (5). The shell (1) is also provided with a sensor assembly for collecting full-dimensional hull status data of the deep-sea unmanned surface vessel and connected to the first processor (3). The first processor (3) is also connected to the lighting assembly, monitoring assembly, radar transceiver assembly, photovoltaic assembly and sonar assembly (19).

4. A deep-sea unmanned surface vessel according to claim 1, characterized in that: The radar transceiver assembly includes a lidar transmitter (6) mounted in the working slot (25) via a support column (9) and located between the main processor (5) and the photovoltaic module, and a plurality of lidar receivers (7) mounted circumferentially along the working slot (25) on the upper surface of the housing (1). A second processor (14) connected to the main processor (5) is provided on the support column (9).

5. A deep-sea unmanned surface vessel according to claim 1, characterized in that: The photovoltaic module includes a photovoltaic panel (8) mounted on the upper surface of the housing (1) and a voltage regulator (15) located on the side of the photovoltaic panel (8). The voltage regulator (15) is connected to the photovoltaic panel (8) and the energy storage box located inside the housing (1).

6. A deep-sea unmanned surface vessel according to claim 1, characterized in that: Inside the housing (1), a pump (21) is provided corresponding to the load bag (10) and is connected to the load bag (10) through an inner pipe (20). The pump (21) is also connected to a transport pipe (22). The end of the transport pipe (22) away from the pump (21) extends to the outside of the housing (1), and a connection port (17) is provided on the end of the transport pipe (22) away from the pump (21). A filter (18) is threaded onto the connection port (17).

7. A deep-sea terrain adaptive cruise path planning method, characterized in that: The deep-sea unmanned surface vessel based on any one of claims 1-6, applied to the main processor (5), comprises the following steps: The system receives multi-source environmental data collected by the monitoring component, radar transceiver component and sonar component (19) sent by the first processor (3) and after data cleaning and preprocessing. Based on the multi-source environmental data after data cleaning and preprocessing, it determines a standardized underwater three-dimensional topographic map and obstacle coordinate model. The cleaned and preprocessed multi-source environmental data is input into the obstacle identification and risk classification model. The obstacle identification and risk classification model, combined with the standardized underwater 3D topographic map and obstacle coordinate model, outputs real-time path planning for the deep-sea unmanned surface vessel. The real-time path planning for the deep-sea unmanned surface vessel is sent to the second processor (14) for verification.

8. The deep-sea terrain adaptive cruise path planning method according to claim 7, characterized in that: The process of outputting real-time path planning for the deep-sea unmanned surface vessel by combining the obstacle recognition and risk classification model with the standardized underwater 3D terrain map and obstacle coordinate model includes: Feature extraction is performed on the standardized sonar data in the multi-source environmental data, and obstacle types, obstacle threat levels, obstacle coverage and restricted areas are identified based on the extracted features; Based on the identified obstacle types, the determined obstacle threat levels, and the calculated obstacle coverage and restricted areas, real-time path planning for deep-sea unmanned surface vessels is obtained.

9. The deep-sea terrain adaptive cruise path planning method according to claim 8, characterized in that: The real-time path planning for the deep-sea unmanned surface vessel (USV) is obtained based on the identified obstacle types, the determined obstacle threat levels, and the calculated obstacle coverage and restricted areas, including: Based on the preset cruise endpoint, and combined with the identified obstacle types, the determined obstacle threat level, and the calculated obstacle coverage area and restricted areas, an initial baseline route is generated; Multiple alternative detour routes are dynamically generated based on the real-time detected obstacle distribution and the hull's navigation status. The core indicators of each alternative detour route are calculated simultaneously, and a weighted scoring mechanism is used to comprehensively evaluate all alternative detour routes. The optimal alternative detour route is obtained based on the evaluation results.

10. The deep-sea terrain adaptive cruise path planning method according to claim 7, characterized in that: The method further includes the following steps: Receive the full-dimensional hull status data collected in real time by the sensors and preprocessed from the first processor (3); The preprocessed full-dimensional hull state data is input into the adaptive buoyancy adjustment model. The adaptive buoyancy adjustment model performs intelligent algorithm iterative calculation on the full-dimensional hull state data to obtain the optimal stable draft under the current sea state, and at the same time determines the flow rate, duration and speed of the load bag (10) for water supply and drainage. The optimal stable draft under the current sea state and the flow rate, duration and rate of water supply and drainage of the load bag (10) are sent to the second processor (14) for verification.

Images