Ship collision risk early warning and avoidance decision-making system

Abstract

The invention relates to the technical field of ship navigation safety, in particular to a ship collision risk early warning and avoidance decision-making system which comprises a data acquisition module, a risk assessment module, a path planning module and an execution control module. According to the system, through multi-source data real-time acquisition and a deep learning algorithm, a collision risk index is quantitatively evaluated, an optimal collision avoidance path is generated, and a ship control device is accurately controlled to complete collision avoidance operation. Meanwhile, a communication module is introduced into the system to achieve information interaction between ships and a shore base, a collision avoidance path is optimized, and remote monitoring is supported. The dynamic adaptability and safety of the ship in a complex water area environment can be improved, the collision risk is effectively reduced, and navigation safety is guaranteed.

Description

Technical Field

[0001] This invention belongs to the field of intelligent shipping and maritime traffic safety technology, specifically a ship collision risk warning and avoidance decision system. Background Technology

[0002] Patent document CN112419786B discloses a ship collision warning system. This system, through modules such as detection devices, sensing devices, and power units, monitors the ship's position, distance, and navigation status, and performs obstacle avoidance maneuvers in conjunction with a path planning device. However, this system relies primarily on static wave measuring devices and simple power output adjustments during operation, failing to adequately consider factors such as multi-ship interactions in dynamic environments, complex weather conditions, and real-time path optimization. Furthermore, its warning mechanism lacks the ability to quantitatively assess collision risk, potentially leading to false alarms or missed alarms, thus affecting the system's reliability and practicality.

[0003] Patent document CN118658338B discloses a ship collision warning method and related equipment. This method acquires ship navigation data (including position, historical trajectory, heading, and speed information), uses generative adversarial networks to predict the ship's future trajectory, and combines this with timeline fitting to assess collision risk. While this method has certain advantages in trajectory prediction and risk assessment, it primarily focuses on collision warnings within the target sea area and lacks specific guidance for avoidance decisions. Furthermore, this approach requires high data acquisition accuracy and computational resources, potentially facing bottlenecks in real-time performance and computational efficiency in practical applications, especially in high-density ship traffic environments where rapid response may be difficult to meet. These issues indicate that existing ship collision warning technologies still have shortcomings in dynamic environmental adaptability, collision risk quantification assessment, avoidance decision support, and real-time performance, urgently requiring improvement to better meet the safety requirements of ship navigation in complex aquatic environments. Summary of the Invention

[0004] The purpose of this invention is to address the problems of insufficient adaptability to dynamic environments, lack of quantitative assessment capabilities for collision risks, and weak support for avoidance decisions in existing technologies, and to provide a ship collision risk early warning and avoidance decision system.

[0005] The objective of this invention is achieved through the following technical solution: A ship collision risk warning and avoidance decision-making system includes a data acquisition module, a risk assessment module, a path planning module, and an execution control module. The data acquisition module is used to acquire multi-source data on the target ship and its surrounding environment in real time, including the target ship's position, heading, and speed information, as well as the navigation status of surrounding ships, meteorological conditions, and water topography data. The risk assessment module receives the data transmitted by the data acquisition module, analyzes the relative position, speed, and acceleration between the target ship and other ships based on a deep learning algorithm, and generates a collision risk index by combining meteorological conditions and water topography data. The deep learning algorithm adopts an LSTM-RNN hybrid model. The input layer receives preprocessed time-series data (sampling frequency 1Hz, features including: relative distance, relative velocity vector, angular velocity, wind speed, wave height, and visibility level), which is then normalized using min-max. The intermediate layer contains three LSTM units (128 nodes per layer) using the ReLU activation function. The output layer outputs the risk probability value using the Sigmoid activation function. Training data was derived from historical AIS data (10,000 samples), meteorological observation data, and simulated scenario data from the XX database, divided into training and test sets in a 7:3 ratio. The model employed a cross-entropy loss function and the Adam optimizer (learning rate 0.001, batch size 32), undergoing 100 training iterations, achieving a validation accuracy of over 95%. The path planning module generated the optimal collision avoidance path based on the collision risk index output by the risk assessment module, combined with the target vessel's current position and navigation task. The execution control module received the collision avoidance path instructions generated by the path planning module and adjusted the vessel's heading and speed through its steering devices to complete the collision avoidance operation. Preferably, the risk assessment module calculates the collision risk index using the following process: The risk assessment module first determines the minimum approach distance between the target vessel and surrounding vessels; if the minimum approach distance is greater than a set safety threshold, the collision risk index is marked as low risk; if the minimum approach distance is less than or equal to the safety threshold, it further determines whether the relative speed between the target vessel and surrounding vessels exceeds a set speed difference threshold; if the relative speed does not exceed the speed difference threshold, the collision risk index is marked as medium risk; if the relative speed exceeds the speed difference threshold, the collision risk index is corrected based on visibility and wave level in the meteorological conditions; if the visibility is lower than a set visibility threshold or the wave level is higher than a set wave level threshold, the collision risk index is marked as high risk.

[0006] Preferably, the path planning module uses the A* algorithm to generate the initial path, with the minimum deviation angle as the heuristic function (weight coefficient 0.7). Feasibility verification is achieved using a water depth topographic grid map (10-meter resolution): the water depth at each point along the path is calculated, and if it is less than the safe draft + 1-meter buffer, the path is deemed infeasible. If infeasible, random sampling optimization (RRT) is used to regenerate the path, with an upper limit of 10 iterations.

[0007] Preferably, after receiving the collision avoidance path command, the execution control module adjusts the ship's course via the rudder and adjusts the speed via the main engine. The rudder deflection angle θ = Kp × Δα + Ki × ∫Δαdt + Kd × d(Δα) / dt (PID control, Kp = 0.8, Ki = 0.1, Kd = 0.05, Δα is the course deviation angle). The main engine power adjustment P = K × Δv (K = 50 kW / knot, Δv is the speed difference), with power limited to ±20% of the rated power.

[0008] Preferably, the data acquisition module acquires data in real time through various sensors installed on the ship, including radar sensors, AIS receivers, weather sensors, and depth sounders. The radar sensors are mounted on the top of the ship's mast with their antennas facing outwards to detect the position and motion of surrounding vessels. The AIS receiver is installed in the bridge and receives automatic identification signals from surrounding vessels via its antenna. The weather sensors are installed above the deck to measure wind speed, wind direction, air pressure, and visibility. The depth sounder is installed on the bottom of the ship and uses sonar beams to detect underwater topography. Preferably, when a target vessel enters a high-density vessel traffic area, the data acquisition module increases the scanning frequency of surrounding vessels and transmits the scanning results to the risk assessment module in real time. The risk assessment module updates the collision risk index based on the high-frequency scanning data and transmits the updated collision risk index to the path planning module to dynamically adjust the collision avoidance path.

[0009] Preferably, the path planning module considers the target vessel's maneuverability limitations, including maximum turning rate and maximum acceleration / deceleration capability, when generating a collision avoidance path. Based on the target vessel's maneuverability parameters, the path planning module smooths the generated collision avoidance path to ensure that the path conforms to the vessel's actual maneuverability.

[0010] Preferably, after completing the collision avoidance maneuver, the execution control module reacquires the target vessel's navigation status via the data acquisition module and feeds this navigation status back to the risk assessment module. The risk assessment module recalculates the collision risk index based on the new navigation status and determines whether further adjustments to the collision avoidance path are necessary.

[0011] Preferably, the system also includes a communication module for information exchange between the target vessel and surrounding vessels. The communication module establishes a communication link with surrounding vessels via a VHF radio, transmits the target vessel's collision avoidance intentions and planned path to the surrounding vessels, and simultaneously receives feedback information from surrounding vessels to optimize the collision avoidance path.

[0012] Preferably, the communication module also establishes a remote connection with the shore-based monitoring center via satellite communication equipment, uploading the target vessel's navigation data and collision avoidance operation records to the shore-based monitoring center so that the shore-based monitoring center can monitor and guide the target vessel's navigation safety in real time.

[0013] The advantages of this invention are mainly reflected in the following aspects: This invention acquires multi-source data in real time through a data acquisition module and combines it with deep learning algorithms to quantitatively assess collision risks, effectively improving the system's accuracy and intelligence. The risk assessment module generates an accurate collision risk index through comprehensive analysis of minimum approach distance, relative speed, meteorological conditions, and water topography data, avoiding false alarms and missed alarms, and improving the system's reliability.

[0014] This invention incorporates water topography data and ship maneuverability constraints into the path planning module, ensuring that the generated collision avoidance path conforms to both actual navigation conditions and the ship's maneuverability requirements. By dynamically adjusting the collision avoidance path, the system can quickly respond to changes in complex aquatic environments, improving the system's real-time performance and adaptability.

[0015] This invention uses an execution control module to precisely control the ship's steering gear and main engine, ensuring the smooth implementation of collision avoidance operations. The execution control module dynamically adjusts control parameters based on the current state of the target ship, achieving both smoothness and efficiency in collision avoidance operations.

[0016] This invention enables information exchange between the target vessel and surrounding vessels, as well as the shore-based monitoring center, through a communication module, enhancing the system's collaborative capabilities and security. The communication module not only optimizes collision avoidance paths but also provides comprehensive navigation data support to the shore-based monitoring center, offering multiple safeguards for vessel navigation safety. Attached Figure Description

[0017] Figure 1 This is a block diagram of the overall structure of the ship collision risk warning and avoidance decision-making system of the present invention, showing the connection relationship between the data acquisition module, risk assessment module, path planning module and execution control module.

[0018] Figure 2 This is a flowchart of the risk assessment module in this invention, which describes in detail the classification and determination process of the collision risk index, including a comprehensive analysis of minimum approach distance, relative speed and weather conditions.

[0019] Figure 3 This is a logical diagram illustrating the path planning module of the present invention that generates collision avoidance paths, showing the main steps of initial path generation, water terrain verification, and path smoothing.

[0020] Figure 4The flowchart of the control module for adjusting the ship's course and speed is shown in the present invention, illustrating the dynamic calculation process of the steering gear deflection angle and the main engine power adjustment.

[0021] Figure 5 This is a schematic diagram of information interaction of the communication module of the present invention, showing the data transmission link between the target vessel and surrounding vessels and the shore-based monitoring center.

[0022] The attached diagram is labeled as follows: 1. Data acquisition module; 2. Risk assessment module; 3. Path planning module; 4. Execution control module; 5. Communication module; 6. Radar sensor; 7. AIS receiver; 8. Meteorological sensor; 9. Water depth sounder; 10. VHF radio station; 11. Satellite communication equipment. Detailed Implementation

[0023] This invention relates to a ship collision risk warning and avoidance decision-making system, the overall structure of which is as follows: Figure 1 As shown, the system includes a data acquisition module 1, a risk assessment module 2, a path planning module 3, an execution control module 4, and a communication module 5. These modules are interconnected and work together via a data transmission link to achieve real-time monitoring, assessment, path planning, and collision avoidance maneuvers for ship collision risks. The specific implementation of this system is described in detail below with reference to the accompanying drawings.

[0024] Data acquisition module 1 is the foundation of the entire system, responsible for acquiring multi-source data on the target vessel and its surrounding environment in real time. This module consists of various sensors, including a radar sensor 6, an AIS receiver 7, a meteorological sensor 8, and a depth sounder 9. Radar sensor 6 is mounted on the top of the ship's mast with its antenna facing outwards, used to detect the position, heading, and speed of surrounding vessels. AIS receiver 7 is installed in the bridge, receiving automatic identification signals from surrounding vessels via its antenna, providing more accurate navigation status data. Meteorological sensor 8 is installed above the ship's deck, used to measure meteorological conditions such as wind speed, wind direction, air pressure, and visibility. Depth sounder 9 is installed on the bottom of the ship, using sonar beams to detect underwater topography, providing water topography data for subsequent path planning. These sensors are connected to the central processing unit of data acquisition module 1 via wired or wireless means, and the collected data is integrated and transmitted to risk assessment module 2.

[0025] Risk assessment module 2 receives data from data acquisition module 1 and analyzes the relative position, velocity, and acceleration of the target vessel compared to other vessels based on deep learning algorithms. Its calculation process is as follows: Figure 2As shown, the minimum approach distance between the target vessel and surrounding vessels is first determined. The safety threshold is set to 500 meters (based on IMO collision avoidance rules), and the speed difference threshold is set to 5 knots (based on vessel type library matching). The visibility threshold is set to 1000 meters (corresponding to dense fog conditions), and the wave level threshold is set to level 4 (corresponding to a wave height of 1.25 meters). These thresholds can be dynamically adjusted via a configuration file, specifically based on: safety threshold = k1 × vessel length (k1 = 3, default value); speed difference threshold = k2 × maximum speed (k2 = 0.2, default value); visibility threshold is fixed at 1000 meters; wave level threshold is fixed at level 4; if the relative speed exceeds the speed difference threshold, the collision risk index is corrected based on visibility and wave level in the meteorological conditions. If visibility is lower than the set visibility threshold or the wave level is higher than the set wave level threshold, the collision risk index is marked as high risk. The risk assessment module 2 generates the collision risk index through the above steps and transmits the results to the path planning module 3.

[0026] The path planning module 3 generates the optimal collision avoidance path based on the collision risk index output by the risk assessment module 2, combined with the target vessel's current position and navigation mission. Its logical diagram is shown below. Figure 3 As shown, the initial path is first generated to minimize the angle at which the target vessel deviates from its current course, and its feasibility is verified using water topography data. If the initial path is infeasible, the path is regenerated and the verification process is repeated until a feasible path is found. The path planning module 3 also considers the target vessel's maneuverability limitations, including maximum turning rate and maximum acceleration / deceleration capability, and smooths the generated collision avoidance path to ensure that the path conforms to the vessel's actual maneuverability. The final generated collision avoidance path command is transmitted to the execution control module 4.

[0027] The execution control module 4 receives the collision avoidance path instructions generated by the path planning module 3, and adjusts the ship's course and speed through the ship's steering devices to complete the collision avoidance operation. Its control flow is as follows: Figure 4 As shown, the execution control module 4 calculates the servo deflection angle based on the deviation angle between the target vessel's current course and the target course of the collision avoidance path, and controls the actual deflection of the servo drive unit. Simultaneously, the execution control module 4 calculates the main engine power adjustment based on the difference between the target vessel's current speed and the target speed of the collision avoidance path, and adjusts the main engine's output power through the main engine control unit. After completing the collision avoidance operation, the execution control module 4 re-acquires the target vessel's navigation status through the data acquisition module 1 and feeds this navigation status back to the risk assessment module 2 to recalculate the collision risk index and determine whether further adjustments to the collision avoidance path are needed.

[0028] Communication module 5 is used to enable information exchange between the target vessel and surrounding vessels, as well as the shore-based monitoring center. A schematic diagram of this information exchange is shown below. Figure 5 As shown, communication module 5 establishes a communication link with surrounding vessels via VHF radio station 10, transmitting the target vessel's collision avoidance intentions and planned path to the surrounding vessels, while simultaneously receiving feedback information from surrounding vessels to optimize the collision avoidance path. Communication module 5 also establishes a remote connection with a shore-based monitoring center via satellite communication equipment 11, uploading the target vessel's navigation data and collision avoidance operation records to the shore-based monitoring center, enabling the center to monitor and guide the target vessel's navigation safety in real time.

[0029] In practical applications, when a target vessel enters a high-density shipping area, the data acquisition module 1 increases the scanning frequency of surrounding vessels and transmits the scan results to the risk assessment module 2 in real time. The risk assessment module 2 updates the collision risk index based on the high-frequency scan data and transmits the updated index to the path planning module 3 to dynamically adjust the collision avoidance path. For example, in busy port waters, a target vessel may face dense navigation from multiple surrounding vessels. In this situation, the system can effectively avoid potential collision risks through high-frequency data acquisition and dynamic path adjustment. Furthermore, the communication module 5 plays a crucial role in this process, coordinating collision avoidance operations through information exchange with surrounding vessels to prevent new risks arising from unilateral path adjustments.

[0030] In adverse weather conditions, such as low visibility or high waves, the risk assessment module 2 uses data from the meteorological sensor 8 to correct the collision risk index, ensuring more accurate assessment results. When generating collision avoidance paths, the path planning module 3 prioritizes paths that avoid high-risk areas and verifies the feasibility of the paths using water topography data. For example, in narrow channels, target vessels may need to navigate around shoals or reefs; the path planning module 3 uses data from the depth sounder 9 to ensure that the generated path is both safe and conforms to actual navigation conditions.

[0031] When adjusting the ship's course and speed, the execution control module 4 fully considers the target ship's maneuverability limitations. For example, for large cargo ships, whose turning rate and acceleration / deceleration capabilities are limited, the execution control module 4 ensures smooth and efficient collision avoidance operations through precise control of the steering gear and main engine. Simultaneously, after completing the collision avoidance operation, the execution control module 4 re-acquires the target ship's navigation status through the data acquisition module 1 and feeds it back to the risk assessment module 2 to recalculate the collision risk index. This closed-loop control mechanism ensures the system's real-time performance and adaptability in complex water environments. The communication module 5 not only enables information exchange between the target ship and surrounding vessels but also establishes a remote connection with the shore-based monitoring center via satellite communication equipment 11, providing multiple safeguards for ship navigation safety. For example, during ocean voyages, the target ship may face sudden weather changes or equipment malfunctions. The communication module 5 uploads navigation data and collision avoidance operation records to the shore-based monitoring center, enabling the center to provide timely guidance and support. Furthermore, the communication module 5 can also receive navigation suggestions or instructions from the shore-based monitoring center, further enhancing the system's collaborative capabilities and safety.

[0032] The various modules of this system work closely together to achieve a complete process from data acquisition to risk assessment, path planning, execution control, and information exchange. Data acquisition module 1 acquires comprehensive navigation data through multiple sensors, providing fundamental support for subsequent modules. Risk assessment module 2 generates an accurate collision risk index through comprehensive analysis of minimum approach distance, relative speed, and weather conditions, avoiding false alarms and missed alarms. Path planning module 3 combines water topography data and ship maneuverability limitations to generate collision avoidance paths that conform to actual navigation conditions, and improves the system's real-time performance and adaptability through dynamic path adjustments. Execution control module 4 ensures the smooth implementation of collision avoidance operations through precise control of the steering gear and main engine, and continuously optimizes the collision avoidance effect through a closed-loop control mechanism. Communication module 5 enhances the system's collaborative capabilities and security through information exchange with surrounding vessels and shore-based monitoring centers, providing multiple safeguards for ship navigation safety.

[0033] Through the specific embodiments described above, this invention achieves real-time monitoring, assessment, path planning, and collision avoidance maneuvers for ship collision risks, significantly improving the system's accuracy, intelligence, and adaptability. The close cooperation and collaborative work between the modules ensures the system's reliability and efficiency in complex aquatic environments.

[0034] To enable those skilled in the art to fully understand and implement this invention, the operating principle and implementation steps of the system are described in detail below in conjunction with specific application scenarios. When a target vessel enters a high-density shipping traffic area, the data acquisition module 1 increases the scanning frequency of surrounding vessels through radar sensor 6 and AIS receiver 7. Radar sensor 6 is installed at the top of the ship's mast, with its antenna facing outwards, to detect the position, heading, and speed of surrounding vessels in real time. AIS receiver 7 receives automatic identification signals from surrounding vessels through its antenna, providing the system with more accurate navigation status data. Meteorological sensor 8 is installed above the ship's deck to continuously monitor meteorological conditions such as wind speed, wind direction, air pressure, and visibility, while depth sounder 9 is installed at the bottom of the ship to detect underwater topography using sonar beams. The data collected by these sensors is integrated and transmitted to risk assessment module 2.

[0035] Risk assessment module 2 analyzes the relative position, speed, and acceleration between the target vessel and surrounding vessels using a deep learning algorithm. First, the module calculates the minimum approach distance between the target vessel and surrounding vessels. If this distance is greater than a set safety threshold, the collision risk index is marked as low risk; if it is less than or equal to the safety threshold, it further determines whether the relative speed exceeds a speed difference threshold. If the relative speed does not exceed the speed difference threshold, the collision risk index is marked as medium risk; if it exceeds the speed difference threshold, the collision risk index is corrected using visibility and wave level data provided by meteorological sensor 8. If the visibility is lower than a set visibility threshold or the wave level is higher than a set wave level threshold, the collision risk index is marked as high risk. The generated collision risk index is then transmitted to path planning module 3.

[0036] The path planning module 3 generates an initial collision avoidance path based on the collision risk index. The initial path prioritizes the one with the smallest deviation from the current course, and its feasibility is verified using water topography data provided by the depth sounder 9. If the initial path is infeasible, a new path is generated, and the verification process is repeated until a feasible path is found. The path planning module 3 also considers the target vessel's maximum turning rate and maximum acceleration / deceleration capability, smoothing the generated path to ensure it conforms to the vessel's actual maneuverability. The final generated collision avoidance path command is transmitted to the execution control module 4. Upon receiving the collision avoidance path command, the execution control module 4 adjusts the servo motor deflection angle via the servo motor drive unit to change the target vessel's course. Simultaneously, the main engine control unit adjusts the main engine output power based on the difference between the target vessel's current speed and the target speed of the collision avoidance path, thereby adjusting the speed. The execution control module 4 dynamically calculates the servo motor deflection based on the deviation angle between the target vessel's current course and the target course of the collision avoidance path, and ensures smooth and efficient collision avoidance operation through main engine power adjustments. After completing the collision avoidance operation, the execution control module 4 reacquires the navigation status of the target vessel through the data acquisition module 1 and feeds the navigation status back to the risk assessment module 2 to recalculate the collision risk index and determine whether further adjustments to the collision avoidance path are needed.

[0037] Communication module 5 plays a crucial role throughout the process. Through VHF radio 10, communication module 5 establishes communication links with surrounding vessels, transmitting the target vessel's collision avoidance intentions and planned path to them, while simultaneously receiving feedback from surrounding vessels to optimize the collision avoidance path. Furthermore, communication module 5 establishes a remote connection with a shore-based monitoring center via satellite communication equipment 11, uploading the target vessel's navigation data and collision avoidance operation records to the shore-based monitoring center, enabling real-time monitoring and guidance of the target vessel's navigation safety.

[0038] In adverse weather conditions, such as low visibility or high waves, the risk assessment module 2 uses data from the meteorological sensor 8 to correct the collision risk index, ensuring more accurate assessment results. When generating collision avoidance paths, the path planning module 3 prioritizes paths that avoid high-risk areas and verifies the feasibility of the paths using water topographic data provided by the depth sounder 9. For example, in narrow channels, target vessels may need to navigate around shoals or reefs; the path planning module 3 uses water topographic data to generate paths that are both safe and conform to actual navigation conditions.

[0039] When adjusting the ship's course and speed, the execution control module 4 fully considers the target ship's maneuverability limitations. For large cargo ships, whose turning rate and acceleration / deceleration capabilities are limited, the execution control module 4 ensures smooth and efficient collision avoidance operations through precise control of the steering gear and main engine. Simultaneously, after completing the collision avoidance operation, the execution control module 4 re-acquires the target ship's navigation status through the data acquisition module 1 and feeds this status back to the risk assessment module 2 to recalculate the collision risk index. This closed-loop control mechanism ensures the system's real-time performance and adaptability in complex aquatic environments.

[0040] Communication module 5 not only enables information exchange between the target vessel and surrounding vessels, but also establishes a remote connection with the shore-based monitoring center via satellite communication equipment 11, providing multiple safeguards for vessel navigation safety. For example, during ocean voyages, the target vessel may face sudden weather changes or equipment malfunctions. Communication module 5 uploads navigation data and collision avoidance operation records to the shore-based monitoring center, enabling the center to provide timely guidance and support. Furthermore, communication module 5 can also receive navigation suggestions or instructions from the shore-based monitoring center, further enhancing the system's collaborative capabilities and safety.

Claims

1. A ship collision risk warning and avoidance decision-making system, comprising a data acquisition module (1), a risk assessment module (2), a path planning module (3), and an execution control module (4), characterized in that: The data acquisition module (1) is used to acquire multi-source data of the target vessel and its surrounding environment in real time, including the target vessel's position, heading, and speed information, as well as the navigation status of surrounding vessels, meteorological conditions, and water topography data; the risk assessment module (2) receives the data transmitted by the data acquisition module (1), analyzes the relative position, speed, and acceleration between the target vessel and other vessels based on a deep learning algorithm, and generates a collision risk index in combination with meteorological conditions and water topography data. The deep learning algorithm adopts an LSTM-RNN hybrid model. The input layer receives preprocessed time series data (sampling frequency 1Hz, features including: relative distance, relative velocity vector, angular velocity, wind speed, wave height, and visibility level), which is normalized by min-max. The intermediate layer contains 3 layers of LSTM units (128 nodes per layer) and uses the ReLU activation function. The output layer outputs the risk probability value through the Sigmoid activation function. The training data comes from historical AIS data (10,000 samples), meteorological observation data, and simulated scenario data from the XX database, and is divided into training and test sets in a 7:3 ratio. The model uses the cross-entropy loss function, Adam optimizer (learning rate 0.001, batch size 32), and 100 training iterations, with a verification accuracy of over 95%. The path planning module (3) generates the optimal collision avoidance path based on the collision risk index output by the risk assessment module (2), combined with the current position of the target vessel and the navigation task. The execution control module (4) receives the collision avoidance path instruction generated by the path planning module (3) and adjusts the course and speed through the vessel's control devices to complete the collision avoidance operation.

2. The ship collision risk early warning and avoidance decision-making system according to claim 1, characterized in that: The risk assessment module (2) calculates the collision risk index using the following process: The risk assessment module (2) first determines the minimum approach distance between the target vessel and surrounding vessels; if the minimum approach distance is greater than the set safety threshold, the collision risk index is marked as low risk; if the minimum approach distance is less than or equal to the safety threshold, it further determines whether the relative speed between the target vessel and surrounding vessels exceeds the set speed difference threshold; if the relative speed does not exceed the speed difference threshold, the collision risk index is marked as medium risk; if the relative speed exceeds the speed difference threshold, the collision risk index is corrected by combining the visibility and wave level in the meteorological conditions; if the visibility is lower than the set visibility threshold or the wave level is higher than the set wave level threshold, the collision risk index is marked as high risk.

3. The ship collision risk early warning and avoidance decision-making system according to claim 1, characterized in that: When generating a collision avoidance path, the path planning module (3) prioritizes the path that minimizes the angle at which the target vessel deviates from its current course as the initial path, and performs a feasibility check on the initial path in conjunction with water topography data. If the initial path is not feasible, the path is regenerated and the check process is repeated until a feasible path is found.

4. The ship collision risk early warning and avoidance decision-making system according to claim 1, characterized in that: After receiving the collision avoidance path instruction, the execution control module (4) adjusts the course of the ship through the rudder and adjusts the speed through the main engine. The execution control module (4) calculates the deflection angle of the rudder based on the deviation angle between the current course of the target ship and the target course of the collision avoidance path, and controls the actual deflection of the rudder through the rudder drive unit. At the same time, the execution control module (4) calculates the power adjustment amount of the main engine based on the difference between the current speed of the target ship and the target speed of the collision avoidance path, and adjusts the output power of the main engine through the main engine control unit.

5. The ship collision risk early warning and avoidance decision-making system according to claim 1, characterized in that: The data acquisition module (1) acquires data in real time through a variety of sensors installed on the ship, including a radar sensor (6), an AIS receiver (7), a weather sensor (8), and a depth sounder (9). The radar sensor (6) is installed on the top of the ship's mast with its antenna facing all directions to detect the position and movement of surrounding ships. The AIS receiver (7) is installed in the cockpit and receives automatic identification signals from surrounding ships through its antenna. The weather sensor (8) is installed above the ship's deck to measure wind speed, wind direction, air pressure, and visibility. The depth sounder (9) is installed on the bottom of the ship and detects underwater topography through sonar beams.

6. The ship collision risk early warning and avoidance decision-making system according to claim 1, characterized in that: When the target vessel enters a high-density vessel traffic area, the data acquisition module (1) increases the scanning frequency of surrounding vessels and transmits the scanning results to the risk assessment module (2) in real time; the risk assessment module (2) updates the collision risk index based on the high-frequency scanning data and transmits the updated collision risk index to the path planning module (3) to dynamically adjust the collision avoidance path.

7. The ship collision risk early warning and avoidance decision-making system according to claim 1, characterized in that: When generating a collision avoidance path, the path planning module (3) considers the maneuverability limitations of the target vessel, including the maximum turning rate and maximum acceleration / deceleration capability. The path planning module (3) smooths the generated collision avoidance path according to the maneuverability parameters of the target vessel, ensuring that the path conforms to the actual maneuverability of the vessel.

8. The ship collision risk early warning and avoidance decision-making system according to claim 1, characterized in that: After completing the collision avoidance operation, the execution control module (4) reacquires the navigation status of the target vessel through the data acquisition module (1) and feeds back the navigation status to the risk assessment module (2); the risk assessment module (2) recalculates the collision risk index based on the new navigation status and determines whether further adjustments to the collision avoidance path are needed.

9. The ship collision risk early warning and avoidance decision-making system according to claim 1, characterized in that: It also includes a communication module (5) for information exchange between the target vessel and surrounding vessels; the communication module (5) establishes a communication link with surrounding vessels through a VHF radio (10), and sends the target vessel’s collision avoidance intention and planned path to surrounding vessels, while receiving feedback information from surrounding vessels to optimize the collision avoidance path.

10. The ship collision risk early warning and avoidance decision-making system according to claim 9, characterized in that: The communication module (5) also establishes a remote connection with the shore-based monitoring center through satellite communication equipment (11) to upload the navigation data and collision avoidance operation records of the target vessel to the shore-based monitoring center so that the shore-based monitoring center can monitor and guide the navigation safety of the target vessel in real time.

Images