Intelligent ship driving assistance system based on multi-source information fusion and AI decision

Abstract

The invention relates to the technical field of intelligent ship driving, and discloses an intelligent ship driving assistance system based on multi-source information fusion and AI decision, which comprises a multi-source information acquisition module, an information data processing module, an information fusion processing module, an AI decision calculation module and a man-machine interaction module, according to the ship intelligent driving assistance system based on multi-source information fusion and AI decision making, perception blind areas of scenes with low visibility and the like are eliminated through a multi-source information collection and vision enhancement technology, and a target is accurately recognized; through data processing and three-level fusion, multi-source information is efficiently integrated, and the manual analysis pressure is reduced; the AI decision calculation module provides accurate guarantee for berthing, stranding, collision avoidance and other scenes, and accident risks are reduced; the system can also optimize the route and speed in combination with data such as weather, reduce energy consumption, shorten the navigation time, take into account the safety and economy of ship driving, and greatly improve the intelligent level of shipping.

Description

Technical Field

[0001] This invention relates to the field of intelligent ship driving technology, specifically to an intelligent ship driving assistance system based on multi-source information fusion and AI decision-making. Background Technology

[0002] With increasingly frequent global trade, the shipping industry, as a key link in international trade, handles more than 90% of the world's cargo transportation volume. Its demand for safety and efficiency is becoming more and more urgent. Ship navigation, as a core link in the shipping industry, is directly related to the safety and efficiency of cargo transportation. Traditional ship navigation methods rely primarily on crew experience and manual operation, which exposes numerous limitations when facing complex and ever-changing marine environments and ever-increasing shipping demands. 1. In low visibility (fog, rain, snow) environments, manual visual observation is limited, which can easily lead to delays or omissions in target recognition and make it impossible to identify surrounding ships, obstacles or shoreline outlines in a timely manner. 2. During navigation, information from multiple sources, such as the dynamics of surrounding vessels, water topography, and hydrological and meteorological conditions, is scattered and lacks effective integration. Crew members need to manually integrate information from multiple sources. Manual integration is not only difficult to analyze, but also prone to data misjudgment or omission. As a result, crew members cannot fully and accurately grasp the dynamic and static environmental information around the vessel, cannot detect potential dangers in time, and have limited environmental perception capabilities. 3. During the berthing and unberthing phases, the judgment of the distance between the vessel and the dock and adjacent vessels relies on human experience. Insufficient accuracy can easily lead to collision accidents. When faced with emergencies (such as sudden obstacles or other vessels making illegal turns), the response speed is slow and it is impossible to make correct decisions quickly, thereby increasing the risk of accidents. 4. Traditional navigation route planning is mostly based on fixed routes, making it difficult to optimize routes and speeds by incorporating dynamic factors such as real-time weather and tides. This not only leads to unnecessary risks for ships during navigation but also results in energy waste and increased sailing time, failing to meet the requirements of modern shipping for high efficiency and energy conservation.

[0003] In view of this, and in response to the above problems, we conducted in-depth research and proposed a ship intelligent driving assistance system based on multi-source information fusion and AI decision-making. Summary of the Invention

[0004] The purpose of this invention is to provide a ship intelligent driving assistance system based on multi-source information fusion and AI decision-making, so as to solve the problems mentioned in the background art, such as insufficient perception of the ship driving environment, low efficiency of multi-source information integration, poor accuracy of berthing, unberthing and emergency response, and weak route and speed optimization capabilities.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a ship intelligent driving assistance system based on multi-source information fusion and AI decision-making, comprising a multi-source information acquisition module, an information data processing module, an information fusion processing module, an AI decision-making calculation module, and a human-computer interaction module; The multi-source information acquisition module is used to acquire dynamic and static environmental information during the ship's navigation and berthing / unberthing processes; The information data processing module uses edge computing technology to process the collected information data, performing noise filtering, time synchronization, data calibration, redundancy removal, and format standardization. The information fusion processing module is used to generate a unified ship surrounding situational awareness dataset, enabling complementarity and synergy among different information data. The AI ​​decision computing module is used to analyze and process the situational awareness dataset around the ship and generate maneuvering decision commands for ship navigation and berthing / departure. The human-computer interaction module is used to display the control decision commands in a visual form that conforms to the human eye's observation habits, eliminating blind spots and allowing staff to keep abreast of the ship's surrounding situation.

[0006] The above technical solution facilitates the construction of a closed-loop intelligent driving assistance system for ships that integrates perception, processing, fusion, decision-making, and interaction. This breaks through the limitations of traditional driving that relies on human experience. Through multi-module collaboration, it achieves comprehensive acquisition, accurate processing, and deep fusion of dynamic and static environmental information. Combined with AI algorithms, it generates scientific decision-making instructions and presents them in a human-centered manner, significantly improving the safety, efficiency, and intelligence of ship navigation and berthing. It effectively solves core problems such as perception blind spots, information fragmentation, and decision delays in complex marine environments.

[0007] As a preferred technical solution of the present invention, the multi-source information acquisition module is configured with a visual perception device, an environmental perception device and a ship status perception device. The visual perception device includes a thermal imaging gimbal and a fixed-focus panoramic camera. The thermal imaging gimbal is used to capture target thermal radiation signals in low visibility environments, and the fixed-focus panoramic camera is used to achieve 360-degree undistorted visual image acquisition. The environmental sensing equipment includes a radar, an AIS receiver, a depth sounder, and a hydro-meteorological instrument. The radar is used to detect the distance and orientation of surrounding targets, the AIS receiver is used to acquire the identity and movement data of surrounding vessels, the depth sounder is used to collect water depth data, and the hydro-meteorological instrument is used to monitor wind speed, wind direction, and water temperature. The ship status sensing device includes a GPS positioning module, a ship attitude sensor, and a ship draft sensor. The GPS positioning module is used to obtain the real-time position coordinates of the ship, the ship attitude sensor is used to monitor the ship's heading angle, roll angle, and pitch angle, and the ship draft sensor is used to collect the draft depth of the ship at the bow, stern, bottom, and both sides.

[0008] The above technical solutions facilitate comprehensive, multi-dimensional, and seamless data collection of the ship's surrounding environment and its own status. The combination of a thermal imaging gimbal and a fixed-focus panoramic camera covers low-visibility and 360° visual scenarios, avoiding the limitations of manual observation. The coordinated use of radar, AIS, depth sounders, and other equipment accurately captures target distance, ship identity, water depth, and meteorological data, eliminating perception biases from individual devices. The integration of GPS with attitude and draft sensors allows for real-time monitoring of the ship's position and attitude dynamics, providing complete and reliable raw data support for subsequent decision-making, ensuring the comprehensiveness and accuracy of information collection from the outset.

[0009] As a preferred technical solution of the present invention, the noise filtering of the information data processing module adopts Kalman filtering or wavelet filtering algorithm to filter interference signals such as radar clutter, visual image noise, and false water depth values ​​of the depth sounder; time synchronization adopts network time protocol combined with hardware synchronization triggering technology to unify all sensor data to the same time reference; data calibration is based on the factory parameters of the data acquisition equipment and field calibration data to calibrate the radar distance accuracy, visual equipment distortion parameters, and water depth deviation of the depth sounder; redundancy removal is achieved by removing duplicate information between multiple information data through data correlation analysis; format standardization converts the processed image data, numerical data, and text data into JSON or Protocol Buffers format.

[0010] The above technical solutions effectively address the problems of "high noise, asynchronous operation, poor accuracy, and disordered format" in multi-source data. Kalman filtering or wavelet filtering algorithms effectively filter out interference such as radar clutter and image noise, improving data purity. Network time protocols and hardware synchronization technologies ensure that all sensor data are time-consistent, avoiding decision-making errors caused by time differences. Data calibration based on equipment parameters and on-site calibration corrects accuracy deviations of radar, vision equipment, etc., ensuring data authenticity. Redundancy removal reduces redundant information, and format standardization achieves unified compatibility of different types of data, laying a high-quality data foundation for subsequent information fusion and improving data processing efficiency and reliability.

[0011] As a preferred technical solution of the present invention, the information fusion processing module includes a data layer fusion unit, a feature layer fusion unit, and a decision layer fusion unit; The data layer fusion unit is used to associate and register the processed information data to achieve spatial matching of radar point cloud data, visual image pixel data and AIS text data. The feature layer fusion unit is used to extract key features from each information data, and integrates visual target contour features, radar target motion features and AIS target identity features into a unified situation feature vector through principal component analysis or deep learning feature fusion network. The decision-level fusion unit, based on the feature-level fusion results and combined with domain knowledge and fuzzy reasoning algorithms, comprehensively judges the static environment, dynamic targets and environmental parameters around the ship, and generates a situational awareness dataset that includes target type, location, movement trend and environmental risk level.

[0012] The above technical solutions facilitate the transformation of multi-source data from scattered and independent to collaborative and complementary. Data layer fusion, through spatial matching, associates radar point clouds, visual pixels, and AIS text data, eliminating spatial misalignment of data. Feature layer fusion extracts key features and integrates them into a unified situation vector, avoiding the one-sidedness of single features. Decision layer fusion combines domain knowledge and fuzzy reasoning to generate a situation dataset containing target type and risk level, allowing crew members to shift from "manually integrating multi-source information" to "directly obtaining a unified situation," significantly reducing the difficulty of information analysis, improving the overall cognitive ability of the ship's surrounding environment, and timely identifying potential risks.

[0013] As a preferred technical solution of the present invention, the AI ​​decision computing module includes a visual enhancement module, a grounding warning module, an intelligent navigation and collision avoidance warning module, a berthing distance assistance module, and a global route and speed optimization module; The visual enhancement module uses GAN image enhancement algorithm and YOLOV8 target detection algorithm to perform noise reduction, contrast enhancement and target contour extraction processing on image data, so as to realize all-time and all-weather monitoring and target recognition and ship name extraction in low visibility environment. The grounding warning module accesses tidal data, combines it with water depth data, ship draft data and electronic nautical chart seabed topography data to calculate the safe water depth margin, and promptly monitors the ship's route for grounding and reef contact. When the actual water depth is less than the safe water depth, an early warning is triggered to avoid the risk of the ship running aground. The intelligent navigation and collision avoidance warning module dynamically simulates collision avoidance paths and situations based on surrounding target situations, environmental parameters, and the ship's own status data, and provides the optimal collision avoidance maneuvering scheme. The berthing distance assist module is used to dynamically mark the ship's position and distance assist lines; The global route and speed optimization module integrates ECMWF meteorological and tidal data, and uses an improved Dijkstra algorithm to automatically generate the optimal route and dynamically optimize the speed.

[0014] The above technical solutions facilitate precise and intelligent decision support for core ship operation scenarios; the vision enhancement module solves the problem of target recognition in low visibility conditions, enabling all-weather, 24 / 7 monitoring; the grounding warning module avoids grounding risks in advance through safe water depth calculation and real-time monitoring; the intelligent navigation and collision avoidance module dynamically calculates routes and provides specific heading and speed adjustment schemes to ensure the scientific and operable nature of collision avoidance decisions; the berthing distance assistance module improves berthing and unberthing accuracy and reduces collision accidents through custom distance markings; and the global route optimization module combines meteorological and tidal data to generate optimal routes, reducing energy consumption and travel time while balancing safety and economy.

[0015] As a preferred technical solution of the present invention, the optimal collision avoidance control scheme in the intelligent navigation and collision avoidance warning module includes a heading adjustment angle and a speed adjustment value.

[0016] The above technical solution facilitates clear heading adjustment angles and speed adjustment values, avoids operational deviations caused by differences in crew experience, and ensures that all collision avoidance actions are executable and verifiable.

[0017] As a preferred technical solution of the present invention, the distance auxiliary line in the berthing distance auxiliary module specifically displays the distance between the bow, stern and both sides of the ship and the dock and the ships ahead and behind. At the same time, the marking distance and density of the distance auxiliary line can be customized according to different ship types or user needs.

[0018] The above technical solution facilitates the dynamic marking of the distances between the bow, stern, and sides of the ship and the dock, as well as the distances between the ship and the fore and aft vessels, allowing crew members to intuitively grasp the ship's positional relationships. The function of customizing the marking distance and density is adaptable to different ship types (such as cargo ships and cruise ships) and berthing scenarios (such as busy ports and narrow docks), meeting diverse operational needs, reducing accidents such as dock collisions and ship scrapes caused by misjudgment of distance, and improving the safety and efficiency of berthing and unberthing operations.

[0019] As a preferred technical solution of the present invention, the human-computer interaction module includes a multi-screen linkage display unit, a voice interaction unit, and an emergency control unit; The multi-screen linkage display unit adopts a high-brightness anti-glare LCD screen, which supports split-screen display of situational awareness map, equipment operating status and decision command details; The voice interaction unit supports bilingual (Chinese and English) command recognition and feedback, and broadcasts warning information and decision-making suggestions via voice. The emergency control unit is equipped with a physical emergency pause button and a touch-screen emergency takeover interface. When staff determine that the AI ​​decision-making command is risky, they can immediately interrupt the automatic decision-making process and switch to manual control mode. At the same time, the system automatically saves all the operating data before the interruption for subsequent analysis.

[0020] The above technical solutions facilitate the realization of a "human-machine collaborative, safe and controllable" interaction mode; multi-screen linkage display allows the situation map, equipment status, and decision-making instructions to be presented on separate screens, which conforms to human eye observation habits, eliminates blind spots, and helps crew members quickly obtain key information; bilingual (Chinese and English) voice interaction is not only suitable for international shipping scenarios, but also strengthens early warning and decision prompts through voice broadcasts, avoiding information omissions; physical emergency pause buttons and touch-based emergency controls ensure that manual mode can be switched immediately when AI decision-making poses a risk, while saving operational data for subsequent review, improving intelligence while ensuring safe redundancy of driving control, and balancing the relationship between intelligent assistance and human dominance.

[0021] Compared with the prior art, the beneficial effects of the present invention are: the intelligent driving assistance system for ships based on multi-source information fusion and AI decision-making; 1. Through multi-source information acquisition modules and visual enhancement technology, accurate environmental perception is achieved all-weather and all-time. On the one hand, the thermal imaging gimbal can capture target thermal radiation signals in low-visibility scenarios such as fog, rain, and snow, while the fixed-focus panoramic camera can complete 360° undistorted visual image acquisition. Combined with GAN image enhancement algorithms and YOLOV8 target detection algorithms, noise reduction, contrast enhancement, and target contour extraction can be performed on image data. This not only clearly identifies surrounding ships, obstacles, or shoreline contours but also accurately extracts ship name information. On the other hand, radar, AIS receivers, hydrological and meteorological instruments, and other equipment work together to simultaneously acquire the distance and azimuth of surrounding targets, ship identity and movement data, and environmental parameters such as wind speed, wind direction, and water temperature, completely eliminating blind spots and solving the problem of limited environmental perception in traditional driving. 2. Through the information data processing module and the information fusion processing module, efficient integration and collaboration of multi-source information are achieved. The information data processing module adopts edge computing technology, combined with Kalman filtering or wavelet filtering algorithms to filter interference signals, unifies the time base through network time protocol, calibrates data accuracy according to equipment parameters, removes redundant information and standardizes data to ensure data quality. The information fusion processing module achieves three-level fusion of data layer, feature layer and decision layer. First, it realizes spatial matching of radar, vision and AIS data, then extracts key features and integrates them into a unified situation feature vector. Finally, it combines domain knowledge and fuzzy reasoning to generate a situational awareness dataset containing target type, location, movement trend and environmental risk level. No manual analysis by crew members is required, which greatly reduces the pressure of manual decision-making and improves the accuracy and efficiency of information judgment. 3. Through multiple sub-modules of the AI ​​decision-making and calculation module, precise support is provided for critical scenarios. The berthing distance assistance module can dynamically mark the distances between the bow, stern, and both sides of the ship and the dock, as well as the ships ahead and behind, and the marked distances and densities can be customized, solving the problem of insufficient accuracy in distance judgment during berthing; the grounding warning module integrates tidal data and calculates the safe water depth margin by combining water depth, ship draft, and electronic chart data. When the actual water depth is less than the safe water depth, a warning is triggered to avoid grounding and reefs; the intelligent navigation and collision avoidance warning module dynamically deduces collision avoidance paths based on the surrounding situation, environmental parameters, and the ship's own status, and provides the optimal collision avoidance scheme, including course adjustment angles and speed adjustment values, in accordance with international maritime collision avoidance rules. It responds quickly to emergencies and significantly reduces the risk of collisions and other safety accidents. 4. The global route and speed optimization module integrates ECMWF meteorological and tidal data, and uses an improved Dijkstra algorithm to automatically generate the optimal route while dynamically optimizing speed. This effectively avoids the impact of severe weather, reduces unnecessary detours, lowers ship fuel consumption, and shortens sailing time. Whether it is long-distance transoceanic voyages or short-distance coastal transport, it can improve shipping efficiency and reduce operating costs while ensuring safety, meeting the modern shipping industry's demand for high efficiency and energy conservation, and creating higher economic benefits for shipping companies. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the overall system of the present invention; Figure 2 This is a schematic diagram of the system structure of the multi-source information acquisition module of the present invention; Figure 3 This is a schematic diagram of the information fusion processing module structure of the present invention; Figure 4 This is a schematic diagram of the AI ​​decision computing module structure of the present invention; Figure 5 This is a schematic diagram of the human-computer interaction module structure of the present invention. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0024] Please see Figure 1 - Figure 5 The technical solution of this invention is: a ship intelligent driving assistance system based on multi-source information fusion and AI decision-making, including a multi-source information acquisition module, an information data processing module, an information fusion processing module, an AI decision-making calculation module, and a human-computer interaction module; The multi-source information acquisition module is used to acquire dynamic and static environmental information during the ship's navigation and berthing / unberthing processes; The information data processing module uses edge computing technology to process the collected information data, performing noise filtering, time synchronization, data calibration, redundancy removal, and format standardization. The information fusion processing module is used to generate a unified ship surrounding situational awareness dataset, enabling complementarity and synergy among different information data. The AI ​​decision computing module is used to analyze and process the situational awareness dataset around the ship and generate maneuvering decision commands for ship navigation and berthing / departure. The human-computer interaction module is used to display the control decision commands in a visual form that conforms to the human eye's observation habits, eliminating blind spots and allowing staff to keep abreast of the ship's surrounding situation.

[0025] The multi-source information acquisition module is equipped with visual sensing devices, environmental sensing devices, and ship status sensing devices; The visual perception device includes a thermal imaging gimbal and a fixed-focus panoramic camera. The thermal imaging gimbal is used to capture target thermal radiation signals in low visibility environments, and the fixed-focus panoramic camera is used to achieve 360-degree undistorted visual image acquisition. The environmental sensing equipment includes a radar, an AIS receiver, a depth sounder, and a hydro-meteorological instrument. The radar is used to detect the distance and orientation of surrounding targets, the AIS receiver is used to acquire the identity and movement data of surrounding vessels, the depth sounder is used to collect water depth data, and the hydro-meteorological instrument is used to monitor wind speed, wind direction, and water temperature. The ship status sensing device includes a GPS positioning module, a ship attitude sensor, and a ship draft sensor. The GPS positioning module is used to obtain the real-time position coordinates of the ship, the ship attitude sensor is used to monitor the ship's heading angle, roll angle, and pitch angle, and the ship draft sensor is used to collect the draft depth of the ship at the bow, stern, bottom, and both sides.

[0026] The noise filtering in the information data processing module employs Kalman filtering or wavelet filtering algorithms to filter interference signals from radar clutter, visual image noise, and false depth values ​​from the depth sounder. Time synchronization utilizes a combination of network time protocol and hardware synchronization triggering technology to unify all sensor data to the same time reference. Data calibration is based on the factory parameters of the data acquisition equipment and on-site calibration data to calibrate radar distance accuracy, visual equipment distortion parameters, and depth sounder depth deviation. Redundancy removal removes duplicate information between multiple data sets through data correlation analysis. Format standardization converts processed image data, numerical data, and text data into JSON or Protocol Buffers format.

[0027] The information fusion processing module includes a data layer fusion unit, a feature layer fusion unit, and a decision layer fusion unit; The data layer fusion unit is used to associate and register the processed information data to achieve spatial matching of radar point cloud data, visual image pixel data and AIS text data. The feature layer fusion unit is used to extract key features from each information data, and integrates visual target contour features, radar target motion features and AIS target identity features into a unified situation feature vector through principal component analysis or deep learning feature fusion network. The decision-level fusion unit, based on the feature-level fusion results and combined with domain knowledge and fuzzy reasoning algorithms, comprehensively judges the static environment, dynamic targets and environmental parameters around the ship, and generates a situational awareness dataset that includes target type, location, movement trend and environmental risk level.

[0028] The AI ​​decision computing module includes a vision enhancement module, a grounding warning module, an intelligent navigation and collision avoidance warning module, a berthing distance assistance module, and a global route and speed optimization module. The visual enhancement module uses GAN image enhancement algorithm and YOLOV8 target detection algorithm to perform noise reduction, contrast enhancement and target contour extraction processing on image data, so as to realize all-time and all-weather monitoring and target recognition and ship name extraction in low visibility environment. The grounding warning module accesses tidal data, combines water depth data, ship draft data, and electronic nautical chart seabed topography data to calculate the safe water depth margin (safe water depth margin = ship draft + margin water depth). The margin water depth (i.e., the safe distance that needs to be reserved between the ship's keel and the seabed) is dynamically adjusted according to the seabed soil quality and the ship's tonnage. It monitors the ship's route for grounding and reef contact in a timely manner. When the actual water depth is less than the safe water depth, an early warning is triggered to avoid the risk of the ship running aground. The intelligent navigation and collision avoidance warning module dynamically simulates collision avoidance paths and situations based on surrounding target situations, environmental parameters, and the ship's own status data. It calculates the collision avoidance risk in conjunction with international maritime collision avoidance rules and provides the optimal collision avoidance maneuvering scheme, which includes course adjustment angle and speed adjustment value. The berthing distance assist module is used to dynamically mark the position of the ship and the distance assist line. The distance assist line specifically displays the distance between the bow, stern and both sides of the ship and the dock and the ships in front and behind. At the same time, the marking distance and density of the distance assist line can be customized according to different ship types or user needs. The global route and speed optimization module integrates ECMWF meteorological and tidal data, and uses an improved Dijkstra algorithm to automatically generate the optimal route and dynamically optimize the speed.

[0029] The human-computer interaction module includes a multi-screen linkage display unit, a voice interaction unit, and an emergency control unit; The multi-screen linkage display unit adopts a high-brightness anti-glare LCD screen, which supports split-screen display of situational awareness map, equipment operating status and decision command details; The voice interaction unit supports bilingual (Chinese and English) command recognition and feedback, and broadcasts warning information and decision-making suggestions via voice. The emergency control unit is equipped with a physical emergency pause button and a touch-screen emergency takeover interface. When staff determine that the AI ​​decision-making command is risky, they can immediately interrupt the automatic decision-making process and switch to manual control mode. At the same time, the system automatically saves all the operating data before the interruption for subsequent analysis.

[0030] The contents not described in detail in this specification are existing technologies known to those skilled in the art.

[0031] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A ship intelligent driving assistance system based on multi-source information fusion and AI decision-making, characterized in that, It includes a multi-source information acquisition module, an information data processing module, an information fusion processing module, an AI decision computing module, and a human-computer interaction module; The multi-source information acquisition module is used to acquire dynamic and static environmental information during the ship's navigation and berthing / unberthing processes; The information data processing module uses edge computing technology to process the collected information data, performing noise filtering, time synchronization, data calibration, redundancy removal, and format standardization. The information fusion processing module is used to generate a unified ship surrounding situational awareness dataset, enabling complementarity and synergy among different information data. The AI ​​decision computing module is used to analyze and process the situational awareness dataset around the ship and generate maneuvering decision commands for ship navigation and berthing / departure. The human-computer interaction module is used to display maneuvering decision commands in a visual form that conforms to human eye observation habits, eliminating blind spots and allowing staff to keep abreast of the ship's surrounding situation.

2. The ship intelligent driving assistance system based on multi-source information fusion and AI decision-making according to claim 1, characterized in that, The multi-source information acquisition module is equipped with visual sensing devices, environmental sensing devices, and ship status sensing devices. The visual perception device includes a thermal imaging gimbal and a fixed-focus panoramic camera. The thermal imaging gimbal is used to capture target thermal radiation signals in low visibility environments, and the fixed-focus panoramic camera is used to achieve 360-degree undistorted visual image acquisition. The environmental sensing equipment includes a radar, an AIS receiver, a depth sounder, and a hydro-meteorological instrument. The radar is used to detect the distance and orientation of surrounding targets, the AIS receiver is used to acquire the identity and movement data of surrounding vessels, the depth sounder is used to collect water depth data, and the hydro-meteorological instrument is used to monitor wind speed, wind direction, and water temperature. The ship status sensing device includes a GPS positioning module, a ship attitude sensor, and a ship draft sensor. The GPS positioning module is used to obtain the real-time position coordinates of the ship, the ship attitude sensor is used to monitor the ship's heading angle, roll angle, and pitch angle, and the ship draft sensor is used to collect the draft depth of the ship at the bow, stern, bottom, and both sides.

3. The ship intelligent driving assistance system based on multi-source information fusion and AI decision-making according to claim 2, characterized in that, The noise filtering of the information data processing module adopts Kalman filtering or wavelet filtering algorithms to filter interference signals such as radar clutter, visual image noise, and false water depth values ​​of the depth sounder; the time synchronization adopts network time protocol combined with hardware synchronization triggering technology to unify all sensor data to the same time reference; the data calibration is based on the factory parameters of the data acquisition equipment and the field calibration data to calibrate the radar distance accuracy, the distortion parameters of the visual equipment, and the water depth deviation of the depth sounder. Redundancy removal removes duplicate information between multiple data sets through data association analysis; format standardization converts processed image data, numerical data, and text data into JSON or Protocol Buffers format.

4. The ship intelligent driving assistance system based on multi-source information fusion and AI decision-making according to claim 2, characterized in that, The information fusion processing module includes a data layer fusion unit, a feature layer fusion unit, and a decision layer fusion unit; The data layer fusion unit is used to associate and register the processed information data to achieve spatial matching of radar point cloud data, visual image pixel data and AIS text data. The feature layer fusion unit is used to extract key features from each information data, and integrates visual target contour features, radar target motion features and AIS target identity features into a unified situation feature vector through principal component analysis or deep learning feature fusion network. The decision-level fusion unit, based on the feature-level fusion results and combined with domain knowledge and fuzzy reasoning algorithms, comprehensively judges the static environment, dynamic targets and environmental parameters around the ship, and generates a situational awareness dataset that includes target type, location, movement trend and environmental risk level.

5. A ship intelligent driving assistance system based on multi-source information fusion and AI decision-making according to claim 1, characterized in that, The AI ​​decision-making and computing module includes a vision enhancement module, a grounding warning module, an intelligent navigation and collision avoidance warning module, a berthing distance assistance module, and a global route and speed optimization module. The visual enhancement module uses GAN image enhancement algorithm and YOLOV8 target detection algorithm to perform noise reduction, contrast enhancement and target contour extraction processing on image data, so as to realize all-time and all-weather monitoring and target recognition and ship name extraction in low visibility environment. The grounding warning module accesses tidal data, combines it with water depth data, ship draft data and electronic nautical chart seabed topography data to calculate the safe water depth margin, and promptly monitors the ship's route for grounding and reef contact. When the actual water depth is less than the safe water depth, an early warning is triggered to avoid the risk of the ship running aground. The intelligent navigation and collision avoidance warning module dynamically simulates collision avoidance paths and situations based on surrounding target situations, environmental parameters, and the ship's own status data, and provides the optimal collision avoidance maneuvering scheme. The berthing distance assist module is used to dynamically mark the ship's position and distance assist lines; The global route and speed optimization module integrates ECMWF meteorological and tidal data, and uses an improved Dijkstra algorithm to automatically generate the optimal route and dynamically optimize the speed.

6. A ship intelligent driving assistance system based on multi-source information fusion and AI decision-making according to claim 5, characterized in that, The optimal collision avoidance maneuver scheme in the intelligent navigation and collision avoidance warning module includes heading adjustment angle and speed adjustment value.

7. A ship intelligent driving assistance system based on multi-source information fusion and AI decision-making according to claim 5, characterized in that, The berthing distance assist module displays the distances between the bow, stern, and sides of the ship and the dock, as well as the ships ahead and behind. The distance and density of the assist lines can be customized according to different ship types or user needs.

8. A ship intelligent driving assistance system based on multi-source information fusion and AI decision-making according to claim 1, characterized in that, The human-computer interaction module includes a multi-screen linkage display unit, a voice interaction unit, and an emergency control unit; The multi-screen linkage display unit adopts a high-brightness anti-glare LCD screen, which supports split-screen display of situational awareness map, equipment operating status and decision command details; The voice interaction unit supports bilingual (Chinese and English) command recognition and feedback, and broadcasts warning information and decision-making suggestions via voice. The emergency control unit is equipped with a physical emergency pause button and a touch-screen emergency takeover interface. When staff determine that the AI ​​decision-making command is risky, they can immediately interrupt the automatic decision-making process and switch to manual control mode. At the same time, the system automatically saves all the operating data before the interruption for subsequent analysis.

Images