A method for safety management of navigable waters and a safety manager system

Abstract

The application discloses a kind of navigable water area safety management method and safety housekeeper system, to improve the fine and intelligent level of navigable water area safety management.Method covers data acquisition and preprocessing, intelligent analysis and forecast, daily supervision application, emergency rescue disposal, information interaction and feedback five steps, obtain multi-source data through multi-platform data acquisition system, after cleaning and integration, with data assimilation, machine learning Intelligent algorithms such as achieve hydrological fine prediction and ship anomaly identification, through drift prediction model and intelligent scheduling algorithm optimization search and rescue scheme in emergency scenarios, integrate multiple databases and terminal equipment, support multi-channel information release and double communication mode.The application eliminates monitoring blind area, reduces false and missed report rate, reduces search and rescue range, improves rescue efficiency, adapts to complex navigation environment, and helps to improve shipping safety and port operation efficiency.

Description

Technical Field

[0001] This invention relates to the field of shipping management technology, and in particular to a method and system for safety management of navigable waters. Background Technology

[0002] The rapid development of the global shipping industry has led to a surge in vessel traffic in navigable waterways, making the waterway environment increasingly complex and creating a growing demand for more refined, intelligent, and efficient safety management. Currently, safety management in navigable waterways largely relies on AIS, VTS radar, and hydrological monitoring buoys to build a basic monitoring system, supplemented by drones and unmanned surface vessels in some scenarios. While this has achieved initial multi-source data collection and application, it suffers from numerous technical shortcomings in actual operation, making it difficult to adapt to complex management needs. These problems are particularly prominent in core areas such as data processing and intelligent analysis.

[0003] In the data acquisition and processing stage, the deployment density of existing fixed monitoring equipment is limited, failing to fully cover deep-sea channels, narrow shoals, and other areas. Furthermore, the poor data coordination between mobile and fixed equipment easily leads to monitoring loopholes. This weak data foundation further results in low efficiency in intelligent analysis and emergency response. Hydrodynamic forecasting models rely on fixed parameter settings, lacking dynamic integration with real-time data and autonomous optimization capabilities, resulting in poor forecast accuracy and timeliness. Ship anomaly identification is based solely on simple threshold judgments, failing to incorporate multi-dimensional scenario information, leading to false alarms and missed alarms. Emergency rescue still relies on human experience; drift prediction does not fully integrate multi-field hydrological data such as wind and current fields, resulting in excessively large search and rescue areas and a lack of global coordination in resource allocation, hindering efficient and coordinated response.

[0004] Furthermore, inherent flaws in the system architecture also hinder overall performance improvement. Each functional module operates independently with poor interoperability, preventing deep integration of data and algorithms. The data storage solution struggles to adapt to diverse, heterogeneous data types, including structured and unstructured data, and the access and backup mechanisms are inadequate, further impacting system efficiency. In summary, existing technologies are insufficient to meet the management needs of complex navigation environments, making the development of an integrated intelligent management solution an urgent requirement in the field of shipping safety. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a method for safety management of navigable waters, characterized by comprising the following steps: Step S1, data acquisition and preprocessing: Hydrological environment data, vessel navigation status data, and rescue resource data of navigable waters are acquired through a multi-platform data acquisition system; data cleaning and integration technology is used to standardize the above data to obtain standardized data; Step S2, Intelligent Analysis and Forecasting: The standardized data is input into the hydrodynamic simulation and forecasting model. The forecasting parameters of the hydrodynamic simulation and forecasting model are optimized by combining the data assimilation algorithm and the adaptive parameter optimization algorithm to generate hydrological forecasting results. At the same time, the ship navigation status data in the standardized data is input into the anomaly identification algorithm to identify abnormal ship behavior and generate alarm information. Step S3, daily supervision application, pushes hydrological forecast results, vessel navigation status data and abnormal alarm information to the navigable waters management entity and vessel users through information release channels; Step S4, emergency rescue and disposal: When an accident occurs, the hydrological forecast results are input into the drift prediction model to determine the search area. Relying on the visualization scheduling platform that integrates hydrological, ship, rescue resources and geographic information data, the search and rescue plan is planned and the rescue forces are allocated through intelligent scheduling algorithms. The search and rescue area, rescue force deployment and command instructions are pushed in real time through information release channels. Step S5, Information Interaction and Feedback: Receive user operation instructions and forecast result verification data submitted through the visualization scheduling platform, use the operation instructions and verification data as feedback input, and iteratively optimize and update the parameters of the hydrodynamic simulation forecast model based on the data assimilation algorithm and adaptive parameter optimization algorithm.

[0006] Furthermore, in step S1, the equipment configuration of the multi-platform data acquisition system is as follows: the buoy integrates a GNSS PPP positioning unit and an IMU attitude sensor; the unmanned vessel is equipped with an ADCP Doppler current meter; and the unmanned aerial vehicle is equipped with a radar current sensor. It also includes crowdsourced depth sounding technology based on telemetry and remote reporting, which dynamically collects positioning, depth sounding, attitude, and ship draft data on the navigation trajectory by installing a small automatic depth acquisition device on the ship and transmits it to the data center for processing.

[0007] Furthermore, in step S1, the standardized data includes water depth data corrected for sound velocity and tide level; the tide level correction employs an nth-order polynomial moving average fitting model. In the formula, h(t) is the estimated tide level at time t, and t is the time variable. ε represents the fitting coefficients within the sliding window, n is the polynomial order, and ε is the fitting residual. The tidal level estimation results are dynamically corrected by updating the observed data and fitting coefficients through the sliding window. The sound velocity correction model based on the measured sound velocity profile data of the region is combined to complete the sound velocity correction of the water depth data.

[0008] Furthermore, in step S2, the data assimilation algorithm employs a three-dimensional variational assimilation algorithm, through an objective function. To eliminate simulation bias caused by data heterogeneity, where x is the analytical variable field. Let B be the prior background field, y be the background field error covariance matrix, H be the observation operator, and R be the observation error covariance matrix. The adaptive parameter optimization algorithm adopts a reinforcement learning algorithm. Based on the reinforcement learning framework, the spatial accuracy, statistical error, and tidal level calculation accuracy of the hydrological forecast results are used as feedback reward items to construct an adaptive optimization mechanism for model parameters, so as to realize the iterative update of the velocity field, tidal field, and wave field forecast parameters of the hydrodynamic simulation and forecast model without human intervention.

[0009] Furthermore, in step S4, the drift prediction model is constructed using a particle tracking method based on the Lagrange tracking method. The wind field, flow field, and wave field data from the hydrological forecast results are input, and the model is constructed using a particle position iteration formula. Calculate the real-time position of the target, where Let be the particle position at time t. For wind-induced drift speed, For ocean current speed, The wave residual velocity is Δt, the time step is Δt, and the drift trajectory is corrected by combining the ship's draft and the windward area coefficient of the target falling into the water. The intelligent scheduling algorithm adopts an improved genetic algorithm, with the shortest rescue response time, the highest resource matching degree, and the lowest navigation risk as the multi-objective optimization function. The scheduling path and task allocation scheme are represented by real number encoding. The population is optimized by adaptive crossover operator and mutation operator to generate the main scheme and at least two backup schemes. The scheduling path and task allocation scheme are updated according to the dynamic drift trajectory.

[0010] Furthermore, the information dissemination channels include VHF voice, AIS short messages, APP push, and WeChat mini-program push; the visualization dispatch platform is interconnected through heterogeneous and geographically dispersed networks, and overlays the location and status of rescue vessels, on-site drift trajectories, and command instruction data on the same geographic information layer, enabling multi-terminal synchronous operation and instruction issuance.

[0011] The present invention also discloses a navigation waterway safety management system that implements any of the above methods, including a data acquisition layer, a data processing layer, an application service layer, a terminal interaction layer and a data storage layer; The data acquisition layer is used to perform the collection of hydrological environment data, ship navigation status data and rescue resource data in navigable waters through a multi-platform data acquisition system. The data processing layer includes a built-in data preprocessing module, a hydrodynamic simulation and forecasting module, a ship anomaly identification module, a drift prediction module, and a model parameter optimization module. It employs a hybrid architecture of parallel computing clusters and cloud computing platforms to perform calculations. The application service layer includes a water traffic monitoring platform, a visual dispatching platform, and an information release platform. The water traffic monitoring platform is used to receive and display vessel location, speed, heading, and abnormal alarm information. The visual dispatching platform is used to receive drift prediction results, generate search and rescue plans, and issue dispatching instructions. The information release platform is used to release hydrological forecast results and alarm information to external terminals through information release channels. The terminal interaction layer includes mobile smart terminals and shipborne terminals, which are used to provide multiple types of terminal operation interfaces and dual communication modes, and are used to perform data query, situation display, command and dispatch, information reception and feedback submission operations. The dual communication modes are Beidou short message and mobile network. The data storage layer is used to store data using multiple storage methods and supports hierarchical data backup and recovery.

[0012] Furthermore, the mobile smart terminal of the terminal interaction layer supports mobile phone APP and WeChat mini program; the shipborne terminal includes a real-time hydrological data display module, a ship dynamic query module, an alarm receiving module, a positioning and navigation module, a dynamic water depth visualization module, and a tide calculation module. The positioning and navigation module plans the navigation route by combining electronic charts and real-time hydrological data, the dynamic water depth visualization module displays the channel water depth distribution through color gradients, and the tide calculation module calculates the time window for entering and leaving the port at high tide based on tide level forecast data and ship draft data.

[0013] Furthermore, the data storage layer employs multiple storage methods, including relational databases, non-relational databases, and distributed storage. It utilizes asynchronous multithreading technology to concurrently process multiple data access requests, employs database sharding and table partitioning technology to split data by time and region, and uses SQL index optimization technology to optimize query statements in relational databases. The data storage layer deploys data replicas in different regions, and when a storage node in one region fails, it switches to other regional nodes to provide data services.

[0014] The present invention also discloses a navigation waterway safety steward system, characterized in that it further includes a navigation safety data management system, which is deployed between the data acquisition layer and the data processing layer, and is used to uniformly access and format the multi-source heterogeneous data collected by the data acquisition layer and then output it to the data preprocessing module of the data processing layer. The navigation safety data management system integrates basic geographic information data, port and shipping facility data, hydrological and meteorological data, ship dynamic data and rescue resource data. It includes data access interfaces covering five dimensions: land, water surface, air, space-based and underwater, as well as data cleaning module, data conversion module and data fusion module. The data cleaning module is used to remove duplicate data, correct erroneous data, and mark missing data in the raw data imported through the data access interface; the data conversion module is used to convert the heterogeneous formats and coordinate systems of data from different sources into a standardized format specified by the system; the data fusion module is used to associate and integrate the cleaned and format-converted multi-source data according to geographic coordinates and timestamps, generate a basic dataset with a unified format, and output it to the data preprocessing module of the data processing layer.

[0015] This invention significantly improves the accuracy and efficiency of safety management in navigable waters, addressing multiple shortcomings of existing technologies. Through multi-platform data collection and systematic standardized processing, it eliminates monitoring blind spots, improves data accuracy, and provides reliable support for subsequent analysis. The integration of intelligent algorithms enables autonomous optimization of hydrological forecasts and accurate identification of vessel anomalies, reducing false alarm rates. In emergency situations, precise drift prediction and intelligent dispatching narrow the search and rescue area and improve rescue efficiency. Deep integration of system modules adapts to various data storage needs, ensures stable operation, and fully adapts to complex navigation environments, contributing to improved shipping safety and port operational efficiency. Attached Figure Description

[0016] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0017] Figure 1 A flowchart for safety management methods in navigable waters. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0019] Figure 1 This is a flowchart of a navigable waterway safety management method provided by an embodiment of the present invention. See also... Figure 1 This invention provides a method for safety management of navigable waters, comprising the following steps: Step S1, data acquisition and preprocessing: Hydrological environment data, vessel navigation status data, and rescue resource data of navigable waters are acquired through a multi-platform data acquisition system; data cleaning and integration technology is used to standardize the above data to obtain standardized data; Specifically, buoys are deployed at key nodes in waterways and port entrances to continuously capture basic hydrological parameters such as tidal current direction and velocity, wave period, and height. Buoy data acquisition and processing follow a complete chain: first, precise IGS ephemeris, satellite clock bias, satellite positioning signals, and IMU attitude data are acquired; then, the precise three-dimensional coordinates of the buoy are calculated using a Kalman filter algorithm to correct attitude deviations caused by wind and waves; finally, the raw hydrological data is transmitted to the shore-based data receiving system via BeiDou communication or the internet, and after gross error removal and separation of wave and tidal data, it is pushed to the data center.

[0020] In addition, unmanned surface vessels (USVs) were deployed to conduct mobile surveys along pre-set routes, focusing on covering channel cross-sections and shoal areas not reached by buoys. Unmanned aerial vehicles (UAVs) were deployed to quickly acquire information on wide-area wind speed distribution, sea wave patterns, and ship swarm dynamics from an aerial perspective.

[0021] Vessel dynamic information is cross-verified through multiple sources. VTS radar is used to capture the position of vessels at long distances, AIS data is used to supplement attribute information such as vessel name, draft, and cargo, and CCTV video is used to assist in verifying the actual status of the vessel. Rescue resource data synchronously includes key parameters such as vessel endurance, rescue equipment configuration, drone operating radius, and ROV diving depth, ensuring accurate access during emergencies.

[0022] During the data processing stage, the filtering algorithm prioritizes the removal of outliers caused by wind and wave interference, the gross error removal model filters out extreme deviation data based on the 3σ criterion, and the format normalization technology converts heterogeneous data output from different devices into the JSON standard format, ultimately forming a standardized dataset that is time-series coherent and accurate.

[0023] Step S2, Intelligent Analysis and Forecasting: The standardized data is input into the hydrodynamic simulation and forecasting model. The forecasting parameters of the hydrodynamic simulation and forecasting model are optimized by combining the data assimilation algorithm and the adaptive parameter optimization algorithm to generate hydrological forecasting results. At the same time, the ship navigation status data in the standardized data is input into the anomaly identification algorithm to identify abnormal ship behavior and generate alarm information.

[0024] Specifically, high-performance parallel computing clusters provide powerful computing support for hydrodynamic simulation and forecasting, enabling rapid processing of massive amounts of standardized data and the construction of three-dimensional flow field models by combining basic data such as sea topography and shoreline morphology. Data assimilation algorithms dynamically fuse real-time observation data with model simulation results, while adaptive parameter optimization algorithms automatically adjust key parameters in the model, such as roughness and friction coefficient, by continuously comparing forecast and measured values. This iterative improvement in forecast accuracy can be achieved without manual intervention, ultimately outputting hydrological forecast products such as tidal currents, tidal levels, and waves with a spatial resolution better than 30m for the next 72 hours. In the abnormal behavior identification stage, the abnormal identification algorithm is implemented using machine learning algorithms. First, based on historical track data, a normal navigation mode library is constructed for different ship types, such as cargo ships, passenger ships, and dangerous goods ships, under different hydrological conditions and time periods. Then, by comparing in real time indicators such as the deviation between the ship's position and the planned route, the matching degree between the ship's speed and the channel speed limit, and the magnitude of sudden changes in course, various abnormal situations are accurately identified. For example, scenarios such as a cargo ship staying in a restricted area for more than 10 minutes, a passenger ship's speed suddenly dropping by more than 50%, or a ship continuing to move after its AIS signal is interrupted will all trigger graded alarms.

[0025] Step S3, daily monitoring application, pushes hydrological forecast results, vessel navigation status data and abnormal alarm information to the navigable waters management entity and vessel users through information dissemination channels.

[0026] Specifically, maritime authorities can use a Web GIS PC-based visualization interface to view real-time vessel distribution, spatiotemporal changes in hydrological elements, and lists of abnormal alarms within their jurisdiction. They can initiate remote inquiries or on-site verifications for high-risk vessels. Port management relies on the platform to receive 24-hour tide forecasts and, combined with vessel draft data, plan berthing sequences to avoid grounding due to insufficient tide. Vessel operators can use mobile devices to receive advance warnings of wind and waves, channel construction notices, and other information along their routes. For example, ocean-going cargo ships can adjust their speed based on forecasts to pass through dangerous channels during periods of calm. Inland waterway vessels can view real-time water depth changes and avoid shallow areas. Information delivery is precisely tailored to user type; maritime authorities prioritize receiving abnormal alarms and updates on rescue resources, while vessel operators focus on receiving hydrological forecasts and route-related notices, ensuring efficient and accurate information transmission.

[0027] Step S4, Emergency Rescue and Response: When an accident occurs, the hydrological forecast results are input into the drift prediction model to determine the search area. Relying on the visualization scheduling platform that integrates hydrological, vessel, rescue resource and geographic information data, the search and rescue plan is planned and rescue forces are allocated through intelligent scheduling algorithms. The search and rescue area, rescue force deployment and command instructions are pushed in real time through information release channels.

[0028] Specifically, after an accident occurs, the system automatically extracts key information such as the location, time, and type of vessel. It inputs the wind field, flow field, and wave field data from the hydrological forecast results output in step S2 into the drift prediction model. The system simulates the movement trajectory of the person or vessel that fell into the water using the Lagrange tracking method. Combined with the characteristics of the target itself, such as the windward area coefficient of the person in the water and the windward area coefficient of the vessel wreckage, the drift parameters are corrected, and the system quickly outputs the high-probability search area for the next 24 hours. This area will be dynamically adjusted as the real-time hydrological data is updated.

[0029] The visualization and dispatch platform intuitively presents information such as basic nautical charts of the incident area, shoreline topography, the location and status of nearby rescue vessels, the operating range of drones, and the deployment status of ROVs. The intelligent dispatch algorithm comprehensively considers factors such as the distance between rescue forces and the incident site, sailing speed, and the matching degree of rescue capabilities to generate the optimal search and rescue plan. For example, it prioritizes dispatching the closest vessels with nighttime search and rescue capabilities, combining them with drones for aerial reconnaissance, and using ROVs for underwater target detection. Simultaneously, based on the dynamic trajectory output by the drift prediction model, it adjusts the routes and search areas of rescue vessels in real time to ensure that rescue resources are precisely focused on the target area.

[0030] Step S5, Information Interaction and Feedback: Receive user operation instructions and forecast result verification data submitted through the visualization scheduling platform, use the operation instructions and verification data as feedback input, and iteratively optimize and update the parameters of the hydrodynamic simulation forecast model based on the data assimilation algorithm and adaptive parameter optimization algorithm.

[0031] Specifically, maritime law enforcement personnel can provide feedback on the accuracy of abnormal alarms, such as false alarms and missed alarms, through the terminal. Ship operators can submit data on the deviation between hydrological forecasts and actual conditions, and rescue personnel can evaluate the rationality of search and rescue plans and the efficiency of rescue resource allocation. After classifying and organizing this feedback data, the system inputs it into data assimilation algorithms and adaptive parameter optimization algorithms. For example, for feedback indicating a large deviation in tide level forecasts for a certain sea area, the algorithm will automatically adjust the boundary condition parameters of the hydrodynamic model for that area. For feedback regarding missed alarms of specific types of abnormal ship behavior, the system will optimize the feature extraction dimensions and recognition thresholds of the anomaly identification algorithm. Simultaneously, based on user-reported functional requirements, such as adding customized forecasts for specific waterways and optimizing the priority of rescue command pushes, the system's functional configuration will be improved to enhance its adaptability and practicality in different scenarios.

[0032] Furthermore, in step S1, the equipment configuration of the multi-platform data acquisition system is as follows: the buoy integrates a GNSS PPP positioning unit and an IMU attitude sensor; the unmanned vessel is equipped with an ADCP Doppler current meter; and the unmanned aerial vehicle is equipped with a radar current sensor. It also includes crowdsourced depth sounding technology based on telemetry and remote reporting, which dynamically collects positioning, depth sounding, attitude, and ship draft data on the navigation trajectory by installing a small automatic depth acquisition device on the ship and transmits it to the data center for processing.

[0033] Specifically, the GNSS PPP positioning unit integrated into the buoy can achieve centimeter-level positioning. Combined with the IMU attitude sensor, it corrects measurement errors caused by the buoy's own swaying in real time, ensuring the accuracy of data acquisition for tide levels, waves, etc. The ADCP Doppler current meter on the unmanned surface vessel (USV) can simultaneously measure the velocity and direction of flow at different water layers, accurately capturing the flow field distribution characteristics of the channel cross-section. The radar current sensor on the UAV transmits microwave signals and receives the Doppler frequency shift of reflected water waves to quickly calculate surface flow velocity, making it particularly suitable for areas difficult for USVs to access, such as shoals and narrow channels. Crowdsourced depth sounding technology selects regularly navigating merchant ships and fishing vessels within the jurisdiction as crowdsourcing carriers. Small, low-power automatic depth acquisition devices are installed on the bottom of these vessels. These devices simultaneously collect GNSS positioning data, sonar depth sounding data, attitude sensor data, and ship draft information from the ship's navigation trajectory, transmitting the data in real time to the data center via BeiDou short message service or mobile network. The data center integrates crowdsourced data from multiple vessels, eliminating errors caused by vessel attitude interference and changes in sound speed, and generates high-density, wide-range channel depth data, effectively compensating for the shortcomings of traditional professional depth sounding vessels, such as long measurement cycles and limited coverage.

[0034] Furthermore, in step S1, the standardized data includes water depth data corrected for sound velocity and tide level; the tide level correction employs an nth-order polynomial moving average fitting model. In the formula, h(t) is the estimated tide level at time t, and t is the time variable. ε represents the fitting coefficients within the sliding window, n is the polynomial order, and ε is the fitting residual. The tidal level estimation results are dynamically corrected by updating the observed data and fitting coefficients through the sliding window. The sound velocity correction model based on the measured sound velocity profile data of the region is combined to complete the sound velocity correction of the water depth data.

[0035] Specifically, during the water depth data acquisition process, the speed of sound fluctuates with changes in seawater temperature and salinity, directly affecting the depth measurement accuracy. The sound speed correction model established based on regional measured sound speed profile data establishes the correlation between sound speed and temperature and salinity parameters by collecting temperature and salinity data of different sea areas and different water layers in real time, and automatically corrects the sound speed error in sonar depth measurement data.

[0036] For example, in the tide level correction process, the sliding window length is set to 1-2 hours, with a step size of 0.5 hours. Based on the real-time tide level observation data within the window, the fitting coefficients a0~a0 of the nth-order polynomial are solved using the least squares method. n The value of n is taken from 2 to 4 to adapt to the nonlinear variation characteristics of the tide level, and the fitting residual ε is controlled within ±0.05m.

[0037] As time progresses, the sliding window continuously removes outdated data and incorporates new observations, dynamically updating the fitting coefficients to ensure the real-time accuracy and reliability of the calculated tide level h(t) at time t. The water depth data, after dual correction for sound velocity and tide level, accurately reflects actual water depth changes in the channel. For example, in port siltation areas, it can promptly capture trends of decreasing water depth, providing data support for channel dredging planning. During ship navigation, it can assist navigators in determining whether sufficient water depth is available, avoiding the risk of grounding.

[0038] Ultimately, this improves the accuracy of water depth data by more than 30% compared to traditional methods, providing high-precision input data for the hydrodynamic simulation and forecasting model in step S2, while also supporting accurate decision-making for ship navigation safety assessment and subsequent emergency rescue.

[0039] Furthermore, in step S2, the data assimilation algorithm employs a three-dimensional variational assimilation algorithm, through an objective function. To eliminate simulation bias caused by data heterogeneity, where x is the analytical variable field. Let B be the prior background field, y be the background field error covariance matrix, H be the observation operator, and R be the observation error covariance matrix. The adaptive parameter optimization algorithm adopts a reinforcement learning algorithm. Based on the reinforcement learning framework, the spatial accuracy, statistical error, and tidal level calculation accuracy of the hydrological forecast results are used as feedback reward items to construct an adaptive optimization mechanism for model parameters, so as to realize the iterative update of the velocity field, tidal field, and wave field forecast parameters of the hydrodynamic simulation and forecast model without human intervention.

[0040] Specifically, in the three-dimensional variational assimilation algorithm, the prior background field The background field error covariance matrix B, generated from the initial simulation results of the hydrodynamic model, is derived from the statistical analysis of historical simulation errors and is used to quantify the uncertainty of the prior background field. The observation operator H transforms the analytical variable field x into simulated values ​​of the same dimension as the observed data y, enabling comparison between the model simulation results and the observed data. The observation error covariance matrix R is constructed based on the accuracy indicators of different observation equipment, such as the accuracy of buoy tide level measurement and the error of UAV wind speed measurement, reflecting the reliability of the observed data. The analytical variable field x and the prior background field are calculated using the objective function J(x). By taking into account the comprehensive deviation of the observation data y, and finding the x value that minimizes J(x), the optimal analysis field after fusing the observation data can be obtained. This field serves as the initial field for the hydrodynamic simulation and prediction model, effectively eliminating the deviation caused by the heterogeneity of multi-source data.

[0041] Furthermore, in the adaptive parameter optimization algorithm, the agent of the reinforcement learning framework uses key parameters of the hydrodynamic model, such as roughness and bottom friction coefficient, as its action space, and uses the spatial accuracy, statistical error, and tidal level calculation accuracy of the hydrological forecast as reward items. The spatial accuracy can be the spatial matching degree between the tidal field forecast and the measured value, and the statistical error can be the mean value of the tidal level forecast error. These are not listed in detail in this embodiment. When the forecast accuracy meets the standard, a positive reward is given; when the accuracy does not meet the standard, a negative reward is given. Through continuous iterative learning, the agent automatically adjusts the model parameters to achieve autonomous optimization of the forecast accuracy. The entire process does not require human intervention.

[0042] Furthermore, in step S4, the drift prediction model is constructed using a particle tracking method based on the Lagrange tracking method. The wind field, flow field, and wave field data from the hydrological forecast results are input, and the model is constructed using a particle position iteration formula. Calculate the real-time position of the target, where Let be the particle position at time t. For wind-induced drift speed, For ocean current speed, The wave residual velocity is Δt, the time step is Δt, and the drift trajectory is corrected by combining the ship's draft and the windward area coefficient of the target falling into the water. The intelligent scheduling algorithm adopts an improved genetic algorithm, with the shortest rescue response time, the highest resource matching degree, and the lowest navigation risk as the multi-objective optimization function. The scheduling path and task allocation scheme are represented by real number encoding. The population is optimized by adaptive crossover operator and mutation operator to generate the main scheme and at least two backup schemes. The scheduling path and task allocation scheme are updated according to the dynamic drift trajectory.

[0043] Specifically, in the Lagrange tracking method, the particle tracking model discretizes the target or wrecked vessel into a series of particles for simulation. Among them, u a u is calculated by combining wind field data with the target's windward area coefficient. s The flow field velocity provided by hydrodynamic forecasting is directly adopted, while u v The time step Δt is derived from wave parameters such as wave height and period using empirical formulas. It is typically set between 0.5 and 1 hour to balance computational efficiency and trajectory accuracy.

[0044] The model will also make corresponding adjustments for different types of targets. For example, for wrecked ships, the model will analyze the flow field effect on the underwater parts in combination with its draft and adjust the drift speed according to the windward area of ​​the wreckage; for people who have fallen into the water, the model will consider factors such as whether they are wearing life jackets and seawater temperature to adjust and evaluate the wind-induced drift coefficient and survival time.

[0045] In the intelligent scheduling algorithm, an improved genetic algorithm is used for real-number encoding, where each encoded bit corresponds to a scheduling parameter for the rescue vessel, such as departure time, key nodes on the route, and operation duration. The multi-objective optimization function then sets weights to comprehensively weigh three indicators—rescue response time, resource matching degree, and navigation risk—to optimize the overall scheduling effect.

[0046] The formula for calculating the multi-objective optimization function is as follows: In the formula, To optimize the objective function for multiple objectives, These are weighted coefficients for rescue response time, resource matching degree, and navigation risk, respectively, to meet the following requirements. , For rescue response time, This represents the resource matching degree, with a value ranging from 0 to 1. This represents the navigation risk value.

[0047] In the algorithm design, an adaptive crossover mechanism is introduced. The adaptive crossover operator dynamically adjusts the crossover probability based on the number of population iterations. The mutation operator also adopts an adaptive strategy to adjust the mutation probability accordingly, thereby reducing the risk of getting trapped in a local optimum.

[0048] During the iterative calculation process, the system receives the target drift trajectory obtained based on the Lagrange tracking method in real time, and dynamically adjusts the navigation path and search area of ​​the rescue vessel accordingly. In the overall plan, the main plan prioritizes improving rescue efficiency and success rate; at the same time, a backup plan is designed in advance to deal with emergencies, such as ship failure or sudden changes in sea conditions, to ensure that the rescue operation can be carried out continuously and stably.

[0049] Furthermore, the information dissemination channels include VHF voice, AIS short messages, APP push notifications, and WeChat mini-program push notifications. The visualization dispatch platform is interconnected through heterogeneous and geographically dispersed networks, overlaying and displaying the location and status of rescue vessels, on-site drift trajectories, and command instruction data on the same geographic information layer, enabling multi-terminal synchronous operation and instruction issuance.

[0050] Specifically, the system achieves hierarchical information transmission and coverage through multiple communication methods. VHF voice is primarily used for real-time communication between maritime authorities and vessels, such as issuing emergency warnings or dispatch instructions, playing a crucial role, especially in situations where mobile network coverage is unavailable in the open ocean. AIS short messages, relying on shipboard AIS equipment, directly send brief warning messages or channel restrictions to the bridge, ensuring that information is received by the crew immediately.

[0051] For users on shore or in management positions, such as ship operators and port managers, the system pushes information through an app and WeChat mini-program, providing more detailed content, including hydrological forecasts, ship dynamic lists, and rescue progress, and supports various presentation formats such as text, audio, and video.

[0052] In terms of dispatching and coordination, the visual dispatching platform combines multiple networks such as BeiDou short message service, mobile communication, and satellite communication to achieve data interconnection between different regions. For example, rescue vessels located at sea can transmit real-time location and on-site video via satellite link, while the maritime command center can access the platform via mobile network to monitor and issue instructions.

[0053] Furthermore, the platform supports simultaneous online operation by multiple users, allowing various rescue units to share real-time situational information. Command center dispatch instructions can be rapidly distributed to relevant forces through the platform, thus forming a unified command and coordinated emergency response mechanism.

[0054] This invention also discloses a navigational waterway safety management system for implementing the above-described method, characterized in that, It includes a data acquisition layer, a data processing layer, an application service layer, a terminal interaction layer, and a data storage layer; The data acquisition layer is used to perform the collection of hydrological environment data, ship navigation status data and rescue resource data in navigable waters through a multi-platform data acquisition system. The data processing layer includes a built-in data preprocessing module, a hydrodynamic simulation and forecasting module, a ship anomaly identification module, a drift prediction module, and a model parameter optimization module. It employs a hybrid architecture of parallel computing clusters and cloud computing platforms to perform calculations. The application service layer includes a water traffic monitoring platform, a visual dispatching platform, and an information release platform. The water traffic monitoring platform is used to receive and display vessel location, speed, heading, and abnormal alarm information. The visual dispatching platform is used to receive drift prediction results, generate search and rescue plans, and issue dispatching instructions. The information release platform is used to release hydrological forecast results and alarm information to external terminals through information release channels. The terminal interaction layer includes mobile smart terminals and shipborne terminals, which are used to provide multiple types of terminal operation interfaces and dual communication modes, and are used to perform data query, situation display, command and dispatch, information reception and feedback submission operations. The dual communication modes are Beidou short message and mobile network. The data storage layer is used to store data using multiple storage methods and supports hierarchical data backup and recovery.

[0055] Specifically, the equipment in the data acquisition layer is deployed according to the idea of ​​combining fixed and mobile equipment, and coordinating air, surface and underwater operations. Buoys are evenly distributed according to the waterway, unmanned ships and drones are flexibly dispatched according to the mission, and crowdsourced depth sounding equipment covers ships on major routes, thereby building a comprehensive data acquisition system.

[0056] At the data processing layer, an architecture combining high-performance parallel computing clusters and cloud computing is adopted. The former mainly undertakes computationally intensive and complex tasks such as hydrodynamic simulation and forecasting, while the latter is responsible for data preprocessing and model parameter optimization. Through cross-platform technology, data can flow smoothly between different devices and systems. The application service layer consists of three platforms operating collaboratively. For example, once the water traffic monitoring platform detects an anomaly, it can automatically link with the visual dispatching platform for pre-response, while the information publishing platform will also push relevant content to the corresponding users.

[0057] In terms of terminal interaction layer, the WebGIS PC terminal focuses on professional applications, supporting functions such as multi-layer overlay, data export, and report generation; while the mobile terminal emphasizes lightweight convenience, suitable for information query and feedback in outdoor or maritime scenarios, and ensures that critical data transmission can still be completed in the absence of signal at sea through dual communication modes.

[0058] The data storage layer selects appropriate methods based on different data types. For example, structured data such as ship static information and rescue resource parameters are stored in relational databases, while unstructured data such as video and audio are stored in non-relational databases and distributed storage systems. At the same time, a hierarchical backup mechanism is used to set different backup frequencies and storage locations according to the importance of the data in order to reduce the risk of data loss due to hardware failure or natural disasters.

[0059] The navigation waterway safety management system includes a mobile smart terminal that supports mobile APP and WeChat mini-program; the shipborne terminal includes a real-time hydrological data display module, a ship dynamic query module, an alarm receiving module, a positioning and navigation module, a dynamic water depth visualization module, and a tide calculation module. The positioning and navigation module combines electronic charts and real-time hydrological data to plan navigation routes, the dynamic water depth visualization module displays the channel water depth distribution through color gradients, and the tide calculation module calculates the time window for entering and leaving the port at high tide based on tide level forecast data and ship draft data.

[0060] Specifically, the mobile app supports both iOS and Android systems. Users can select the waterways, vessels, or hydrological parameters they wish to focus on and receive personalized information push notifications. Feedback can also be submitted via voice. The WeChat mini-program requires no download or installation and is more suitable for temporary use, allowing users to quickly access basic information such as hydrological forecasts and vessel status. The shipborne terminal adopts an industrial-grade design, possessing waterproof, shockproof, and electromagnetic interference resistance capabilities. Its real-time hydrological data display function visually presents changes in currents, tide levels, and wind speeds around the vessel; the vessel status query function allows users to view the location, heading, and speed of nearby vessels, helping to reduce the risk of collisions.

[0061] In terms of navigation, the system combines electronic nautical charts with real-time hydrological information to provide vessels with more reasonable navigation path suggestions. The dynamic water depth visualization function displays the water depth distribution in the channel through color changes; for example, red indicates risky areas such as shoals, while green marks relatively safe areas. In addition, the tide calculation function integrates tide forecasts and vessel draft to automatically provide the optimal time window for entering and leaving port, helping navigators to plan their voyages more efficiently.

[0062] Furthermore, the data storage layer employs multiple storage methods, including relational databases, non-relational databases, and distributed storage; the relational database is PostgreSQL or Oracle, the non-relational database is NoSQL, and the distributed storage is HDFS, HBase, MongoDB, or ES; the data storage layer uses asynchronous multi-threading technology to concurrently process multiple data access requests, uses database sharding and table partitioning technology to split data by time and region dimensions, and uses SQL index optimization technology to optimize the query statements of the relational database; the data storage layer deploys data replicas in different regions, and switches to other regional nodes to provide data services when a storage node in one region fails.

[0063] Specifically, in terms of data storage solutions, the system comprehensively adopts multiple databases and distributed technologies: PostgreSQL or Oracle is selected for relational databases, NoSQL system is adopted for non-relational databases, and distributed storage and retrieval tools such as HDFS, HBase, MongoDB and Elasticsearch are combined.

[0064] PostgreSQL is primarily used to manage structured data such as ship static information, waterway basic data, and rescue resource parameters, and it excels particularly in geospatial data processing, efficiently supporting the management and analysis of coordinate-based information. Oracle, on the other hand, is better suited for handling large-scale historical hydrological data and ship tracking data, exhibiting strong stability under high-concurrency access and transaction processing scenarios.

[0065] For unstructured data, the system stores it using NoSQL databases, such as CCTV video and ship voice communication records. This type of data does not rely on a fixed format and can flexibly adapt to diverse data types. Meanwhile, HDFS is primarily used to store massive datasets, such as hydrological simulation results and satellite remote sensing imagery, and supports distributed parallel processing.

[0066] In real-time data processing, HBase is used to store data with obvious time-series characteristics, such as buoy observation data and ship dynamic information, enabling efficient writing and fast querying; MongoDB is suitable for storing semi-structured data such as user feedback and system logs, and supports flexible conditional searches. Elasticsearch mainly handles full-text search functions, making it convenient for users to quickly find ship information, accident records, and hydrological forecast reports.

[0067] In terms of performance optimization, the system introduces an asynchronous multi-threading mechanism, enabling multiple data access requests to be processed in parallel, avoiding the blockage of overall performance by a single request. Simultaneously, through database sharding and table partitioning strategies, massive amounts of data are split according to time or region, reducing the pressure on a single database and thus improving query efficiency. Combined with SQL optimization techniques, such as query statement optimization and index design, the access performance of the relational database is further improved. Furthermore, the system enhances reliability by deploying data replicas across regions. If a storage node in one region fails, it can quickly switch to other nodes, ensuring the continuous and stable operation of data services.

[0068] The navigation waterway safety steward system also includes a navigation safety data management system. The navigation safety data management system is deployed between the data acquisition layer and the data processing layer. It is used to uniformly access and format the multi-source heterogeneous data collected by the data acquisition layer and then output it to the data preprocessing module of the data processing layer. The navigation safety data management system integrates basic geographic information data, port and shipping facility data, hydrological and meteorological data, ship dynamic data and rescue resource data. It includes data access interfaces covering five dimensions: land, water surface, air, space-based and underwater, as well as data cleaning module, data conversion module and data fusion module. The data cleaning module is used to remove duplicate data, correct erroneous data, and mark missing data in the raw data imported through the data access interface; the data conversion module is used to convert the heterogeneous formats and coordinate systems of data from different sources into a standardized format specified by the system; the data fusion module is used to associate and integrate the cleaned and format-converted multi-source data according to geographic coordinates and timestamps, generate a basic dataset with a unified format, and output it to the data preprocessing module of the data processing layer.

[0069] It is understood that the system and units provided in this embodiment can also be used to implement the steps in the methods provided in other embodiments of the present invention.

[0070] The present invention also provides a computer device. The computer device is manifested in the form of a general-purpose computing device. The components of the computer device may include, but are not limited to: one or more processors or processing units, system memory, and buses connecting different system components.

[0071] Computer devices typically include a variety of computer system-readable media. These media can be any available media that can be accessed by a computer device, including volatile and non-volatile media, and removable and non-removable media.

[0072] The system memory may include a computer system readable medium in the form of volatile memory, and the memory may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of the embodiments of the present invention.

[0073] The processing unit executes various functional applications and data processing by running programs stored in the system memory, such as implementing the methods provided in other embodiments of the present invention.

[0074] The present invention also provides a storage medium containing computer-executable instructions and storing a computer program thereon, which, when executed by a processor, implements the methods provided in other embodiments of the present invention.

[0075] Note that the above description is merely a preferred embodiment of the present invention and the technical principles employed. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of the present invention, the scope of which is determined by the scope of the appended claims.

Claims

1. A method for safety management of navigable waters, characterized in that, Includes the following steps: Step S1, data acquisition and preprocessing: Hydrological environment data, vessel navigation status data, and rescue resource data of navigable waters are acquired through a multi-platform data acquisition system; data cleaning and integration technology is used to standardize the above data to obtain standardized data; Step S2, Intelligent Analysis and Forecasting: The standardized data is input into the hydrodynamic simulation and forecasting model. The forecasting parameters of the hydrodynamic simulation and forecasting model are optimized by combining the data assimilation algorithm and the adaptive parameter optimization algorithm to generate hydrological forecasting results. At the same time, the ship navigation status data in the standardized data is input into the anomaly identification algorithm to identify abnormal ship behavior and generate alarm information. Step S3, daily monitoring application, pushes the hydrological forecast results, vessel navigation status data and abnormal alarm information to the navigable waters management entity and vessel users through information release channels; Step S4, emergency rescue and disposal: When an accident occurs, the hydrological forecast results are input into the drift prediction model to determine the search area. Relying on the visualization scheduling platform that integrates hydrological, ship, rescue resources and geographic information data, the search and rescue plan is planned and the rescue forces are allocated through intelligent scheduling algorithms. The search and rescue area, rescue force deployment and command instructions are pushed in real time through information release channels. Step S5, Information Interaction and Feedback: Receive user operation instructions and forecast result verification data submitted through the visualization scheduling platform, use the operation instructions and verification data as feedback input, and iteratively optimize and update the parameters of the hydrodynamic simulation forecast model based on the data assimilation algorithm and adaptive parameter optimization algorithm.

2. The method for safety management of navigable waters according to claim 1, characterized in that, In step S1, the multi-platform data acquisition system consists of buoys, shipborne equipment, unmanned surface vessels (USVs), unmanned aerial vehicles (UAVs), and satellite remote sensing. The hydrological environment data includes tide level, current velocity, water depth, wind speed, and wind direction data. The ship navigation status data includes ship position, speed, heading, and draft data. The rescue resource data includes the position and status data of rescue vessels. The specific equipment configuration of the multi-platform data acquisition system is as follows: the buoy integrates a GNSS PPP positioning unit and an IMU attitude sensor; the USV is equipped with an ADCP Doppler current meter; and the UAV is equipped with a radar current sensor. The multi-platform data acquisition system also includes crowdsourced depth sounding technology based on telemetry and remote reporting, which dynamically collects positioning, depth, attitude, and ship draft data along the navigation trajectory by installing a small automatic depth acquisition device on the ship and transmits it to the data center for processing.

3. The method for safety management of navigable waters according to claim 1, characterized in that, In step S1, the standardized data includes water depth data corrected for sound velocity and tide level; the tide level correction uses an nth-order polynomial sliding fitting model. In the formula, h(t) is the estimated tide level at time t, and t is the time variable. ε represents the fitting coefficients within the sliding window, n is the polynomial order, and ε is the fitting residual. The tidal level estimation results are dynamically corrected by updating the observed data and fitting coefficients through the sliding window. The sound velocity correction model based on the measured sound velocity profile data of the region is combined to complete the sound velocity correction of the water depth data.

4. The method for safety management of navigable waters according to claim 1, characterized in that, In step S2, the data assimilation algorithm employs a three-dimensional variational assimilation algorithm, through the objective function. To eliminate simulation bias caused by data heterogeneity, where x is the analytical variable field. Let B be the prior background field, B be the background field error covariance matrix, y be the standardized multi-source observation data, H be the observation operator, and R be the observation error covariance matrix. The adaptive parameter optimization algorithm adopts a reinforcement learning algorithm. Based on the reinforcement learning framework, it uses the spatial accuracy, statistical error and tidal level calculation accuracy of the hydrological forecast results as feedback reward items to construct an adaptive optimization mechanism for model parameters, so as to realize the iterative update of the velocity field, tidal field and wave field forecast parameters of the hydrodynamic simulation forecast model without human intervention.

5. The method for safety management of navigable waters according to claim 1, characterized in that, In step S4, the drift prediction model is constructed using a particle tracking method based on the Lagrange tracking method. The wind field, flow field, and wave field data from the hydrological forecast results are input, and the model is constructed using a particle position iteration formula. Calculate the real-time position of the target, where Let be the particle position at time t. For wind-induced drift speed, For ocean current speed, The wave residual velocity is Δt, the time step is Δt, and the drift trajectory is corrected by combining the ship's draft and the windward area coefficient of the target falling into the water. The intelligent scheduling algorithm adopts an improved genetic algorithm, with the shortest rescue response time, the highest resource matching degree, and the lowest navigation risk as the multi-objective optimization function. The scheduling path and task allocation scheme are represented by real number encoding. The population is optimized by adaptive crossover operator and mutation operator to generate the main scheme and at least two backup schemes. The scheduling path and task allocation scheme are updated according to the dynamic drift trajectory.

6. The method for safety management of navigable waters according to claim 1, characterized in that, The information dissemination channels include VHF voice, AIS short messages, APP push and WeChat mini-program push; the navigation waterway management entities include navigation waterway management departments and rescue units, and the vessel users include vessel drivers; the visualization dispatch platform is interconnected through heterogeneous and geographically dispersed networks, and overlays the location and status of rescue vessels, on-site drift trajectories and command instruction data on the same geographic information layer, enabling multi-terminal synchronous operation and instruction issuance.

7. A navigational waterway safety management system implementing the method of any one of claims 1-6, characterized in that, It includes a data acquisition layer, a data processing layer, an application service layer, a terminal interaction layer, and a data storage layer; The data acquisition layer is used to perform the collection of hydrological environment data, ship navigation status data and rescue resource data in navigable waters through a multi-platform data acquisition system. The data processing layer includes a built-in data preprocessing module, a hydrodynamic simulation and forecasting module, a ship anomaly identification module, a drift prediction module, and a model parameter optimization module. It employs a hybrid architecture of parallel computing clusters and cloud computing platforms to perform calculations. The application service layer includes a water traffic monitoring platform, a visual dispatching platform, and an information release platform; the water traffic monitoring platform is used to receive and display vessel location, speed, heading, and abnormal alarm information. The visual dispatching platform is used to receive drift prediction results, generate search and rescue plans, and issue dispatching instructions; the information release platform is used to release hydrological forecast results and alarm information to external terminals through information release channels. The terminal interaction layer includes mobile smart terminals and shipborne terminals, which are used to provide multiple types of terminal operation interfaces and dual communication modes, and are used to perform data query, situation display, command and dispatch, information reception and feedback submission operations. The dual communication modes are Beidou short message and mobile network. The data storage layer is used to store data using multiple storage methods and supports hierarchical data backup and recovery.

8. The navigable waterway safety management system according to claim 7, characterized in that, The mobile smart terminal supports mobile APP and WeChat mini program; the shipborne terminal includes a real-time hydrological data display module, a ship dynamic query module, an alarm receiving module, a positioning and navigation module, a dynamic water depth visualization module, and a tide calculation module. The positioning and navigation module plans the navigation route by combining electronic charts and real-time hydrological data, the dynamic water depth visualization module displays the channel water depth distribution through color gradients, and the tide calculation module calculates the time window for entering and leaving the port at high tide based on tide level forecast data and ship draft data.

9. The navigable waterway safety management system according to claim 7, characterized in that, The data storage layer employs multiple storage methods, including relational databases, non-relational databases, and distributed storage. It uses asynchronous multithreading to concurrently process multiple data access requests, sharding to partition data by time and region, and SQL index optimization to optimize relational database queries. The data storage layer deploys data replicas in different regions, and switches to other regions to provide data services when a storage node in one region fails.

10. The navigable waterway safety management system according to claim 7, characterized in that, It also includes a general aviation safety data management system, which is deployed between the data acquisition layer and the data processing layer. It is used to uniformly access and format the multi-source heterogeneous data collected by the data acquisition layer and then output it to the data preprocessing module of the data processing layer. The navigation safety data management system integrates basic geographic information data, port and shipping facility data, hydrological and meteorological data, ship dynamic data and rescue resource data. It includes data access interfaces covering five dimensions: land, water surface, air, space-based and underwater, as well as data cleaning module, data conversion module and data fusion module. The data cleaning module is used to remove duplicate data, correct erroneous data, and mark missing data in the raw data imported through the data access interface; the data conversion module is used to convert the heterogeneous formats and coordinate systems of data from different sources into a standardized format specified by the system; the data fusion module is used to associate and integrate the cleaned and format-converted multi-source data according to geographic coordinates and timestamps, generate a basic dataset with a unified format, and output it to the data preprocessing module of the data processing layer.

Images