Ship berthing and departing early warning method and system

Abstract

The invention provides a ship berthing and leaving early warning method and system, and relates to the field of port ship management, and the method comprises the steps: obtaining ship position data, converting the ship position data into distribution density, extracting time sequence features, building a motion prediction relation, calculating a congestion risk coefficient, constructing water area risk distribution, and carrying out the route crossing calculation to obtain an avoidance demand value. And determining an avoidance sequence, calculating an avoidance route and a navigation time window, and generating time-sharing avoidance early warning information containing an early warning level and an avoidance instruction. The ship congestion risk can be predicted in advance, the navigation time is reasonably planned, and the navigation safety and efficiency of the port water area are effectively improved.

Description

Technical Field

[0001] This invention relates to port vessel management technology, and more particularly to a method and system for early warning of vessel berthing and departure. Background Technology

[0002] In port waters, vessel berthing and departure operations are frequent and critical aspects of port operations. With the continuous growth of port throughput, the number and density of vessels in the waters are gradually increasing, significantly raising the risk of mutual interference and collisions. Existing berthing and departure control mainly relies on manual scheduling and experience-based judgment, often suffering from information lag, untimely warnings, and a lack of scientific rigor in avoidance measures. The movement of vessels in port waters is uncertain and dynamic; relying solely on vessel position information is insufficient to reflect the overall distribution of traffic risks in a timely manner. Most existing route avoidance methods do not fully consider navigation constraints and time factors, leading to reduced berthing and departure efficiency and potentially causing congestion and safety hazards. Therefore, there is an urgent need for a method that, based on vessel distribution and movement prediction, combined with risk assessment and dynamic planning, can achieve early warning and reasonable avoidance of berthing and departure vessels, thereby improving the safety and navigation efficiency of port operations. Summary of the Invention

[0003] This invention provides a method and system for early warning of ship berthing and departure, which can solve the problems in the prior art.

[0004] A first aspect of the present invention provides a method for early warning of ship berthing and departure, comprising: Obtain vessel location data within the port waters, and convert the vessel location data into vessel distribution density based on gridded water information; Temporal features of ship distribution density are extracted, ship motion prediction relationships are established, and congestion risk coefficients are calculated based on ship motion prediction relationships. Based on the congestion risk coefficient, a risk distribution of port waters is constructed. The planned routes of waiting and departing vessels are cross-calculated with the risk distribution of port waters to obtain the avoidance demand value. The order of ships to avoid obstacles is determined based on the obstacle avoidance demand value. The obstacle avoidance route is calculated using a dynamic programming method with navigation constraints. The navigation time window is then calculated based on the obstacle avoidance route. The navigation passage section is determined according to the navigation time window, and time-sharing avoidance warning information is generated within the navigation passage section. The time-sharing avoidance warning information includes the warning level and avoidance instructions.

[0005] Obtain vessel location data within the port waters, and convert the vessel location data into vessel distribution density based on gridded water information, including: Obtain the geographical boundary location data of the port waters, and divide the port waters into grids of equal size according to the minimum safe distance between ships to obtain gridded waters location data; Obtain vessel position data and vessel weight values ​​within the gridded water area; A multi-scale sliding window is constructed with a preset time interval, and the time decay characteristics of ship position data are calculated at different time scales to obtain a multi-dimensional time feature vector. The ship position data is decomposed based on the multidimensional time feature vector to extract the ship motion trend features, and the density correlation between the ship and the grid is calculated by combining the gridded water position data. A ship spatiotemporal state transition matrix is ​​constructed using density correlation. The density distribution at each time scale is iterated, and the confidence level of the density distribution is calculated based on the state iteration results. The density distributions at different time scales are then fused based on the confidence level to obtain the ship distribution density.

[0006] Temporal features of ship distribution density are extracted to establish ship motion prediction relationships. Based on these relationships, congestion risk coefficients are calculated, including: Arrange the ship distribution density data in chronological order and calculate the change in ship distribution density. The ship distribution density change values ​​are divided into regions, the change trend of ship distribution density in each region is analyzed, and the change trend is compared with the preset navigation density threshold to determine the density fluctuation point. Based on density fluctuation points, the density influence range is constructed, the spatiotemporal evolution characteristics of ship distribution density within the density influence range are extracted, the topological structure of density field changes is established, and the associated paths and transmission intensity of density transmission are identified based on the topological structure. A density propagation network is constructed using associated paths and transmission intensity. The direction of density propagation is identified, the rate of density growth is calculated, and the density diffusion pattern is predicted by combining density transmission efficiency. The ship motion prediction relationship is obtained, and the ship motion state is evaluated based on the ship motion prediction relationship to obtain the congestion risk coefficient.

[0007] Based on density fluctuation points, a density influence range is constructed. The spatiotemporal evolution characteristics of ship distribution density within this range are extracted. A topological structure of density field changes is established, and the associated paths and intensity of density transmission are identified based on this topological structure. Spatial clustering of density fluctuation points yields the core region of density fluctuation. A density influence matrix is ​​constructed based on the core region of density fluctuation, and the range of density influence is determined according to the spatial distribution of the density influence matrix. A spatiotemporal tensor representation matrix is ​​constructed within the density influence range. The spatiotemporal tensor representation matrix is ​​decomposed to obtain the principal direction vector and secondary direction vector of the ship distribution density. The spatiotemporal evolution characteristics of the ship distribution density are extracted based on the principal direction vector and secondary direction vector. The spatiotemporal evolution characteristics of ship distribution density are discretized into a simple complex sequence, and the continuous homology characteristics of the simple complex sequence are calculated. Based on the continuous homology characteristics, the topological structure of density field changes is obtained. A topological connectivity graph is obtained by performing connectivity analysis on the topological structure of the density field change. The set of edges in the topological connectivity graph is extracted as the associated paths of density transmission, and the transmission strength of the associated paths is obtained by calculating the connectivity of each edge in the topological connectivity graph.

[0008] Based on the congestion risk coefficient, a risk distribution for port waters is constructed. The planned routes of vessels waiting to berth and departing are then cross-calculated with the risk distribution for port waters to obtain avoidance demand values, including: Based on the congestion risk coefficient, a two-way distance transformation is performed to obtain the risk transfer function. Spatial interpolation is then performed using the risk transfer function, and local noise is filtered out to obtain the risk distribution of the port water area. Based on the risk distribution of the port water area, a family of risk contour lines is extracted. The system receives the planned routes of vessels waiting to berth or depart, uses ray tracing to identify the route intersections between the planned routes and the risk contour line family, extracts the congestion risk coefficients at each route intersection, and determines the route intersection sections based on adjacent route intersections. The angle between the tangent vector of the route crossing section and the normal vector of the risk contour line is calculated to obtain the route crossing angle. The route crossing angle and the length of the route crossing section are then converted into route crossing risk. The time to pass through the crossroads is calculated based on the ship's speed. The crossroads risk is then weighted and calculated with the time to pass through the crossroads to obtain the avoidance requirement value.

[0009] The vessel avoidance sequence is determined based on the avoidance demand value. A dynamic programming method with navigation constraints is used to calculate the avoidance route, and the navigation time window is calculated based on the avoidance route, including: The avoidance demand value is compared with a preset safety threshold. When the avoidance demand value exceeds the preset safety threshold, the waiting time of the ship is calculated. A time compensation factor is generated based on the waiting time. The avoidance demand value and the time compensation factor are combined to obtain the avoidance priority. The order of ships to avoid the obstacle is determined based on the avoidance priority. Construct a state space with heading angle and speed as state variables, set route deviation cost, travel time cost and navigation constraint violation cost in the state space, and construct a comprehensive evaluation function. Based on the order of ship avoidance, the avoidance route is obtained by using dynamic programming in the state space, and the avoidance route minimizes the comprehensive evaluation function. The earliest and latest arrival times to the avoidance zone are calculated based on the ship's current position and maneuverability. The passage time is then calculated by combining the length of the avoidance route and the ship's speed. The navigation time window is determined based on the earliest arrival time, the latest arrival time, and the passage time.

[0010] Based on the navigation time window, the passage section of the waterway is determined, and time-sharing avoidance warning information is generated within the passage section. The time-sharing avoidance warning information includes a warning level and avoidance instructions, including: Based on the navigation time window and avoidance route, the ship motion is discretized into time-series position points. A safety buffer is constructed based on the ship maneuvering characteristics. The safety buffer is then expanded in the spatiotemporal coordinate system to obtain the waterway passage range. The waterway passage section is divided into spatiotemporal grid units. The ship density of each grid unit is calculated. The initial risk value is obtained by combining the avoidance priority. The risk transmission coefficient is calculated based on the initial risk values ​​of adjacent grid units to generate the navigation risk index. A dynamic risk threshold is established based on the spatiotemporal distribution of the navigation risk index. The navigation risk index is compared with the dynamic risk threshold to obtain the warning level. An avoidance command is generated based on the warning level and the ship maneuvering response characteristics. Within the navigation channel, the warning level and avoidance instructions are combined to generate time-sharing avoidance warning information.

[0011] A second aspect of the present invention provides a ship berthing and departure early warning system, comprising: The first unit is used to acquire vessel location data in the port waters and convert the vessel location data into vessel distribution density based on the gridded water information. The second unit is used to extract time-series features of ship distribution density, establish ship motion prediction relationships, and calculate the congestion risk coefficient based on the ship motion prediction relationships. The third unit is used to construct the risk distribution of port waters based on the congestion risk coefficient. It calculates the route cross-cutting between the planned routes of waiting and departing vessels and the risk distribution of port waters to obtain the avoidance demand value. The fourth unit is used to determine the order of ships to avoid obstacles based on the obstacle avoidance demand value, calculate the obstacle avoidance route using a dynamic programming method with navigation constraints, and calculate the navigation time window based on the obstacle avoidance route. The fifth unit is used to determine the channel passage section according to the navigation time window, and generate time-sharing avoidance warning information within the channel passage section. The time-sharing avoidance warning information includes the warning level and avoidance instructions.

[0012] A third aspect of the present invention provides an electronic device, comprising: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0013] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0014] In this embodiment, by collecting vessel position data and converting it into distribution density, and extracting temporal features to establish a motion prediction model, a precise assessment of port waterway congestion risk is achieved, effectively improving the scientific nature and accuracy of berthing and departure arrangements. A waterway risk distribution map is constructed based on the congestion risk coefficient and cross-calculated with planned routes to determine the avoidance sequence and optimal route. This solves the problems of arbitrary and unsystematic vessel avoidance decisions in traditional berthing and departure management, significantly reducing the risk of vessel collisions. By calculating navigation time windows and generating time-sharing avoidance warning information, dynamic control of the vessel berthing and departure process is achieved. Compared with traditional berthing and departure management methods, this approach can provide early warnings of potential risks, optimize navigation sequences, and significantly improve port navigation efficiency and safety. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating the ship berthing and departure early warning method according to an embodiment of the present invention; Figure 2 This is a flowchart of the ship collision avoidance decision-making and route planning process according to an embodiment of the present invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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 skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0018] Figure 1 This is a flowchart illustrating the ship berthing and departure early warning method according to an embodiment of the present invention, as shown below. Figure 1 As shown, the method includes: Obtain vessel location data within the port waters, and convert the vessel location data into vessel distribution density based on gridded water information; Temporal features of ship distribution density are extracted, ship motion prediction relationships are established, and congestion risk coefficients are calculated based on ship motion prediction relationships. Based on the congestion risk coefficient, a risk distribution of port waters is constructed. The planned routes of waiting and departing vessels are cross-calculated with the risk distribution of port waters to obtain the avoidance demand value. The order of ships to avoid obstacles is determined based on the obstacle avoidance demand value. The obstacle avoidance route is calculated using a dynamic programming method with navigation constraints. The navigation time window is then calculated based on the obstacle avoidance route. The navigation passage section is determined according to the navigation time window, and time-sharing avoidance warning information is generated within the navigation passage section. The time-sharing avoidance warning information includes the warning level and avoidance instructions.

[0019] In one optional implementation, acquiring vessel location data within the port waters and converting the vessel location data into vessel distribution density based on gridded water information includes: Obtain the geographical boundary location data of the port waters, and divide the port waters into grids of equal size according to the minimum safe distance between ships to obtain gridded waters location data; Obtain vessel position data and vessel weight values ​​within the gridded water area; A multi-scale sliding window is constructed with a preset time interval, and the time decay characteristics of ship position data are calculated at different time scales to obtain a multi-dimensional time feature vector. The ship position data is decomposed based on the multidimensional time feature vector to extract the ship motion trend features, and the density correlation between the ship and the grid is calculated by combining the gridded water position data. A ship spatiotemporal state transition matrix is ​​constructed using density correlation. The density distribution at each time scale is iterated, and the confidence level of the density distribution is calculated based on the state iteration results. The density distributions at different time scales are then fused based on the confidence level to obtain the ship distribution density.

[0020] In this implementation method, the geographical boundary location data of the port waters first needs to be obtained. This data can be measured by GPS equipment and recorded as a series of geographical coordinate points, such as latitude and longitude pairs in the form of (121.123456, 31.123456). The port waters are then divided into grids of equal size according to the minimum safe distance between vessels. For example, if the minimum safe distance between vessels is 200 meters, the entire port waters can be divided into square grids with sides of 200 meters. Each grid is defined by its center point coordinates and side length, forming gridded waters location data. For example, the center point coordinates of grid number G1 are (121.123456, 31.123456), and the side length is 200 meters.

[0021] Obtain vessel location data within the gridded waterway, including vessel ID, latitude and longitude coordinates, and timestamp. For example, vessel A's location at time t1 is (121.234567, 31.234567). Simultaneously, obtain a weight value for each vessel. The weight value can be determined comprehensively based on factors such as vessel tonnage, length, and dangerous goods carrying capacity. For example, a large oil tanker has a weight value of 5, a medium-sized cargo ship has a weight value of 3, and a small fishing vessel has a weight value of 1.

[0022] A multi-scale sliding window is constructed at preset time intervals to extract time-dimensional features from ship position data. Three different time scales are set: short-term (15 minutes), medium-term (1 hour), and long-term (4 hours). For each time scale, the sliding window sizes are set to 15 minutes, 60 minutes, and 240 minutes, respectively, with a sliding step of 5 minutes. Within each sliding window, the time decay feature of the ship position data is calculated. The time decay feature considers the timeliness of the data, with more recent data having a higher weight than older data. The specific calculation method is as follows: for the ship position data at each time point within the window, a time decay factor is calculated based on the difference between the data and the current time. For example, if the data time differs from the current time by t minutes, the decay factor can be set to exp(-0.05*t). The weighted ship position data at all time points are summed to obtain the time feature vector at that time scale.

[0023] For a specific moment, such as 10:00:00 on June 1, 2023, corresponding ship position data sets are acquired at three time scales: short-term, medium-term, and long-term. The short-term window includes data from 9:45:00 to 10:00:00, the medium-term window includes data from 9:00:00 to 10:00:00, and the long-term window includes data from 6:00:00 to 10:00:00. For the ship position data within each window, a time decay factor is applied for weighting to generate feature vectors V1, V2, and V3 at the three time scales.

[0024] Feature decomposition of ship position data is performed using multidimensional time feature vectors to extract ship motion trend features. For each ship, its position changes at different time scales are analyzed, and motion features such as speed and direction are extracted. For example, ship A has an average speed of 5 knots and a heading of 045 degrees in the short-term window; an average speed of 4.8 knots and a heading of 050 degrees in the medium-term window; and an average speed of 5.2 knots and a heading of 048 degrees in the long-term window. The density correlation between ships and grids is calculated by combining gridded water position data. For each grid, the number of ships that may appear in that grid at each time scale and their weighted sum are counted to form a density correlation matrix. For example, the density correlation of grid G1 is 15 (representing the weighted number of ships) in the short-term time scale, 18 in the medium-term, and 20 in the long-term.

[0025] A spatiotemporal state transition matrix for ships is constructed using density correlation. This matrix describes the probability of a ship moving from one grid cell to another. Based on historical ship trajectory data, the probability Pij of a ship moving from grid i to grid j is calculated. For example, the probability P12 = 0.3 of moving from grid G1 to grid G2 indicates that 30% of ships will move from G1 to G2. State iteration is performed on the density distribution at each time scale. The density distribution at the current time t is multiplied by the state transition matrix to predict the density distribution at time t+1. Through multiple iterations, the convergence of the density distribution is observed; faster convergence indicates higher stability and greater confidence in the distribution.

[0026] The confidence level of the density distribution is calculated based on the state iteration results. For example, the density distribution converges after 5 iterations at a short-term time scale, with a confidence level of 0.9; it converges after 8 iterations at a medium-term time scale, with a confidence level of 0.7; and it converges after 12 iterations at a long-term time scale, with a confidence level of 0.5. The density distributions at different time scales are then weighted and fused based on the confidence level. For grid G1, its final density value is 0.9*15 + 0.7*18 + 0.5*20 = 33.1, indicating a high ship distribution density in this grid.

[0027] The calculated vessel density data is visualized to generate a vessel density heatmap for port waters. Darker areas on the heatmap indicate higher vessel density and require greater attention. This method can be applied to port traffic management, collision risk assessment, and waterway planning to improve the safety and efficiency of port waters.

[0028] In one optional implementation, time-series features of ship distribution density are extracted to establish ship motion prediction relationships. Congestion risk coefficients are then calculated based on these ship motion prediction relationships, including: Arrange the ship distribution density data in chronological order and calculate the change in ship distribution density. The ship distribution density change values ​​are divided into regions, the change trend of ship distribution density in each region is analyzed, and the change trend is compared with the preset navigation density threshold to determine the density fluctuation point. Based on density fluctuation points, the density influence range is constructed, the spatiotemporal evolution characteristics of ship distribution density within the density influence range are extracted, the topological structure of density field changes is established, and the associated paths and transmission intensity of density transmission are identified based on the topological structure. A density propagation network is constructed using associated paths and transmission intensity. The direction of density propagation is identified, the rate of density growth is calculated, and the density diffusion pattern is predicted by combining density transmission efficiency. The ship motion prediction relationship is obtained, and the ship motion state is evaluated based on the ship motion prediction relationship to obtain the congestion risk coefficient.

[0029] In one embodiment, a ship motion prediction relationship can be established by extracting time-series features from ship distribution density, and a congestion risk coefficient can be calculated based on this relationship. First, the collected ship distribution density data is arranged chronologically, with the density data recorded at each time point denoted as D(t), where t represents the time point and D represents the ship density value. For two adjacent time points t1 and t2, the density change ΔD = D(t2) - D(t1) is calculated. For example, in a certain sea area grid region, if the ship density at time t1 is 8 ships / km² and the ship density at time t2 is 12 ships / km², then the density change ΔD for this region during this time period is 4 ships / km². By calculating for all time points of the entire observation time series, a complete density change sequence can be obtained.

[0030] When processing the changes in ship density by region, the study area is divided into several regional grids, each 0.5 nautical miles × 0.5 nautical miles in size. For each grid region, the changing trend of ship density within that region is analyzed. The preset navigation density threshold can be determined based on historical data of the sea area, for example, set at 15 ships per square kilometer. When the rate of change in ship density in a certain area exceeds 20% of the preset threshold, i.e., the density growth rate is greater than 3 ships / square kilometer / hour, that point is marked as a density fluctuation point. For example, within a monitored grid area, if the ship density increases from 10 ships / square kilometer to 14 ships / square kilometer within 1 hour, the rate of change reaches 40%, exceeding the 20% threshold, therefore that grid is marked as a density fluctuation point.

[0031] When constructing the density influence range based on density fluctuation points, the initial influence range is defined by considering eight neighboring grid regions centered on the density fluctuation point. Subsequently, a broader influence range is determined by calculating the density correlation between each grid. Density correlation is determined by the consistency of density changes in adjacent grids at multiple time points; grids with a correlation coefficient greater than 0.6 are considered to have a density transfer relationship. Spatiotemporal evolution feature extraction involves recording the density changes of grids within each influence range at consecutive time points. For example, if grid A shows a density increase at time t, and the adjacent grid B also shows a density increase at time t+1, with similar magnitudes of change, then a density transfer relationship from A to B is considered to exist.

[0032] Based on these spatiotemporal evolution characteristics, a topological structure of density field changes is established. This topological structure can be represented as a directed graph, where nodes represent grid regions and edges represent density transfer relationships. The edge weights represent the transfer intensity, calculated as the ratio of the density change in the target grid to the density change in the source grid. For example, if the density of grid A increases by 4 g / km², resulting in a subsequent increase of 3 g / km² in grid B, then the transfer intensity from A to B is 0.75.

[0033] When constructing a density propagation network using associated paths and transmission intensity, all grid points with transmission relationships are connected to form the network structure. The directionality within the network is determined by the temporal sequence of density transmission. Analyzing the network structure can identify the main direction of density propagation. For example, if density changes in multiple adjacent areas show a trend of propagation from the port towards the channel, it can be determined that ships are moving from the port area towards the channel. The rate of density increase can be determined by calculating the rate of density change per unit time for each grid point; for example, the density in a grid area at the entrance of a channel increases by 2.5 ships per square kilometer per hour.

[0034] Density transfer efficiency refers to the time required for density to transfer from one region to another and the density retention rate during the transfer process. For example, if the density in region A increases, and the density in region B increases 30 minutes later, with the increase in density in region B being 70% of that in region A, then the transfer efficiency is 70% / 30 minutes. These parameters can be used to predict changes in ship density in each grid region over a future period, establishing predictive relationships for ship motion.

[0035] Based on the predicted density distribution changes, the ship movement status is assessed. When the future density of a certain area is predicted to exceed the safety threshold, a congestion risk coefficient is calculated. The congestion risk coefficient can be expressed as the ratio of the predicted density to the safety threshold multiplied by the density growth rate. For example, if the safety density threshold for a certain waterway area is 20 ships / km², and the predicted density for that area will reach 28 ships / km² in 2 hours, and the current density growth rate is 2 ships / km² / hour, then the congestion risk coefficient is calculated as (28 / 20)×2=2.8. A risk coefficient greater than 2.5 indicates a high-risk area, requiring traffic control measures.

[0036] The aforementioned technical solutions can accurately capture the temporal variation characteristics of ship distribution density in dynamic port environments. By identifying density fluctuation points and their impact ranges, local anomalies in ship movement trends can be reflected, thus avoiding biases caused by static analysis at a single moment. By constructing the topological structure of the density field and identifying associated paths and transmission strengths, the diffusion direction and transmission patterns of ship distribution changes can be revealed, enabling proactive predictions of congestion situations. The density propagation network established based on this not only reflects the overall trend of ship movement but also quantifies the risk transmission relationships between different areas, thereby providing accurate congestion risk coefficients for ship berthing and departure processes, improving the timeliness and reliability of early warnings.

[0037] In one optional implementation, a density influence range is constructed based on density fluctuation points, the spatiotemporal evolution characteristics of ship distribution density within the density influence range are extracted, a topological structure of density field changes is established, and the associated paths and transmission intensities of density transfer are identified based on the topological structure, including: Spatial clustering of density fluctuation points yields the core region of density fluctuation. A density influence matrix is ​​constructed based on the core region of density fluctuation, and the range of density influence is determined according to the spatial distribution of the density influence matrix. A spatiotemporal tensor representation matrix is ​​constructed within the density influence range. The spatiotemporal tensor representation matrix is ​​decomposed to obtain the principal direction vector and secondary direction vector of the ship distribution density. The spatiotemporal evolution characteristics of the ship distribution density are extracted based on the principal direction vector and secondary direction vector. The spatiotemporal evolution characteristics of ship distribution density are discretized into a simple complex sequence, and the continuous homology characteristics of the simple complex sequence are calculated. Based on the continuous homology characteristics, the topological structure of density field changes is obtained. A topological connectivity graph is obtained by performing connectivity analysis on the topological structure of the density field change. The set of edges in the topological connectivity graph is extracted as the associated paths of density transmission, and the transmission strength of the associated paths is obtained by calculating the connectivity of each edge in the topological connectivity graph.

[0038] This invention provides an implementation scheme that constructs a density influence range based on density fluctuation points, extracts the spatiotemporal evolution characteristics of ship distribution density within the density influence range, establishes a topological structure of density field changes, and identifies the associated paths and transmission intensity of density transmission based on the topological structure.

[0039] In practical applications, the first step is to acquire ship location data within the study area, including the ship's spatial coordinates (x, y) and corresponding timestamp t. The study area is then divided into grid cells, each 500m × 500m in size. The number of ships in each grid cell is calculated over different time periods, forming a ship distribution density value. By comparing the changes in density values ​​of each grid cell within adjacent time periods, grid cells whose density changes exceed a set threshold are marked as density fluctuation points. For example, if a grid cell has a ship density of 10 ships / km² at time t and 15 ships / km² at time t+1, representing a density change rate of 50%, and if a threshold of 30% is set, then this grid cell is marked as a density fluctuation point.

[0040] The DBSCAN clustering algorithm is applied to the marked density fluctuation points to group spatially similar density fluctuation points into one cluster. In implementation, the neighborhood distance parameter is set to 1000 meters, and the minimum sample size is 5. Multiple density fluctuation point clusters are obtained through clustering, and the center point of each cluster is defined as the density fluctuation core region. For example, in the strait region, three density fluctuation core regions may be identified, located at (121.5°E, 31.2°N), (121.7°E, 31.1°N), and (121.6°E, 31.3°N), respectively.

[0041] A density influence matrix is ​​calculated for each density fluctuation core region. This matrix represents the degree of influence of the core region on the density changes of the surrounding area. Specifically, the gradient values ​​of density changes in each direction are calculated within a 5-kilometer radius of the core region. The larger the gradient value, the stronger the density influence in that direction. For example, if the density gradient from the core region to the northeast is 0.8 g / km² / km and the density gradient to the southwest is 0.3 g / km² / km, then the density influence in the northeast direction is stronger.

[0042] Based on the spatial distribution of the density influence matrix, the density influence range is determined. An influence threshold of 0.2 g / km² / km is set; when the density gradient value in a certain direction exceeds this threshold, that direction is included in the density influence range. For example, for the three core regions mentioned above, irregularly shaped influence ranges are obtained, with a total coverage area of ​​approximately 78 km².

[0043] Within a defined density influence range, a spatiotemporal tensor representation matrix is ​​constructed. This matrix has dimensions m×n×p, where m and n correspond to the spatial dimension, and p corresponds to the temporal dimension. The matrix element values ​​are the ship densities at the corresponding spatiotemporal locations. For example, within a 24-hour observation period, a 100×100×24 tensor representation matrix can be constructed with 1-hour time intervals.

[0044] Tensor decomposition can be performed on the constructed spatiotemporal tensor representation matrix using the Tucker decomposition method, which decomposes the original tensor into a product of a core tensor and a factor matrix. After decomposition, principal and secondary direction vectors are extracted. The principal direction vector represents the spatiotemporal direction of the most significant density change, while the secondary direction vector represents the less significant direction of change. In this example, the principal direction vector might point to (0.7, 0.3, 0.65), representing a density change pattern towards the northeast with a temporal increasing trend.

[0045] Based on the primary and secondary direction vectors, the spatiotemporal evolution characteristics of ship distribution density are extracted. These characteristics include the main direction of density change, the rate of change, and periodicity. For example, in busy waterway areas, a tidal effect with a 12-hour cycle may be observed in density changes, with an average density of 15 ships / km² during the day and decreasing to 8 ships / km² at night.

[0046] The extracted spatiotemporal evolution features are discretized into a sequence of simple complexes. A simple complex is a topological structure composed of geometric elements such as points, lines, and surfaces. In practice, density values ​​are binarized according to different thresholds to obtain binary density fields at different density thresholds. For example, setting 10 density threshold points of 2, 4, 6, 8, 10, 12, 14, 16, 18, and 20 points per square kilometer generates 10 corresponding binary density fields.

[0047] Calculate the persistent homology features of simple complex sequences. Extract topological features for each binary density field, including the number of connected components and the number of holes. These topological features change as the density threshold changes; the pattern of change is the persistent homology feature. When the density threshold increases from 4 nodes / km² to 8 nodes / km², the number of connected components decreases from 12 to 7, indicating that some low-density regions are filtered out.

[0048] The topological structure of density field changes is obtained based on persistent homology features. Topological features at different density thresholds are connected to form a complete topological structure. This structure reflects the organization of the density field at different scales. In an example, a topological structure consisting of 35 nodes and 42 edges might be obtained.

[0049] Connectivity analysis is performed on the topological structure of the density field variation to obtain a topological connectivity graph. By calculating the connectivity between nodes, it is determined which regions have density transfer relationships. For example, if node A and node B are directly connected in the topological structure by an edge, then a density transfer relationship is considered to exist between A and B.

[0050] Extract the set of edges in the topologically connected graph as the associated paths for density propagation. For example, the extracted set of associated paths may contain 42 paths such as {(A, B), (B, C), (A, D), (C, E), ...}, where each path represents the density propagation relationship between two regions.

[0051] The transit strength of associated paths is obtained by calculating the connectivity of each edge in the topological graph. Connectivity can be determined by the topological distance and density correlation between two nodes. For example, the transit strength of path (A, B) is 0.85, indicating a strong density transit relationship between regions A and B; while the transit strength of path (C, E) is 0.32, indicating a weak density transit relationship between regions C and E.

[0052] The above implementation scheme achieves a complete technical solution for constructing the density influence range based on density fluctuation points, extracting the spatiotemporal evolution characteristics of ship distribution density, establishing the density field change topology, and identifying density transmission correlation paths and transmission intensity.

[0053] In one optional implementation, a port waterway risk distribution is constructed based on a congestion risk coefficient. The planned routes of vessels waiting to berth or departing are then cross-calculated with the port waterway risk distribution to obtain avoidance demand values, including: Based on the congestion risk coefficient, a two-way distance transformation is performed to obtain the risk transfer function. Spatial interpolation is then performed using the risk transfer function, and local noise is filtered out to obtain the risk distribution of the port water area. Based on the risk distribution of the port water area, a family of risk contour lines is extracted. The system receives the planned routes of vessels waiting to berth or depart, uses ray tracing to identify the route intersections between the planned routes and the risk contour line family, extracts the congestion risk coefficients at each route intersection, and determines the route intersection sections based on adjacent route intersections. The angle between the tangent vector of the route crossing section and the normal vector of the risk contour line is calculated to obtain the route crossing angle. The route crossing angle and the length of the route crossing section are then converted into route crossing risk. The time to pass through the crossroads is calculated based on the ship's speed. The crossroads risk is then weighted and calculated with the time to pass through the crossroads to obtain the avoidance requirement value.

[0054] In this embodiment, the process of obtaining the risk transfer function based on the congestion risk coefficient through bidirectional distance transformation includes: collecting ship navigation data in the port waters, including ship position, speed, and heading; calculating the ship density within each grid cell based on the ship navigation data, setting a threshold of 3 ships per square kilometer, and defining the area as a congested area when the density exceeds this threshold; performing connectivity analysis on the congested areas and merging adjacent congested areas to form congestion risk sources; calculating the distance from the surrounding areas to the risk source with the congestion risk source as the center, forming a distance field; setting a risk attenuation factor of 0.15, calculating the risk transfer value based on the distance field and the attenuation factor, with the risk transfer value decreasing as the distance from the risk source increases; and superimposing the risk transfer values ​​of all congestion risk sources to obtain the risk transfer function for the entire port waters.

[0055] The process of obtaining the risk distribution of port waters through spatial interpolation using the risk transfer function and filtering out local noise includes: performing Kriging interpolation on a 100m × 100m grid to obtain the initial distribution of the risk field; smoothing the initial distribution using a Gaussian filter with a radius of 300m to remove local noise; normalizing the filtered risk values ​​to the range of 0-1 to form a risk distribution map of the port waters; setting risk levels according to the risk values, for example, risk values ​​of 0-0.2 are low-risk areas, 0.2-0.5 are medium-risk areas, 0.5-0.8 are high-risk areas, and 0.8-1 are extremely high-risk areas; and extracting a family of risk contour lines based on the above risk distribution, setting the contour line interval to 0.1, and extracting a total of 10 risk contour lines.

[0056] The process of receiving planned routes for vessels waiting to berth or departing, and identifying the intersections of the planned routes with the risk contour line family using ray tracing, includes: discretizing the planned route into a series of line segments, each no longer than 50 meters; applying the ray tracing algorithm to each line segment to detect its intersections with the risk contour lines; recording the coordinates of each intersection and the corresponding risk contour line value; eliminating duplicate intersections and retaining those with significant changes in risk value, with a threshold set at a risk value change greater than 0.05; and arranging the remaining intersections in the order of the routes to form a sequence of route intersections.

[0057] The process of determining a route crossing section based on adjacent route intersections includes: for any two adjacent route intersections, if the difference in risk value between them is greater than 0.1, then a route crossing section is considered to be formed between these two points; calculate the coordinates of the start and end points of the route crossing section, as well as the section length; record the maximum risk value, average risk value, and rate of change of risk value within the section; classify the route crossing sections according to the rate of change of risk value: those with a rate of change greater than 0.01 / m are high-risk sections, those with a rate of change between 0.005 and 0.01 / m are medium-risk sections, and those with a rate of change less than 0.005 / m are low-risk sections.

[0058] The process of calculating the angle between the tangent vector of the route crossing section and the normal vector of the risk contour line to obtain the route crossing angle includes: at the midpoint of the route crossing section, calculating the tangent vector of the route, i.e., the heading vector; extracting the normal vector of the risk contour line at the same point, i.e., the risk gradient direction; calculating the angle θ between the tangent vector and the normal vector, with a value ranging from 0 to 90 degrees; defining the route crossing angle coefficient as sin(θ), when θ = 90 degrees, the route crosses the risk contour line perpendicularly, with a coefficient of 1, indicating the minimum risk; when θ is close to 0 degrees, the route is almost parallel to the risk contour line, with a coefficient close to 0, indicating a higher risk; in practical applications, when the route crossing angle is less than 30 degrees, it is recommended that the vessel adjust its route to increase the crossing angle.

[0059] The process of converting the route crossing angle and route crossing section length into route crossing risk includes: setting a base risk coefficient as the average risk value of the route crossing section; calculating the route crossing risk by multiplying the base risk coefficient by the length of the route crossing section (in meters), and then multiplying by (1 - route crossing angle coefficient); for example, for a section with an average risk value of 0.6, a length of 200 meters, and a route crossing angle of 45 degrees, the route crossing angle coefficient is sin(45°)≈0.7071, and the route crossing risk is 0.6×200×(1-0.7071)≈35.3; the risk value range is generally between 0 and 100, and a value greater than 50 is considered a high-risk area, requiring route adjustments.

[0060] The process of calculating the time to pass through a crossroads based on the ship's speed, and then weighting the crossroads risk with the time to pass through the crossroads to obtain the avoidance requirement value, includes: obtaining the ship's current speed in knots (nautical miles per hour); converting the speed to meters per second (1 knot is approximately 0.5144 meters per second); calculating the time to pass through the crossroads, which is equal to the length of the crossroads divided by the ship's speed; setting a time weighting coefficient, usually 1.5, indicating that the longer the time, the higher the risk; and calculating the avoidance requirement value as the crossroads risk multiplied by (the square root of the passage time multiplied by the time weighting coefficient). For example, if the ship's speed is 10 knots (approximately 5.144 meters per second), and it takes approximately 38.9 seconds to pass through a 200-meter-long crossroads, then the avoidance requirement value is 35.3 × (√38.9 × 1.5) ≈ 329.7. When the avoidance requirement value is greater than 300, the system will issue an avoidance suggestion to the ship, prompting it to adjust its course or speed.

[0061] In this embodiment, the distribution of congestion risks within port waters can be visualized as risk contour lines, enabling quantitative identification of potential risk areas in conjunction with planned vessel routes. By accurately calculating the intersection of routes and risk areas using ray tracing, high-risk sections within the navigation path can be effectively identified. Furthermore, the intersection angle and section length are transformed into route intersection risks, and a time factor is introduced into the weighted calculation based on vessel speed, ensuring that the avoidance demand value reflects both spatial risk and navigation timing characteristics. This approach better aligns with the dynamic characteristics of the actual navigation environment, providing a scientific basis for quantifying risks during berthing and departure, and enhancing the rationality and safety of avoidance decisions.

[0062] Figure 2 This embodiment demonstrates the ship avoidance decision-making and route planning process.

[0063] In one optional implementation, the vessel avoidance sequence is determined based on the avoidance demand value, a dynamic programming method with navigation constraints is used to calculate the avoidance route, and the navigation time window is calculated based on the avoidance route, including: The avoidance demand value is compared with a preset safety threshold. When the avoidance demand value exceeds the preset safety threshold, the waiting time of the ship is calculated. A time compensation factor is generated based on the waiting time. The avoidance demand value and the time compensation factor are combined to obtain the avoidance priority. The order of ships to avoid the obstacle is determined based on the avoidance priority. Construct a state space with heading angle and speed as state variables, set route deviation cost, travel time cost and navigation constraint violation cost in the state space, and construct a comprehensive evaluation function. Based on the order of ship avoidance, the avoidance route is obtained by using dynamic programming in the state space, and the avoidance route minimizes the comprehensive evaluation function. The earliest and latest arrival times to the avoidance zone are calculated based on the ship's current position and maneuverability. The passage time is then calculated by combining the length of the avoidance route and the ship's speed. The navigation time window is determined based on the earliest arrival time, the latest arrival time, and the passage time.

[0064] In determining the avoidance order, this implementation first calculates the avoidance requirement value for each vessel. Taking a three-vessel intersection scenario as an example, vessels A, B, and C have initial positions of (0 nautical miles, 0 nautical miles), (5 nautical miles, 3 nautical miles), and (8 nautical miles, -2 nautical miles), respectively, with headings of 045°, 270°, and 180°, and speeds of 12 knots each. By calculating the closest distance point between the vessels and the collision risk index, the avoidance requirement values ​​are 0.85, 0.76, and 0.69, respectively. Assuming a safety threshold of 0.7, all three vessels must avoid each other.

[0065] The waiting time for each vessel is calculated based on its distance from the avoidance zone and its current speed. Vessel A is 4 nautical miles from the avoidance zone and will take 20 minutes to reach it at a speed of 12 knots; Vessel B is 2.5 nautical miles from the avoidance zone and will take 12.5 minutes; Vessel C is 3.2 nautical miles from the avoidance zone and will take 16 minutes. Based on the waiting time, a time compensation factor is generated. Using an exponential decay function, the time compensation factors for vessels A, B, and C are 0.82, 0.93, and 0.87, respectively.

[0066] Multiplying the avoidance requirement value by the time compensation factor yields the final avoidance priority: Ship A has a priority of 0.85 × 0.82 = 0.697, Ship B has a priority of 0.76 × 0.93 = 0.707, and Ship C has a priority of 0.69 × 0.87 = 0.600. Sort by priority from highest to lowest, the avoidance order is B, A, C.

[0067] In the obstacle avoidance route calculation phase, a state space is constructed with heading angle and speed as state variables. The heading angle ranges from 0° to 359°, with intervals of 5°; the speed ranges from 6 knots to 14 knots, with intervals of 1 knot. Multiple cost functions are set within this state space.

[0068] The course deviation cost reflects the degree to which a vessel deviates from its original course. It is calculated by multiplying the deviation angle between the actual course and the target course by the deviation weight. For example, if vessel B's target course is 270° and its actual course is 290°, the deviation is 20°, and the deviation weight is set to 0.4, then the course deviation cost is 0.4 × 20 = 8.

[0069] The time cost of navigation reflects the extra time required for a ship to complete a collision avoidance maneuver. It is calculated by multiplying the time of navigation under the avoidance path with the time of the original path, and then multiplying it by the time weight. For example, if the length of the avoidance path is 6.2 nautical miles, it takes 37.2 minutes at a speed of 10 knots, while the original path only takes 33 minutes. The time difference is 4.2 minutes. If the time weight is set to 0.3, then the time cost of navigation is 0.3 × 4.2 = 1.26.

[0070] The cost of violating navigation constraints reflects the degree of compliance with safety rules, including maintaining a minimum safe distance from other vessels and avoiding entry into restricted areas. Assuming a safe distance is set at 1 nautical mile, if a waypoint is 0.7 nautical miles away from other vessels, which is less than the safe distance, the violation level is 0.3 nautical miles. With a safety violation weight of 5, the cost of violating navigation constraints at that point is 5 × 0.3 = 1.5.

[0071] The comprehensive evaluation function consists of a weighted sum of the three costs mentioned above, and its goal is to find the avoidance route that minimizes the evaluation function. Based on the previously determined avoidance order, the optimal avoidance route is first calculated for vessel B. A dynamic programming method is used, starting from the initial state, calculating the cost of all possible paths to the next state at each state node, and retaining the path with the minimum cost. For example, vessel B starts from the initial state (270°, 12 knots), considers all possible next states, such as (265°, 12 knots), (275°, 12 knots), (270°, 11 knots), etc., calculates the cost of reaching these states, and selects the state with the minimum cost to proceed to the next iteration.

[0072] After multiple iterative calculations, the optimal avoidance route for vessel B was determined to be: maintain the initial heading of 270° for 5 minutes, then turn 290° and decelerate to 10 knots for 8 minutes, finally resuming the original heading of 270° and speed of 12 knots. This route resulted in a comprehensive evaluation function value of 15.8, lower than other feasible options.

[0073] Using the same method, the optimal avoidance routes are calculated for vessels A and C in turn, taking into account the navigation constraints formed by the avoidance path of the preceding vessel B.

[0074] In the navigation time window calculation stage, the time range for reaching the avoidance zone is first determined based on the ship's current position and maneuverability. Assuming ship B's maximum speed is 16 knots and its minimum speed is 6 knots, the earliest arrival time is 2.5 nautical miles ÷ 16 knots × 60 = 9.4 minutes, and the latest arrival time is 2.5 nautical miles ÷ 6 knots × 60 = 25 minutes. Combining this with the previously calculated avoidance route, ship B's transit time through the avoidance zone is the total length of the avoidance route divided by its average speed, i.e., 6.2 nautical miles ÷ 11 knots × 60 = 33.8 minutes. Therefore, ship B's navigation time window is from the 9.4th minute to the 25th minute, with a transit duration of 33.8 minutes. Similarly, the navigation time windows for ships A and C are calculated and adjusted appropriately to avoid overlapping time windows leading to secondary avoidance issues. The final determined navigation time window ensures that ships can pass through the avoidance zone efficiently and safely.

[0075] In this embodiment, the priority of vessel avoidance can be dynamically adjusted based on a comprehensive consideration of avoidance demand and safety thresholds, making the avoidance sequence more reasonable and flexible. By introducing a time compensation factor, the contradiction between waiting time and the urgency of avoidance can be effectively balanced, avoiding scheduling imbalances caused by a single risk assessment. Combining dynamic programming methods to find the optimal solution in the state space allows the avoidance route to achieve a comprehensive optimal result between safety, efficiency, and navigation constraints. Furthermore, by calculating the earliest and latest arrival times and transit times, the vessel navigation time window can be reasonably defined, avoiding new congestion or risks caused by time conflicts, thereby achieving efficient, safe, and orderly traffic organization during berthing and unberthing processes.

[0076] In one optional implementation, a channel passage section is determined based on a navigation time window, and time-sharing avoidance warning information is generated within the channel passage section. The time-sharing avoidance warning information includes a warning level and avoidance instructions, including: Based on the navigation time window and avoidance route, the ship motion is discretized into time-series position points. A safety buffer is constructed based on the ship maneuvering characteristics. The safety buffer is then expanded in the spatiotemporal coordinate system to obtain the waterway passage range. The waterway passage section is divided into spatiotemporal grid units. The ship density of each grid unit is calculated. The initial risk value is obtained by combining the avoidance priority. The risk transmission coefficient is calculated based on the initial risk values ​​of adjacent grid units to generate the navigation risk index. A dynamic risk threshold is established based on the spatiotemporal distribution of the navigation risk index. The navigation risk index is compared with the dynamic risk threshold to obtain the warning level. An avoidance command is generated based on the warning level and the ship maneuvering response characteristics. Within the navigation channel, the warning level and avoidance instructions are combined to generate time-sharing avoidance warning information.

[0077] In this embodiment, the navigation time window information and avoidance route data of the vessel are first acquired. The navigation time window refers to the start and end times of a vessel's planned passage through a specific segment of waterway, for example, a vessel plans to pass through the segment from Port A to Port B between 08:00 and 10:00. The avoidance route includes preset standard waypoints and yaw restrictions. The continuous motion of the vessel is discretized into a series of temporal position points, that is, the vessel is sampled at certain time intervals (e.g., 30 seconds) to obtain the expected position coordinates of the vessel at each sampling time. For example, for a vessel traveling at a speed of 12 knots, a position point is recorded every 30 seconds, with an adjacent point spacing of approximately 185 meters.

[0078] A safety buffer zone is constructed based on the ship's maneuverability. The size of the safety buffer zone is related to the ship's length, width, maneuverability, and speed. Specifically, for a 100-meter-long cargo ship, the forward safety distance is set at 300 meters (3 times the ship's length), and the lateral safety distance is set at 50 meters (approximately 2 times the ship's width). The safety buffer zone can be elliptical in shape, with the major axis along the ship's course and the minor axis perpendicular to the course. This safety buffer zone is unfolded in a spatiotemporal coordinate system to form a channel passage area. This area is represented as a buffer zone for the ship's trajectory on the plane and as a time window for the ship to pass through each point in the time dimension.

[0079] The navigation channel is divided into spatiotemporal grid units. Spatially, the channel is divided into 50m x 50m grids; temporally, time periods are divided into 5-minute intervals. For each grid unit, the system calculates a vessel density index. Vessel density is defined as the ratio of the number of vessels passing through the grid per unit time to the grid area. For example, if a grid area is 2500 square meters and 3 vessels pass through within 5 minutes, the vessel density is 0.00024 vessels / square meter / minute.

[0080] The initial risk value is calculated by weighting the vessel density based on the vessel's avoidance priority. The avoidance priority is determined by factors such as vessel type, load capacity, and dangerous goods transport status. For example, passenger ships have a priority of 5, dangerous goods transport ships have a priority of 4, large cargo ships have a priority of 3, medium-sized vessels have a priority of 2, and small vessels have a priority of 1. When two vessels with priorities of 3 and 2 appear simultaneously in the same grid, the initial risk value for that grid is calculated as vessel density × (3 + 2).

[0081] The initial risk value distribution of adjacent grid cells is analyzed, and the risk transfer coefficient is calculated. Risk transfer takes into account factors such as channel shape, current conditions, and ship maneuverability. In practice, for adjacent grid cells, if the navigation conditions between the two grid cells are poor (e.g., sharp bends, narrow channels), the risk transfer coefficient is set to 0.8; if it is a standard straight section, the risk transfer coefficient is set to 0.5; if there is an obstacle separating the two grid cells, the risk transfer coefficient is 0. Through risk transfer calculation, a complete navigation risk index distribution is generated.

[0082] Based on the spatiotemporal distribution of the air traffic risk index, a dynamic risk threshold is established. The risk threshold is dynamically adjusted according to historical air traffic data for different time periods and flight segments. For example, during peak air traffic periods (such as 7:00 AM to 9:00 AM), the risk threshold is increased by 20%; when visibility is below 500 meters, the risk threshold is decreased by 15%. The calculated air traffic risk index is compared with the dynamic risk threshold to determine the warning level. The warning levels are divided into four levels: Normal, Caution, Warning, and Danger. A risk index below 50% of the threshold is considered Normal; between 50% and 80% is Caution; between 80% and 100% is Warning; and exceeding the threshold is Danger.

[0083] Based on the warning level and the ship's maneuvering response characteristics, specific avoidance instructions are generated. These instructions include operational suggestions such as course adjustments, speed changes, and route replanning. For the "Caution" level, it is recommended to maintain course and reduce speed by 10%-20%; for the "Warning" level, it is recommended to adjust course by 5-15 degrees and reduce speed by 20%-30%; for the "Danger" level, it is required to adjust course by more than 15 degrees or temporarily stop and wait. The ship's maneuvering response time is also considered; for example, large ships may require 2-3 minutes to achieve the desired turning effect, therefore sufficient operational response time is allowed in the avoidance instructions.

[0084] Ultimately, within the navigation channel, the warning level and avoidance instructions are combined to generate a time-based avoidance warning message. The warning message includes: current time, vessel position, estimated arrival time at the conflict area, risk level, recommended avoidance actions, and duration of avoidance. For example: "On [date] at 10:15, your vessel will arrive at high-risk segment A15 in 5 minutes. Warning level: Warning. It is recommended to adjust course to starboard by 12 degrees, reduce speed by 25%, and resume original course and speed after 15 minutes." This warning information is pushed to the vessel's navigator in real time via the vessel's communication terminal, ensuring that the vessel can take avoidance measures in advance and reduce navigation channel traffic risks.

[0085] A second aspect of the present invention provides a ship berthing and departure early warning system, the system comprising: The first unit is used to acquire vessel location data in the port waters and convert the vessel location data into vessel distribution density based on the gridded water information. The second unit is used to extract time-series features of ship distribution density, establish ship motion prediction relationships, and calculate the congestion risk coefficient based on the ship motion prediction relationships. The third unit is used to construct the risk distribution of port waters based on the congestion risk coefficient. It calculates the route cross-cutting between the planned routes of waiting and departing vessels and the risk distribution of port waters to obtain the avoidance demand value. The fourth unit is used to determine the order of ships to avoid obstacles based on the obstacle avoidance demand value, calculate the obstacle avoidance route using a dynamic programming method with navigation constraints, and calculate the navigation time window based on the obstacle avoidance route. The fifth unit is used to determine the channel passage section according to the navigation time window, and generate time-sharing avoidance warning information within the channel passage section. The time-sharing avoidance warning information includes the warning level and avoidance instructions.

[0086] A third aspect of the present invention provides an electronic device, comprising: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0087] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0088] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0089] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for early warning of ship berthing and departure, characterized in that, include: Obtain vessel location data within the port waters, and convert the vessel location data into vessel distribution density based on gridded water information; Temporal features of ship distribution density are extracted, ship motion prediction relationships are established, and congestion risk coefficients are calculated based on ship motion prediction relationships. Based on the congestion risk coefficient, a risk distribution of port waters is constructed. The planned routes of waiting and departing vessels are cross-calculated with the risk distribution of port waters to obtain the avoidance demand value. The order of ships to avoid obstacles is determined based on the obstacle avoidance demand value. The obstacle avoidance route is calculated using a dynamic programming method with navigation constraints. The navigation time window is then calculated based on the obstacle avoidance route. The navigation passage section is determined according to the navigation time window, and time-sharing avoidance warning information is generated within the navigation passage section. The time-sharing avoidance warning information includes the warning level and avoidance instructions.

2. The method according to claim 1, characterized in that, Obtain vessel location data within the port waters, and convert the vessel location data into vessel distribution density based on gridded water information, including: Obtain the geographical boundary location data of the port waters, and divide the port waters into grids of equal size according to the minimum safe distance between ships to obtain gridded waters location data; Obtain vessel position data and vessel weight values ​​within the gridded water area; A multi-scale sliding window is constructed with a preset time interval, and the time decay characteristics of ship position data are calculated at different time scales to obtain a multi-dimensional time feature vector. The ship position data is decomposed based on the multidimensional time feature vector to extract the ship motion trend features, and the density correlation between the ship and the grid is calculated by combining the gridded water position data. A ship spatiotemporal state transition matrix is ​​constructed using density correlation. The density distribution at each time scale is iterated, and the confidence level of the density distribution is calculated based on the state iteration results. The density distributions at different time scales are then fused based on the confidence level to obtain the ship distribution density.

3. The method according to claim 1, characterized in that, Temporal features of ship distribution density are extracted to establish ship motion prediction relationships. Based on these relationships, congestion risk coefficients are calculated, including: Arrange the ship distribution density data in chronological order and calculate the change in ship distribution density. The ship distribution density change values ​​are divided into regions, the change trend of ship distribution density in each region is analyzed, and the change trend is compared with the preset navigation density threshold to determine the density fluctuation point. Based on density fluctuation points, the density influence range is constructed, the spatiotemporal evolution characteristics of ship distribution density within the density influence range are extracted, the topological structure of density field changes is established, and the associated paths and transmission intensity of density transmission are identified based on the topological structure. A density propagation network is constructed using associated paths and transmission intensity. The direction of density propagation is identified, the rate of density growth is calculated, and the density diffusion pattern is predicted by combining density transmission efficiency. The ship motion prediction relationship is obtained, and the ship motion state is evaluated based on the ship motion prediction relationship to obtain the congestion risk coefficient.

4. The method according to claim 3, characterized in that, Based on density fluctuation points, a density influence range is constructed. The spatiotemporal evolution characteristics of ship distribution density within this range are extracted. A topological structure of density field changes is established, and the associated paths and intensity of density transmission are identified based on this topological structure. Spatial clustering of density fluctuation points yields the core region of density fluctuation. A density influence matrix is ​​constructed based on the core region of density fluctuation, and the range of density influence is determined according to the spatial distribution of the density influence matrix. A spatiotemporal tensor representation matrix is ​​constructed within the density influence range. The spatiotemporal tensor representation matrix is ​​decomposed to obtain the principal direction vector and secondary direction vector of the ship distribution density. The spatiotemporal evolution characteristics of the ship distribution density are extracted based on the principal direction vector and secondary direction vector. The spatiotemporal evolution characteristics of ship distribution density are discretized into a simple complex sequence, and the continuous homology characteristics of the simple complex sequence are calculated. Based on the continuous homology characteristics, the topological structure of density field changes is obtained. A topological connectivity graph is obtained by performing connectivity analysis on the topological structure of the density field change. The set of edges in the topological connectivity graph is extracted as the associated paths of density transmission, and the transmission strength of the associated paths is obtained by calculating the connectivity of each edge in the topological connectivity graph.

5. The method according to claim 1, characterized in that, Based on the congestion risk coefficient, a risk distribution for port waters is constructed. The planned routes of vessels waiting to berth and departing are then cross-calculated with the risk distribution for port waters to obtain avoidance demand values, including: Based on the congestion risk coefficient, a two-way distance transformation is performed to obtain the risk transfer function. Spatial interpolation is then performed using the risk transfer function, and local noise is filtered out to obtain the risk distribution of the port water area. Based on the risk distribution of the port water area, a family of risk contour lines is extracted. The system receives the planned routes of vessels waiting to berth or depart, uses ray tracing to identify the route intersections between the planned routes and the risk contour line family, extracts the congestion risk coefficients at each route intersection, and determines the route intersection sections based on adjacent route intersections. The angle between the tangent vector of the route crossing section and the normal vector of the risk contour line is calculated to obtain the route crossing angle. The route crossing angle and the length of the route crossing section are then converted into route crossing risk. The time to pass through the crossroads is calculated based on the ship's speed. The crossroads risk is then weighted and calculated with the time to pass through the crossroads to obtain the avoidance requirement value.

6. The method according to claim 1, characterized in that, The vessel avoidance sequence is determined based on the avoidance demand value. A dynamic programming method with navigation constraints is used to calculate the avoidance route, and the navigation time window is calculated based on the avoidance route, including: The avoidance demand value is compared with a preset safety threshold. When the avoidance demand value exceeds the preset safety threshold, the waiting time of the ship is calculated. A time compensation factor is generated based on the waiting time. The avoidance demand value and the time compensation factor are combined to obtain the avoidance priority. The order of ships to avoid the obstacle is determined based on the avoidance priority. Construct a state space with heading angle and speed as state variables, set route deviation cost, travel time cost and navigation constraint violation cost in the state space, and construct a comprehensive evaluation function. Based on the order of ship avoidance, the avoidance route is obtained by using dynamic programming in the state space, and the avoidance route minimizes the comprehensive evaluation function. The earliest and latest arrival times to the avoidance zone are calculated based on the ship's current position and maneuverability. The passage time is then calculated by combining the length of the avoidance route and the ship's speed. The navigation time window is determined based on the earliest arrival time, the latest arrival time, and the passage time.

7. The method according to claim 1, characterized in that, Based on the navigation time window, the passage section of the waterway is determined, and time-sharing avoidance warning information is generated within the passage section. The time-sharing avoidance warning information includes a warning level and avoidance instructions, including: Based on the navigation time window and avoidance route, the ship motion is discretized into time-series position points. A safety buffer is constructed based on the ship maneuvering characteristics. The safety buffer is then expanded in the spatiotemporal coordinate system to obtain the waterway passage range. The waterway passage section is divided into spatiotemporal grid units. The ship density of each grid unit is calculated. The initial risk value is obtained by combining the avoidance priority. The risk transmission coefficient is calculated based on the initial risk values ​​of adjacent grid units to generate the navigation risk index. A dynamic risk threshold is established based on the spatiotemporal distribution of the navigation risk index. The navigation risk index is compared with the dynamic risk threshold to obtain the warning level. An avoidance command is generated based on the warning level and the ship maneuvering response characteristics. Within the navigation channel, the warning level and avoidance instructions are combined to generate time-sharing avoidance warning information.

8. A ship berthing and departure early warning system, used to implement the method of any one of claims 1-7, characterized in that, include: The first unit is used to acquire vessel location data in the port waters and convert the vessel location data into vessel distribution density based on the gridded water information. The second unit is used to extract time-series features of ship distribution density, establish ship motion prediction relationships, and calculate the congestion risk coefficient based on the ship motion prediction relationships. The third unit is used to construct the risk distribution of port waters based on the congestion risk coefficient. It calculates the route cross-cutting between the planned routes of waiting and departing vessels and the risk distribution of port waters to obtain the avoidance demand value. The fourth unit is used to determine the order of ships to avoid obstacles based on the obstacle avoidance demand value, calculate the obstacle avoidance route using a dynamic programming method with navigation constraints, and calculate the navigation time window based on the obstacle avoidance route. The fifth unit is used to determine the channel passage section according to the navigation time window, and generate time-sharing avoidance warning information within the channel passage section. The time-sharing avoidance warning information includes the warning level and avoidance instructions.

9. An electronic device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by the processor, they implement the method described in any one of claims 1 to 7.

Images