Ship collision prediction method based on shipborne navigation data

Abstract

The invention provides a ship collision prediction method based on shipborne navigation data, and relates to the technical field of ship monitoring, and the method comprises the steps: collecting multi-source interaction data for a multi-ship interaction region; a multi-ship motion state matrix is generated through relative motion analysis; taking the multi-ship motion state matrix as input, and constructing a multi-ship interactive game model; performing ship trajectory prediction based on the multi-ship interactive game model to generate a multi-ship prediction trajectory, and correcting the multi-ship prediction trajectory based on the real-time marine environment data to generate a multi-ship future trajectory set; and according to the multi-ship future trajectory set, performing collision probability calculation, based on a collision probability calculation result, constructing an avoidance strategy game model, performing avoidance strategy game solving, and generating a multi-ship collaborative avoidance strategy. The technical problem that the accuracy of the prediction result is insufficient and the ship navigation safety is affected due to the lack of consideration of multi-ship interaction and real-time environment in the prior art is solved, and the collision prediction accuracy and the ship navigation safety are improved.

Description

Technical Field

[0001] This application relates to the technical field of ship monitoring, and specifically relates to a ship collision prediction method based on on-board navigation data. Background Art

[0002] Ship collision is one of the most serious safety hazards in maritime navigation, which may lead to significant casualties, property losses, and environmental pollution. Currently, ship collision prediction methods mainly predict potential collision risks between ships based on Automatic Identification System (AIS) data or radar monitoring data through trajectory analysis or simple collision risk assessment models. With the growth of global shipping volume, especially in areas with heavy traffic such as ports and straits, the complexity of interactions between ships has increased significantly. However, the mainstream collision prediction methods mostly focus on single-ship or two-ship collision analysis, and do not fully consider the dynamic uncertainty and behavioral impacts caused by complex multi-ship interactions, making it difficult to handle complex interaction behaviors in multi-ship systems. At the same time, these methods do not fully consider the impact of real-time marine environmental data in collision prediction, thus reducing the accuracy of prediction and unable to meet the increasingly complex navigation requirements. Summary of the Invention

[0003] This application provides a ship collision prediction method based on on-board navigation data, which solves the technical problem that the prior art lacks consideration of multi-ship interaction and real-time environment in ship collision prediction, resulting in insufficient accuracy of prediction results and affecting ship navigation safety, and achieves the technical effect of improving the accuracy of collision prediction and thus enhancing ship navigation safety.

[0004] In view of the above problems, this application provides a ship collision prediction method based on on-board navigation data. The method includes: collecting multi-source interaction data for a multi-ship interaction area, where the multi-source interaction data includes ship navigation data, neighboring radar data, and marine environmental data; after correcting and fusing the multi-source interaction data, generating a multi-ship motion state matrix through relative motion analysis; using the multi-ship motion state matrix as input to construct a multi-ship interaction game model, where the multi-ship interaction game model embeds a multi-ship interaction utility function; predicting ship trajectories based on the multi-ship interaction game model to generate multi-ship predicted trajectories, and correcting the multi-ship predicted trajectories based on real-time marine environmental data to generate a set of multi-ship future trajectories; calculating collision probabilities based on the set of multi-ship future trajectories, and constructing an avoidance strategy game model based on the collision probability calculation results; and solving the avoidance strategy game based on the avoidance strategy game model to generate a multi-ship collaborative avoidance strategy.

[0005] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0006] By collecting ship navigation data, neighborhood radar data, and marine environment data in the multi-ship interaction area, and correcting and fusing these data, it provides comprehensive and accurate information input for subsequent analysis and prediction. Through relative motion analysis, a multi-ship motion state matrix is generated, which can capture the dynamic relationships between ships and provide accurate input for the subsequent game model. Using the multi-ship motion state matrix as input, a multi-ship interaction game model is constructed to simulate the complex interaction behaviors between ships, providing decision support for ship trajectory prediction. Based on the multi-ship interaction game model, ship trajectory prediction is carried out to generate multi-ship predicted trajectories, and the multi-ship predicted trajectories are corrected based on real-time marine environment data to generate a set of multi-ship future trajectories, improving the accuracy and environmental adaptability of ship trajectory prediction and ensuring accurate prediction of ship trajectories in complex marine environments. According to the set of multi-ship future trajectories, collision probability calculation is performed, and based on the results of the collision probability calculation, an avoidance strategy game model is constructed, and the avoidance strategy game is solved to generate a multi-ship collaborative avoidance strategy, optimizing the decision-making process of ships in complex navigation scenarios, improving the collaborative avoidance ability of ships in a multi-ship environment, and reducing the collision risk.

[0007] In summary, through a series of steps such as the collection and correction and fusion of multi-source data, construction of a multi-ship interaction game model, combination of real-time marine environment data for trajectory prediction and correction, calculation of collision probability, and solution of avoidance strategies, this application comprehensively considers the ship's own state, the situation of surrounding ships, the dynamic changes of the marine environment, and the mutual relationships among multiple ships, can more accurately predict ship trajectories and calculate collision probabilities, and then formulate effective multi-ship collaborative avoidance strategies, significantly improving the safety and navigation efficiency of ship navigation and effectively reducing the occurrence of ship collision accidents.

[0008] The above description is only an overview of the technical solution of this application. In order to be able to understand the technical means of this application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features, and advantages of this application more obvious and understandable, the following specifically gives the specific implementation manners of this application. Brief Description of the Drawings

[0009] Figure 1 It is a flowchart showing the method for predicting ship collisions based on on-board navigation data provided by an embodiment of this application.

[0010] Figure 2 It is a flowchart showing the process of generating a multi-ship motion state matrix in the method for predicting ship collisions based on on-board navigation data provided by an embodiment of this application.

[0011] Figure 3 It is a flowchart showing the process of generating an avoidance strategy game model in the method for predicting ship collisions based on on-board navigation data provided by an embodiment of this application. Detailed Description of the Invention

[0012] By providing a ship collision prediction method based on on-board navigation data, the embodiments of the present application collect multi-source interaction data, combine a game model to predict the collision risk during multi-ship interaction, and generate a multi-ship collaborative avoidance strategy, realizing the accurate prediction of ship collision risk, solving the technical problem that the prior art lacks consideration of multi-ship interaction and real-time environment during ship collision prediction, resulting in insufficient accuracy of prediction results and affecting the safety of ship navigation, and improving the safety of ship navigation.

[0013] As Figure 1 shown, the embodiments of the present application provide a ship collision prediction method based on on-board navigation data, and the method includes:

[0014] Step S1: For a multi-ship interaction area, collect multi-source interaction data, where the multi-source interaction data includes ship navigation data, neighborhood radar data, and marine environment data.

[0015] Specifically, the multi-ship interaction area refers to a sea area where the navigation of two or more ships affects each other. For example, areas near busy ports, narrow straits, etc. In these areas, the distance between ships is relatively close, and the navigation decisions of each other will affect each other. In the multi-ship interaction area, multiple data sources such as the Automatic Identification System (AIS) for interactive ships, ship radar systems, and marine environment monitoring systems are used to obtain multi-source interaction data including ship navigation data, neighborhood radar data, and marine environment data. Exemplarily, through the Automatic Identification System (AIS) for ships, information such as the identity, position, speed, and heading of ships can be obtained in real time; by scanning the surrounding area with a ship radar system in real time, relevant data of surrounding ships can be obtained, such as the distance and azimuth of surrounding ships relative to the own ship; through a marine environment monitoring system (such as a marine buoy station, a meteorological satellite data receiving system, etc.), data on external environmental factors affecting ship navigation such as the speed and direction of ocean currents, wind direction, wind speed, and wave height can be obtained in real time.

[0016] Through the collection of multi-source data, the states of ships and the environment can be monitored from multiple dimensions, providing a comprehensive data basis for subsequent ship trajectory prediction, collision probability calculation, etc.

[0017] Step S2: After correcting and fusing the multi-source interaction data, generate a multi-ship motion state matrix through relative motion analysis.

[0018] Specifically, the multi-ship motion state matrix is a mathematical structure representing the motion states of multiple ships in matrix form. Each row of the matrix corresponds to a ship, and each column represents ship data such as position, speed, and heading. For example, a 3×3 matrix represents the position, speed, and heading information of 3 ships.

[0019] The multi-source interaction data collected may have errors due to factors such as environmental interference and equipment accuracy. It is necessary to calibrate and integrate the multi-source interaction data to make these data consistent in terms of time, space, etc., and remove noise and errors to improve the accuracy of the data. For example, calibrate the speed in ship navigation data and compensate for the error of the distance measurement value in radar data. Exemplarily, data fusion algorithms (such as Kalman filtering, particle filtering, etc.) can be used to correct different data sources to ensure the consistency and accuracy of the data. Based on the fused data, relative motion analysis is performed, and the relative position, relative speed, and motion trend between each ship are calculated using vector calculation methods, and these states are represented in the form of a matrix to generate a multi-ship motion state matrix. This matrix systematically represents the motion information of all ships, facilitating subsequent collision risk analysis and avoidance strategy calculation.

[0020] Step S3: Use the multi-ship motion state matrix as input to construct a multi-ship interaction game model, which embeds a multi-ship interaction utility function.

[0021] Specifically, the multi-ship interaction game model is a mathematical model used to describe the interaction and decision-making processes of multiple ships in a multi-ship interaction area. The actions of each ship may affect the actions of other ships and the final result. The multi-ship interaction utility function is an important part of the multi-ship interaction game model, which is used to measure the gains or losses of ships under different decisions and is related to risk, cost, and navigation efficiency.

[0022] Input the data in the multi-ship motion state matrix into the model construction framework, and determine the rules and strategies of the game according to the mutual relationship between ships (such as distance, relative speed, etc.) and marine environmental factors. Then, construct a multi-ship interaction utility function based on factors such as ship collision risk, cost, and navigation efficiency, and construct a multi-ship interaction game model in combination with game theory algorithms.

[0023] Through the multi-ship interaction game model, the behavior decisions of ships in a complex multi-ship environment can be effectively simulated, providing a theoretical basis for ship trajectory prediction and avoidance strategy formulation.

[0024] Step S4: Based on the multi-ship interaction game model, perform ship trajectory prediction to generate a multi-ship predicted trajectory, and correct the multi-ship predicted trajectory based on real-time marine environmental data to generate a multi-ship future trajectory set.

[0025] Specifically, the multi-ship predicted trajectory is a set of predicted trajectories of all ships obtained by performing trajectory prediction on multiple ships respectively through the multi-ship interaction game model. The multi-ship future trajectory set is a set of future trajectories of multiple ships that is more in line with the actual situation obtained by correcting the predicted trajectory based on real-time marine environmental data on the basis of the multi-ship predicted trajectory.

[0026] Substitute the multi-ship motion state matrix into the multi-ship interaction game model, and calculate the predicted trajectories of each ship within a certain period in the future according to the rules and algorithms in the model to obtain the multi-ship predicted trajectories. Then obtain real-time ocean environment data, such as the current ocean current, wind speed, wind direction, etc. According to these data, combined with the ocean dynamics model to simulate the impact of the ocean environment on the ship trajectories, adjust the previous predicted trajectories to obtain the multi-ship future trajectory set.

[0027] Ship trajectory prediction can understand the sailing directions of ships in advance. The multi-ship future trajectory set corrected by combining ocean environment data takes into account the impact of the real-time ocean environment on ship trajectories, which is more in line with the actual situation, thus providing a more accurate data basis for collision probability calculation.

[0028] Step S5: According to the multi-ship future trajectory set, calculate the collision probability, and based on the collision probability calculation result, construct an avoidance strategy game model.

[0029] Specifically, the avoidance strategy game model is a mathematical model constructed on the basis of known collision probabilities for determining ship avoidance strategies. This model takes into account factors such as the mutual relationship between ships and collision risks to seek the optimal avoidance strategy.

[0030] Analyze the trajectories of each ship in the multi-ship future trajectory set. According to factors such as the geometric shape and speed of the ships, use probability and statistics methods (such as the Monte Carlo simulation method) to calculate the collision probability between ships. Then, based on the collision probability calculation result, combined with the mutual relationship between ships (such as relative position, speed, etc.) and ocean environment factors, construct an avoidance strategy game model to determine the rules and objective function in the model.

[0031] Collision probability calculation can quantify the collision risk between ships. The avoidance strategy game model provides a theoretical framework for formulating reasonable avoidance strategies. Through collision probability calculation and the avoidance strategy game model, the avoidance behaviors in the multi-ship system can be evaluated and optimized, reducing the ship collision risk.

[0032] Step S6: Based on the avoidance strategy game model, perform avoidance strategy game solution to generate a multi-ship cooperative avoidance strategy.

[0033] Specifically, the multi-ship cooperative avoidance strategy refers to the optimal avoidance strategy generated among multiple ships based on the game model, enabling the ships to cooperate with each other during the avoidance process, reducing the collision risk and improving the navigation efficiency. Based on the avoidance strategy game model, use game solution algorithms (such as Nash equilibrium or other optimization algorithms) to generate cooperative avoidance strategies for each ship. These strategies take into account the relative behaviors of multiple ships, enabling the ships to coordinate with each other during the avoidance process, thus avoiding the generation of new collision risks.

[0034] Further, as Figure 2 shown, step S2 of the embodiment of the present application includes:

[0035] Step S21: Perform time synchronization and spatial alignment on the multi-source interaction data to generate standard multi-source interaction data.

[0036] Step S22: Generate a multi-ship motion state matrix based on the standard multi-source interaction data.

[0037] Step S23: Calculate the relative position vectors of each ship with adjacent ships to determine the relative directions between the ships.

[0038] Step S24: Based on the relative directions, use the relative velocity vectors to calculate the time closest point and the distance closest point.

[0039] Step S25: Perform a dynamic change trend analysis on the time closest point and the distance closest point to generate a real-time evaluation index reflecting the collision risk degree, and embed it into the multi-ship motion state matrix.

[0040] Step S26: Introduce an environmental correction factor to dynamically update the multi-ship motion state matrix.

[0041] Specifically, the multi-source interaction data comes from different devices or systems, and there may be differences in the acquisition time and spatial coordinate systems of these data. For example, ship navigation data may be collected every 10 seconds, while ocean environment data may be updated every 5 minutes, and the coordinate system used by the ship's own positioning system may be different from that of radar data. Before using the multi-source interaction fusion data for subsequent analysis, it is necessary to perform time synchronization and spatial alignment on these multi-source interaction data. For time synchronization, first determine a reference time, and then adjust the data to the same time scale through interpolation, extrapolation, etc. according to the acquisition frequency and timestamp of each data source. For example, if the timestamps of ship navigation data are 10:00:00, 10:00:10, etc., and the timestamps of ocean environment data are 10:00:00, 10:05:00, it is necessary to perform interpolation processing on the ocean environment data at moments such as 10:00:10. For spatial alignment, it is necessary to determine a unified spatial coordinate system, such as the longitude and latitude coordinate system, and then convert the data in different coordinate systems to this unified coordinate system through coordinate conversion formulas. After time synchronization and spatial alignment, standard multi-source interaction data is generated, and these standard multi-source interaction data are consistent in time and space, laying a foundation for accurately analyzing the relationships between ships and the relationships between ships and the environment in the subsequent process.

[0042] Extract the motion state information of each ship from the standard multi-source interaction data, such as the position, speed, heading, etc. of the ship, and then organize this information into a matrix form according to certain rules to generate a multi-ship motion state matrix. For example, if there are 3 ships, the matrix may be a 3×3 matrix, where the first column represents the position (longitude, latitude) of the ship, the second column represents the speed, and the third column represents the heading.

[0043] According to the ship position information in the multi-ship motion state matrix, for each ship, calculate the coordinate difference between it and the neighboring ships to obtain the relative position vector. Then, convert the relative position vector into a relative direction through mathematical methods such as trigonometric functions. Among them, the relative position vector is a vector describing the position of one ship relative to another ship. For example, taking ship A as the reference point, the position vector of ship B relative to ship A can be represented by a directed line segment pointing from ship A to ship B, including the magnitude (the distance between the two ships) and the direction (the direction angle from A to B). The relative direction refers to the azimuth direction of one ship relative to another ship, usually expressed in degrees. For example, the relative direction of ship B relative to ship A is 30 degrees, indicating that ship B is at the 30-degree azimuth of ship A. By calculating the relative position vector and the relative direction, the dynamic relationship between ships can be monitored in real time, so as to judge which ships may have a collision risk and provide a basis for subsequent avoidance decisions.

[0044] The relative velocity vector is a velocity vector representing one ship relative to another ship, including the magnitude and direction of the velocity, which reflects the approaching or separating speed between the two ships. The time of closest approach is the time point when the distance between two ships reaches the minimum during their future voyages. The distance of closest approach is the distance when the distance between two ships reaches the minimum during their future voyages. Calculate the relative velocity vector according to the ship speed information in the multi-ship motion state matrix. Then, based on the relative position vector and the relative velocity vector, solve for the time (time of closest approach) and distance (distance of closest approach) when the distance between the two ships is the smallest through kinematic equations. For example, a quadratic function about time can be established, and its function value represents the square of the distance between the two ships. By taking the derivative to find the minimum point of the function, the time of closest approach can be obtained, and then substituting it into the function to get the distance of closest approach. By calculating the time of closest approach and the distance of closest approach, potential collision risks can be identified in advance, providing crucial decision-making moments and distance information.

[0045] Real-time evaluation metrics are metrics used to quantify the collision risk and can reflect the possibility and urgency of collisions between ships. Record the time nearest point and distance nearest point calculated at different times, and then observe the changing trends of both over time through data analysis methods (such as the difference method, fitting curve slope analysis, etc.). According to the set rules, convert this trend into a real-time evaluation metric. For example, set a function to calculate the metric value based on the rate of decrease of the distance nearest point and the proximity of the time nearest point. This real-time evaluation metric provides dynamic monitoring of the collision risk, helps to judge the safe distance and approaching speed between ships, and provides support for dynamically adjusting the avoidance strategy.

[0046] The environmental correction factor is a coefficient or a set of coefficients that takes into account the influence of marine environmental factors (such as ocean currents, wind directions, sea waves, etc.) on the motion state of ships. For example, ocean currents will push ships to deviate from their original routes, and the environmental correction factor can quantify the magnitude of this influence. Determine the value of the environmental correction factor based on marine environmental data (such as ocean current speed, wind direction, etc.) and the physical characteristics of the ship (such as the draft depth of the ship, ship type, etc.). For example, for the influence of ocean currents on ships, a correction coefficient can be calculated based on the ocean current speed, the hydrodynamic coefficient of the ship, etc. Then apply this correction factor to the multi-ship motion state matrix to adjust information such as the positions and speeds of the ships in the matrix. Introducing the environmental correction factor to dynamically update the multi-ship motion state matrix can respond to changes in the marine environment in a timely manner, more accurately reflect the motion state of ships in the actual marine environment, and help improve the accuracy of subsequent ship trajectory prediction, collision probability calculation, etc.

[0047] Furthermore, step S3 of the embodiment of the present application includes:

[0048] Step S31: Use the multi-ship motion state matrix as the input. The multi-ship motion state matrix contains multiple key metrics, and the multiple key metrics include relative distance, relative speed, time nearest point, distance nearest point, course difference, and environmental correction factor.

[0049] Step S32: Define a multi-ship interaction utility function based on the multiple key metrics. The multi-ship interaction utility function includes a collision risk weight, a navigation efficiency weight, an avoidance cost weight, and a priority weight.

[0050] Step S33: Build the multi-ship interaction game model based on the multi-ship interaction utility function.

[0051] Specifically, extract key metrics such as relative distance, relative speed, time nearest point, distance nearest point, course difference, and environmental correction factor from the multi-ship motion state matrix, and use these key metrics as the input data for building the multi-ship interaction game model, so that the built model can accurately simulate the relative position relationship between ships.

[0052] According to the actual needs and safety standards of ship navigation, multiple weights such as the collision risk weight, navigation efficiency weight, avoidance cost weight, and priority weight are customized. Among them, the collision risk weight is used to measure the importance of collision risk in the ship interaction utility. The navigation efficiency weight reflects the importance of ship navigation efficiency in the utility function. The avoidance cost weight reflects the importance of the costs incurred by the ship in taking avoidance measures (such as time cost, fuel consumption cost, etc.) in the utility function. The priority weight considers the priority situation of different ships in the interaction. For example, large ships or ships on special missions may have a higher priority weight. Combining these weights with multiple key indicators extracted from the multi-ship motion state matrix, a multi-ship interaction utility function is defined, comprehensively considering multiple factors in the ship interaction process, providing a standard for measuring the benefits and losses of ship behavior for constructing a multi-ship interaction game model, enabling the model to make reasonable decisions according to different ship states and requirements.

[0053] Based on the multi-ship interaction utility function, combined with the key indicators in the multi-ship motion state matrix, a multi-ship interaction game model is constructed according to the principles of game theory. For example, according to indicators such as the relative distance and relative speed between ships, the decision space of the ship is determined (such as the range in which the course can be changed, the range in which the speed can be adjusted, etc.), and then according to the utility function, the utility situation of the ship under different decisions is determined. By analyzing the utility situations of the ship under different decision combinations, a multi-ship interaction game model is constructed. This model will simulate how each ship makes the optimal avoidance decision considering the actions of other ships.

[0054] Constructing a multi-ship interaction game model can simulate the complex interaction relationships between multiple ships, providing a theoretical basis for ship trajectory prediction and the formulation of avoidance strategies. Through this model, the optimal decisions of ships in different situations can be analyzed, thereby improving the safety and efficiency of ship navigation.

[0055] Furthermore, step S33 includes:

[0056] Step S331: Based on the multi-ship interaction utility function, regarding each ship as a game participant, a framework for the multi-ship interaction game model is built.

[0057] Step S332: Based on the non-cooperative game theory, a game strategy space is defined, and the game strategy space includes a course adjustment range and a speed adjustment range.

[0058] Step S334: Integrating the multi-ship interaction game model framework and the game strategy space to generate the multi-ship interaction game model.

[0059] Specifically, in the multi-ship interactive game model, each ship is regarded as an independent decision-making entity, that is, a game participant. When constructing the multi-ship interactive game model, the number of ships participating in the game is determined from the multi-ship motion state matrix. For each ship, it is associated with the multi-ship interactive utility function, and it is clear that the goal of each ship in the game is to maximize its own utility.

[0060] Non-cooperative game theory is a game theory that assumes that participants pursue the maximization of their own interests during the game process without considering cooperation with other participants. In the context of multi-ship interaction, each ship mainly focuses on its own safety (avoiding collisions), navigation efficiency and other goals, and does not necessarily actively cooperate with other ships. The game strategy space refers to the set of strategies that each ship can adopt during the game process, that is, the range of the course and speed that the ship can adjust. For example, the course adjustment range of a ship may be 30 degrees to the left or right of the current course, and the speed adjustment range may be plus or minus 5 knots based on the current speed. According to the actual operation ability of the ship and the navigation rules, the course adjustment range and speed adjustment range of each ship are determined. At the same time, according to the navigation safety regulations and the designed speed range of the ship, the speed adjustment range is determined. Then, based on non-cooperative game theory, these adjustment ranges are defined as the game strategy space of each ship. Defining the game strategy space based on non-cooperative game theory can reflect the operation possibilities of ships in actual navigation and help to more realistically simulate the decision-making process of ships in the multi-ship interactive game model.

[0061] Integrate the previously constructed multi-ship interactive game model framework and the defined game strategy space, and add the information of the game strategy space to the model framework, so that the decision-making process of each ship in the model can be carried out within the defined strategy space, thereby generating a multi-ship interactive game model. For example, when calculating the utility of a ship under different decisions, it is necessary to consider that the adjustment of its course and speed must be within the range specified by the strategy space.

[0062] Furthermore, in step S4 of the embodiment of the present application, ship trajectory prediction is performed based on the multi-ship interactive game model to generate a multi-ship predicted trajectory, including:

[0063] Step S41: Input the multi-ship motion state matrix as the initial condition of the multi-ship interactive game model.

[0064] Step S42: The multi-ship interactive game model solves the Nash equilibrium through dynamic iteration to generate an optimal multi-ship interactive behavior strategy.

[0065] Step S43: Based on the optimal multi-ship interactive behavior strategy, output a multi-ship predicted trajectory, where each trajectory records the predicted positions and velocity vectors of the ship at multiple future times.

[0066] Specifically, each element in the multi-ship motion state matrix (such as the position coordinates, speed magnitude, and direction of each ship, etc.) is input into the multi-ship interactive game model according to the format required by the model. The multi-ship motion state matrix provides various state information of the ships at the starting moment, such as the position, speed, and heading of the ships, which can be used as the basic data for the model to start calculations.

[0067] Nash equilibrium refers to a balanced state in a game where each participant has chosen its optimal strategy, and given the strategies of other participants, no participant has the incentive to change its own strategy. Based on the initially input multi-ship motion state matrix, the model makes strategy assumptions for each ship within its game strategy space (such as the heading adjustment range and speed adjustment range). Then, by calculating the utility of each ship under different assumed strategies (according to the multi-ship interactive utility function) and considering the strategies of other ships, dynamic iterative calculations are carried out. In each iteration, the ship adjusts its strategy according to the strategies of other ships to improve its own utility until Nash equilibrium is reached, that is, the strategies of all ships no longer change. The strategies of all ships at this time are output as the multi-ship optimal interactive behavior strategy. Under the combination of the multi-ship optimal interactive behavior strategies, each ship achieves the optimal of its own utility (i.e., the optimal solution of the multi-ship interactive utility function) considering the strategies of other ships.

[0068] The multi-ship predicted trajectory is the navigation trajectory of each ship within a certain period in the future calculated based on the multi-ship optimal interactive behavior strategy. Each trajectory contains the predicted positions and velocity vectors of the ship at multiple future time points. The position can be represented by coordinates (such as longitude and latitude coordinates), and the velocity vector includes the magnitude and direction of the speed. According to the multi-ship optimal interactive behavior strategy, for each ship, based on the initial position and speed, adjust the motion state according to the heading and speed specified in the strategy, and use trajectory simulation to calculate the new position and velocity vector of the ship at each time step (such as every 1 minute). For example, if the initial position of a ship is (x 0 , y 0 ), and the initial velocity vector is (V x0 , V y0 ), and the adjusted velocity vector within the first time step according to the optimal strategy is (V x1 , V y1 ), then the position after the first time step is (x 0 + V x1 × Δt, y 0 + V y1×Δt) (where Δt is the time step). By continuously repeating this process, the predicted positions and velocity vectors of each ship at multiple future times are obtained, thereby generating the multi-ship prediction trajectories. Through these prediction trajectories, the future movement paths of the ships can be predicted, providing a basis for subsequent collision prediction and avoidance strategies.

[0069] Further, as Figure 3 shown, step S5 of the embodiment of the present application includes:

[0070] Step S51: Using the multi-ship future trajectory set as the input, based on the trajectory time series, gradually pair the future trajectory of each ship with the trajectories of other ships to obtain multiple trajectory interference combinations.

[0071] Step S52: Based on the multiple trajectory interference combinations, calculate the relative velocity and relative direction between each pair of ships to generate a dynamic relative motion vector.

[0072] Step S53: According to the dynamic relative motion vector, extract multiple perturbation samples for simulation, calculate the collision probability under each possible condition, and generate a global collision risk matrix.

[0073] Step S54: Treat each ship as a game participant, and all participants share the global collision risk matrix to generate the avoidance strategy game model.

[0074] Specifically, the trajectory time series is the sequence of the predicted trajectories of each ship in time, that is, a series of position and velocity vectors arranged in chronological order. The trajectory interference combination is the pairing combination of the future trajectory of each ship with the trajectories of other ships at different time points, used to analyze the mutual interference situation between ships. Extract the trajectory data of each ship from the multi-ship future trajectory set. Then, according to the trajectory time series, gradually pair the future trajectory of each ship with the trajectories of other ships. For example, for two ships A and B, the position and velocity vectors of A at the first time point can be paired with the position and velocity vectors of B at the first time point, and then the position and velocity vectors of A at the second time point can be paired with the position and velocity vectors of B at the second time point, and so on, to obtain multiple trajectory interference combinations.

[0075] For each trajectory interference combination, calculate the relative velocity and relative direction between the two vessels. Among them, the relative velocity is the velocity difference between the two vessels, including the velocity direction and magnitude. The relative direction is the direction difference between the two vessels, usually expressed in angles. Extract the velocity vectors and heading angles of the two vessels from the trajectory interference combination. Then, calculate the relative velocity vector based on the velocity vectors and the relative direction based on the heading angles. For example, use vector subtraction to calculate the relative velocity vector and the angle difference formula to calculate the relative direction. Combine multiple relative velocities and relative directions to generate a dynamic relative motion vector for describing the dynamic relative motion state between the two vessels.

[0076] The perturbation samples are samples extracted from the dynamic relative motion vector for simulating the vessel motions under different conditions. These samples can be generated by making slight changes to the relative velocity and relative direction to simulate different navigation conditions and vessel behaviors. The global collision risk matrix is a matrix for comprehensively evaluating the multi-vessel collision risk. Each element in the matrix represents the collision probability of a pair of vessels at a specific time step. Extract multiple perturbation samples from the dynamic relative motion vector. Then, simulate each perturbation sample and calculate the collision probability of the two vessels under the conditions represented by the sample. Whether a collision occurs can be determined by defining a collision area (such as a circular area centered on the vessel) and checking whether the two vessels enter each other's collision area during the simulation. Finally, summarize the collision probabilities of all perturbation samples into the global collision risk matrix. The global collision risk matrix can comprehensively reflect the collision risks of the multi-vessel system under different conditions, thus providing a detailed risk assessment for the subsequent avoidance strategy design.

[0077] Define each vessel as a game participant and assign a target function to each participant. The target function can be to minimize the collision risk between itself and other vessels, thereby generating an avoidance strategy game model. In the model, the decision-making range of the participants can be restricted by defining the strategy space (such as the heading adjustment range and speed adjustment range). This process is similar to constructing a multi-vessel interaction game model, and the relevant descriptions of the foregoing steps can be referred to.

[0078] Furthermore, step S53 of the embodiment of the present application includes:

[0079] Step S531: Extract multiple perturbation samples of the future trajectories of each pair of vessels according to the dynamic relative motion vector.

[0080] Step S532: Use the time to closest point and the distance to closest point as risk indicators to define a collision risk function: where P is the collision probability, TCPA is the time to closest point, CPA is the distance to closest point, and α and β are weight factors.

[0081] Step S533: Using the collision risk function, perform simulation calculations for the multiple perturbation samples to generate multiple collision probabilities.

[0082] Specifically, when calculating the collision probability, first, analyze the information such as the speed and direction of each pair of ships according to the dynamic relative motion vector. Perturbation samples are generated by making small changes to this information. For example, the Monte Carlo simulation algorithm can be used to add a small random increment to the ship's velocity vector or make a small angular change in the direction. For each pair of ships, multiple perturbation samples can be generated to cover various possible situations during navigation.

[0083] Using the time to the closest point and the distance to the closest point as risk indicators, define a collision risk function as follows: In this collision risk function, P represents the collision probability. TCPA is the time to the closest point, and CPA is the distance to the closest point, which can be determined based on the relative position vector and relative velocity vector in the perturbation samples using the same operation process as in steps S23 to S24. α and β are the weight factors corresponding to the time to the closest point and the distance to the closest point respectively, which can be customized with reference to historical data based on research and experience in ship navigation safety.

[0084] For each perturbation sample, first obtain the values of the time to the closest point and the distance to the closest point corresponding to this sample (refer to steps S23 to S24). Then substitute these values into the collision risk function and calculate the collision probability according to the operation rules of the function. By simulating and calculating multiple collision probabilities, a more comprehensive understanding of the collision risk of ships in different possible situations can be obtained, providing a data basis for generating the global collision risk matrix.

[0085] The above steps calculate the collision probability of each pair of ships according to the perturbation samples, which can simulate and evaluate the collision risk of ships in various possible situations. This process takes into account various uncertainties during navigation and provides a more accurate risk assessment for the calculation of collision probability, improving the accuracy of collision risk prediction.

[0086] Furthermore, step S6 of the embodiment of the present application includes:

[0087] Step S61: Define a global optimization objective in the collision avoidance strategy game model, where the global optimization objective includes minimizing the collision risk between ships and maximizing the overall navigation efficiency.

[0088] Step S62: Based on the global optimization objective, use distributed game solving. Each ship independently calculates a preliminary collision avoidance plan and updates it to the global optimal plan through multiple rounds of games to generate the multi-ship collaborative collision avoidance strategy.

[0089] Specifically, in the avoidance strategy game model, a global optimization goal is set from an overall perspective. This global optimization goal includes two aspects: minimizing the collision risk between ships and maximizing the overall navigation efficiency. Exemplarily, the theory and methods of multi-objective optimization can be used. By defining the objective function and constraints, the two objectives can be comprehensively considered using the linear weighted method or other multi-objective optimization algorithms. For example, the collision risk can be quantified as a numerical value (such as the collision probability in step S53), and the requirement of minimization can be represented by setting a threshold or an objective function; for the navigation efficiency, factors such as the average speed and voyage of the ship can be considered, and a corresponding mathematical expression can be constructed to represent the maximization goal. Defining the global optimization goal provides a clear direction for the formulation of the multi-ship cooperative avoidance strategy, which can improve the operation efficiency of the entire ship group while ensuring the safety of the ships.

[0090] Distributed game solving is a method for solving game problems. Each ship is regarded as an independent decision-making entity, and calculations are performed based on its own state and the global optimization goal, without the need for a centralized control center to command uniformly. The preliminary avoidance plan is an avoidance plan initially calculated by each ship according to its own information (such as its own position, speed, heading, etc.) and the global optimization goal. The global optimal plan is a combination of avoidance plans reached by all ships after multiple rounds of games. Under this plan, the global optimization goal can be achieved, that is, minimizing the collision risk between ships and maximizing the overall navigation efficiency.

[0091] In the process of distributed game solving, first, calculations are performed independently for each ship. Based on the current state of each ship (information obtained from the multi-ship motion state matrix), the global optimization goal, and the preliminary estimate of the behavior of other ships, a preliminary avoidance plan is calculated. For example, a ship initially decides whether to adjust its heading to the left by a certain angle or reduce its speed based on its relative position and speed with other ships. Then, multiple rounds of games are carried out. In each round of the game, the ship exchanges information about its preliminary avoidance plan with other ships. Each ship re-evaluates its plan based on the plans of other ships to determine whether it can be further optimized to better achieve the global optimization goal. For example, if a ship finds that the avoidance plan of other ships will increase its adjusted collision risk or reduce its navigation efficiency, it will re-adjust its plan. After multiple rounds of such a game process, until the avoidance plans of all ships no longer change, the global optimal plan is obtained at this time. Finally, based on this global optimal plan, a multi-ship cooperative avoidance strategy is generated, specifying the specific avoidance operations of each ship in different situations.

[0092] The above steps enable each ship to make decisions according to its own actual situation through distributed computing, without relying on a complex centralized control system, which improves the decision-making flexibility. Through multiple rounds of games, ships can fully consider each other's avoidance plans, thus obtaining a globally optimal plan, effectively balancing the two important factors of collision risk and navigation efficiency. The generated multi-ship collaborative avoidance strategy can improve the safety and operation efficiency of multiple ships in a complex navigation environment.

[0093] In summary, the ship collision prediction method based on on-board navigation data provided by the embodiments of the present application has the following technical effects:

[0094] The embodiments of the present application comprehensively collect and fuse multi-source data, accurately analyze the relative motion between ships, construct and apply a multi-ship interactive game model for trajectory prediction, adjust the prediction results in real time to adapt to changes in the marine environment, and formulate a multi-ship collaborative avoidance strategy based on the collision probability calculation results, realizing collaborative avoidance in a multi-ship environment. Thus, the accuracy and adaptability of ship collision prediction, as well as the navigation safety and navigation efficiency of ships, are significantly improved, the risk of ship collision accidents is effectively reduced, and the scientific nature and real-time nature of maritime traffic management are enhanced.

[0095] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A ship collision prediction method based on shipborne navigation data, characterized in that: The method comprises: For multi-ship interaction areas, multi-source interaction data are collected, wherein the multi-source interaction data includes ship navigation data, neighborhood radar data, and marine environment data; After correcting and fusing the multi-source interactive data, a multi-ship motion state matrix is ​​generated through relative motion analysis; Taking the multi-ship motion state matrix as input, constructing a multi-ship interactive game model, wherein the multi-ship interactive game model has a multi-ship interactive utility function embedded therein; Predicting ship trajectories based on the multi-ship interactive game model to generate multi-ship predicted trajectories, and correcting the multi-ship predicted trajectories based on real-time ocean environment data to generate a multi-ship future trajectory set; Calculating the collision probability according to the future trajectory set of the multiple ships, and building an avoidance strategy game model based on the collision probability calculation result; Based on the avoidance strategy game model, the avoidance strategy game is solved to generate a multi-ship coordinated avoidance strategy.

2. The ship collision prediction method based on shipborne navigation data according to claim 1, characterized in that: After the multi-source interaction data is corrected and fused, a multi-ship motion state matrix is ​​generated through relative motion analysis, including: Performing time synchronization and spatial alignment on the multi-source interaction data to generate standard multi-source interaction data; Based on the standard multi-source interaction data, a multi-ship motion state matrix is ​​generated; Calculate the relative position vector of each ship and its neighboring ships to determine the relative direction between the ships; Based on the relative direction, using the relative velocity vector, calculating the closest point in time and the closest point in distance; Performing dynamic change trend analysis on the closest point in time and the closest point in distance, generating a real-time evaluation index reflecting the degree of collision risk, and embedding the real-time evaluation index into the multi-ship motion state matrix; An environmental correction factor is introduced to dynamically update the multi-ship motion state matrix.

3. The ship collision prediction method based on shipborne navigation data according to claim 1, characterized in that: The multi-ship motion state matrix is ​​used as input to construct a multi-ship interactive game model, wherein the multi-ship interactive game model is embedded with a multi-ship interactive utility function, including: Taking the multi-ship motion state matrix as input, the multi-ship motion state matrix includes a plurality of key indicators, the plurality of key indicators including relative distance, relative speed, closest point in time, closest point in distance, heading difference and environmental correction factor; Based on the multiple key indicators, a multi-ship interaction utility function is defined, wherein the multi-ship interaction utility function includes a collision risk weight, a navigation efficiency weight, an avoidance cost weight, and a priority weight; Based on the multi-ship interaction utility function, the multi-ship interaction game model is constructed.

4. The ship collision prediction method based on shipborne navigation data according to claim 3, characterized in that: Based on the multi-ship interaction utility function, the multi-ship interaction game model is constructed, including: Based on the multi-ship interaction utility function, each ship is regarded as a game participant, and a multi-ship interaction game model framework is constructed; Based on non-cooperative game theory, a game strategy space is defined, wherein the game strategy space includes a heading adjustment range and a speed adjustment range; The multi-ship interactive game model framework and the game strategy space are integrated to generate the multi-ship interactive game model.

5. The ship collision prediction method based on shipborne navigation data according to claim 1, characterized in that: The ship trajectory prediction is performed based on the multi-ship interactive game model to generate multi-ship predicted trajectories, including: Inputting the multi-ship motion state matrix as the initial condition of the multi-ship interactive game model; The multi-ship interactive game model is used to solve the Nash equilibrium through dynamic iteration to generate the optimal interactive behavior strategy for multiple ships; Based on the multi-ship optimal interactive behavior strategy, multi-ship predicted trajectories are output, wherein each trajectory records the predicted position and velocity vector of a ship at multiple times in the future.

6. The ship collision prediction method based on shipborne navigation data according to claim 1, characterized in that: According to the future trajectory set of multiple ships, a collision probability calculation is performed, and based on the collision probability calculation result, an avoidance strategy game model is constructed, including: Taking the set of future trajectories of multiple ships as input, and gradually pairing the future trajectory of each ship with the trajectories of other ships based on the trajectory time series, to obtain multiple trajectory interference combinations; Based on the multiple trajectory interference combinations, the relative speed and relative direction between each pair of ships are calculated to generate a dynamic relative motion vector; According to the dynamic relative motion vector, a plurality of disturbance samples are extracted for simulation, the collision probability under each possible condition is calculated, and a global collision risk matrix is ​​generated; Each ship is regarded as a game participant, and all participants share the global collision risk matrix to generate the avoidance strategy game model.

7. The ship collision prediction method based on shipborne navigation data according to claim 6, characterized in that: According to the dynamic relative motion vector, multiple disturbance samples are extracted for simulation, and the collision probability under each possible condition is calculated, including: extracting a plurality of disturbance samples of each pair of future trajectories of the ships according to the dynamic relative motion vector; Using the closest point in time and the closest point in distance as risk indicators, define the collision risk function: Among them, P is the collision probability, TCPA is the closest point in time, CPA is the closest point in distance, and α and β are weight factors; The collision risk function is used to perform simulation calculations on the multiple disturbance samples to generate multiple collision probabilities.

8. The ship collision prediction method based on shipborne navigation data according to claim 1, characterized in that: Based on the avoidance strategy game model, the avoidance strategy game is solved to generate a multi-ship coordinated avoidance strategy, including: Defining a global optimization goal in the avoidance strategy game model, wherein the global optimization goal includes minimizing the risk of collision between ships and maximizing the overall navigation efficiency; Based on the global optimization goal, a distributed game solution is adopted, each ship independently calculates a preliminary avoidance plan, and updates it to the global optimal plan through multiple rounds of games to generate the multi-ship collaborative avoidance strategy.

Images