Ship type-based sailing trajectory prediction method, device, equipment and medium

Abstract

The application discloses a ship type-based sailing track prediction method and device, equipment and medium, and relates to the technical field of ship navigation and maritime traffic management. The method comprises the following steps: acquiring sailing data, environment data, other ship information corresponding to the ship, channel information corresponding to the ship, and a sailing track prediction model corresponding to the ship type; determining a target prediction model from three sub-prediction models based on the sailing data, the other ship information and the channel information; inputting the sailing data and the environment data into the target prediction model to obtain a position of the ship in a next time interval output by the target prediction model. The application aims to solve the problem of poor prediction accuracy in predicting the sailing track of the ship in the prior art, and improve the accuracy of the prediction of the sailing track of the ship.

Description

Technical Field

[0001] This application relates to the field of ship navigation and maritime traffic management technology, and in particular to a method, device, equipment and medium for predicting navigation trajectory based on ship type. Background Technology

[0002] With the rapid development of the maritime transport industry and the continuous growth of global shipping volume, competition for waterway resources has become increasingly fierce, leading to a continuous increase in the risk of ship collisions. In order to effectively reduce collision risks and improve the accuracy and efficiency of maritime traffic management, ship trajectory prediction has become one of the key research and application directions.

[0003] In existing technologies, a trajectory prediction model is trained using a large amount of ship operation data, and then the trained model is used to predict ship trajectories. However, this method considers only one factor, resulting in low model prediction accuracy. Summary of the Invention

[0004] In response to the aforementioned problems and technical requirements, the applicant has proposed a method, device, equipment, and medium for predicting navigation trajectories based on ship type, in order to solve the problem of poor prediction accuracy in the prior art when predicting ship navigation trajectories and improve the accuracy of ship navigation trajectory prediction.

[0005] This application provides a method for predicting navigation trajectories based on ship type. The method includes:

[0006] The system acquires the ship's navigation data, environmental data, information on other ships and waterways within a preset range corresponding to the ship's location, and a navigation trajectory prediction model corresponding to the ship's type. The navigation trajectory prediction model includes three sub-prediction models: a time series analysis model, a deep learning model, and a graph neural network model. The environmental data includes wind speed, water flow speed at the ship's location, and wave parameters at the ship's location. The navigation data includes the ship's position, speed, and heading.

[0007] Based on the navigation data, the other vessel information, and the waterway information, a target prediction model is determined from the three sub-prediction models;

[0008] The navigation data and environmental data are input into the target prediction model to obtain the ship position in the next time interval output by the target prediction model. The navigation trajectory prediction model is trained based on navigation data samples, environmental data samples and next ship position samples.

[0009] According to an embodiment of the application, the ship type-based voyage trajectory prediction method comprises the following steps:

[0010] Based on the other ship information and the waterway information, the voyage difficulty of the ship is determined;

[0011] If the voyage difficulty is greater than a preset difficulty, the graph neural network model is determined as the target prediction model;

[0012] If the voyage difficulty is less than or equal to the preset difficulty, the voyage change of the ship is determined based on the voyage data, and the target prediction model is determined from the time series analysis model and the deep learning model based on the voyage change.

[0013] According to an embodiment of the application, the ship type-based voyage trajectory prediction method comprises the following steps:

[0014] The change frequency, change trend of the voyage data and whether the change trend belongs to a linear change trend are determined;

[0015] If the change frequency is less than a preset frequency, and / or the change trend belongs to the linear change trend, the time series analysis model is determined as the target prediction model;

[0016] If the change frequency is greater than or equal to the preset frequency, and / or the change trend does not belong to the linear change trend, the deep learning model is determined as the target prediction model.

[0017] According to an embodiment of the application, the ship type-based voyage trajectory prediction method further comprises the following steps after determining the target prediction model from the three sub-prediction models:

[0018] A time period corresponding to the target prediction model is obtained;

[0019] From the voyage data and environmental data in a current time interval, corresponding voyage data and environmental data in the time period are extracted, wherein the time interval is greater than the time period;

[0020] The voyage data and environmental data are input into the target prediction model to obtain the corresponding ship position in the next time interval output by the target prediction model, comprising:

[0021] The corresponding navigation data and environmental data in the time period are input into the target prediction model to obtain the corresponding ship position in the next time period output by the target prediction model.

[0022] According to the ship type-based navigation trajectory prediction method of one embodiment of the present application, the navigation data and environmental data are input into the target prediction model to obtain the corresponding ship position in the next time interval output by the target prediction model, which includes:

[0023] The navigation data and the environmental data are input into the target prediction model, the ship motion features and ship operation features corresponding to the ship type are extracted based on the navigation data by the target prediction model, and the corresponding ship position in the next time interval is predicted and output based on the ship motion features, the ship operation features, and the environmental data.

[0024] According to the ship type-based navigation trajectory prediction method of one embodiment of the present application, the time periods corresponding to the three sub-prediction models are different.

[0025] According to the ship type-based navigation trajectory prediction method of one embodiment of the present application, the time series analysis model includes an ARIMA model.

[0026] The deep learning model includes any one of an LSTM network model and a GRU network model.

[0027] The present application also provides a ship type-based navigation trajectory prediction device, which includes:

[0028] The acquisition module is configured to acquire the navigation data of a ship in a current time interval, environmental data, other ship information and channel information in a preset range corresponding to the position of the ship, and a ship type corresponding to the ship type-based navigation trajectory prediction model to which the ship belongs, wherein the ship type-based navigation trajectory prediction model includes three sub-prediction models, the three sub-prediction models include a time series analysis model, a deep learning model, and a graph neural network model, the environmental data includes a wind speed, a water flow rate at the position of the ship, and a wave parameter at the position of the ship, and the navigation data includes a ship position, a speed, and a heading.

[0029] The determination module is configured to determine a target prediction model from the three sub-prediction models based on the navigation data, the other ship information, and the channel information.

[0030] The prediction module is configured to input the navigation data and the environmental data into the target prediction model to obtain the corresponding ship position in the next time interval output by the target prediction model, wherein the ship type-based navigation trajectory prediction model is trained based on a navigation data sample, an environmental data sample, and a next ship position sample.

[0031] The embodiment of the present application also provides an electronic device, including a memory, a processor and a computer program stored in the memory and executable on the processor, when the processor executes the program, the steps of the ship type based voyage trajectory prediction method according to any one of the above are implemented.

[0032] The embodiment of the present application also provides a non-transitory computer readable storage medium, having a computer program stored thereon, when the computer program is executed by a processor, the steps of the ship type based voyage trajectory prediction method according to any one of the above are implemented.

[0033] The ship type based voyage trajectory prediction method, device, equipment and medium provided by the embodiment of the present application, by obtaining the voyage data of the ship in the current time interval, the environment data, the other ship information and the channel information in the preset range corresponding to the position of the ship, and the voyage trajectory prediction model corresponding to the ship type of the ship; further, based on the voyage data, the other ship information and the channel information, the target prediction model is determined from the three sub-prediction models (the voyage trajectory prediction model includes three sub-models), it can be seen that the target prediction model corresponding to the overall situation of the ship itself and the external environment at the position is selected, the accuracy of the voyage trajectory output is ensured; finally, the voyage data and the environment data are input into the target prediction model, and the corresponding ship position in the next time interval output by the target prediction model is obtained, the ship itself is not only considered, but also the environment data of the environment where the ship is located, and the accurate prediction of the ship voyage trajectory is realized. BRIEF DESCRIPTION OF DRAWINGS

[0034] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the drawings needed to be used in the embodiment or prior art description will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the present application, and those skilled in the art can also obtain other drawings according to these drawings without creative labor.

[0035] Figure 1 is a flowchart of the ship type based voyage trajectory prediction method provided by the embodiment of the present application;

[0036] Figure 2 is a structural schematic diagram of the ship type based voyage trajectory prediction device provided by the embodiment of the present application;

[0037] Figure 3 is a structural schematic diagram of the electronic device provided by the embodiment of the present application. DETAILED DESCRIPTION

[0038] In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0039] The embodiment of the present application provides a ship type-based navigation track prediction method. The method can be applied in an intelligent terminal, can also be applied in a server, and can also be applied in a controller of a ship. Hereinafter, the method applied in the controller of the ship is taken as an example for description, some other descriptions in the embodiments of the present application are for example description, and are not used for limiting the protection scope of the present application, and then the descriptions are not repeated. The specific implementation of the method is as shown in Figure 1

[0040] In step 101, navigation data of a ship in a current time interval, environment data, other ship information and channel information in a preset range corresponding to a position of the ship, and a navigation track prediction model corresponding to a ship type to which the ship belongs are obtained.

[0041] The navigation track prediction model includes three sub-prediction models, and the three sub-prediction models include a time series analysis model, a deep learning model and a graph neural network model. The environment data includes a wind speed, a water flow rate at the position of the ship and a wave parameter at the position of the ship. The navigation data includes a ship position, a speed and a heading.

[0042] In step 102, a target prediction model is determined from the three sub-prediction models based on the navigation data, the other ship information and the channel information.

[0043] In step 103, the navigation data and the environment data are input into the target prediction model, and a corresponding ship position in a next time interval output by the target prediction model is obtained.

[0044] The navigation track prediction model is trained based on navigation data samples, environment data samples and next ship position samples.

[0045] The next ship position sample can also belong to the navigation data samples.

[0046] ​The ship type-based sailing trajectory prediction method provided by the embodiments of the present application comprises the following steps: obtaining sailing data of a ship in a current time interval, environment data, other ship information and channel information in a preset range corresponding to a position of the ship, and a sailing trajectory prediction model corresponding to a ship type to which the ship belongs; then, based on the sailing data, the other ship information and the channel information, a target prediction model is determined from three sub-prediction models (the sailing trajectory prediction model comprises three sub-models), and it can be seen that the target prediction model corresponding to the overall situation of the ship itself and the external environment at the position is selected, thereby ensuring the accuracy of the sailing trajectory output; finally, the sailing data and the environment data are input into the target prediction model, and a corresponding position of the ship in a next time interval output by the target prediction model is obtained, and the present application not only considers the running condition of the ship itself, but also considers the environment data of the environment where the ship is located, thereby realizing accurate prediction of the sailing trajectory of the ship.

[0047] In one specific embodiment, the time series analysis model comprises an ARIMA model.

[0048] The deep learning model comprises any one of an LSTM network model and a GRU network model.

[0049] The time series analysis model is a theory and method for establishing a mathematical model through curve fitting and parameter estimation according to time series data obtained by system observation. The ARIMA model is an autoregressive moving average model, which is a kind of time series analysis.

[0050] The deep learning model is a model constructed by deep learning. Deep learning specifically refers to machine learning based on deep neural network models and methods. The LSTM network model is a kind of time recurrent network model for solving long-term dependence problems. The GRU network model is an improved recurrent neural network model.

[0051] The graph neural network model refers to an algorithm model for learning graph structure data using a neural network, extracting and mining features and patterns in the graph structure data, and meeting the needs of clustering, classification, prediction, segmentation, generation and other graph learning tasks.

[0052] In one specific embodiment, the specific implementation of determining the target prediction model from the three sub-prediction models based on the sailing data, the other ship information and the channel information comprises the following steps:

[0053] Based on the other ship information and the channel information, the sailing difficulty of the ship is determined; in the case that the sailing difficulty is greater than a preset difficulty, the graph neural network model is determined as the target prediction model; in the case that the sailing difficulty is less than or equal to the preset difficulty, the sailing change of the ship is determined based on the sailing data, and the target prediction model is determined from the time series analysis model and the deep learning model based on the sailing change.

[0054] In one specific embodiment, the navigation change of the ship is determined based on the navigation data, and the specific implementation of the target prediction model is determined from the time series analysis model and the deep learning model based on the navigation change, which includes:

[0055] The change frequency, change trend, and whether the change trend belongs to a linear change trend of the navigation data are determined; in a case where the change frequency is determined to be less than a preset frequency, and / or the change trend belongs to a linear change trend, the time series analysis model is determined as the target prediction model; in a case where the change frequency is determined to be greater than or equal to the preset frequency, and / or the change trend does not belong to a linear change trend, the deep learning model is determined as the target prediction model.

[0056] Specifically, different target prediction models are used for different scenarios, and the target prediction model of the ship during navigation may be constant or may change at any time, and the adjustment of the target prediction model is specifically based on the change of the specific scenario.

[0057] Specifically, the other ship information is used to reflect whether the position of the ship is dense with ships, and the navigation information is used to reflect whether there is a complex channel interaction at the position of the ship, and the navigation difficulty of the ship is determined based on this. Specifically, the determination standard user can set according to the actual needs, and the present application does not make any limitation.

[0058] Specifically, for the training of the graph neural network model, the training sample data needs to be constructed into a graph structure, so that the graph neural network model captures the relative position and dynamic interaction features between ships, and based on this, a connection matrix or a distance graph is constructed, and then the spatial and temporal dependence is extracted using GNN, and finally the graph neural network model is obtained. When predicting using the graph neural network model, the navigation data and the environment data are mapped to the graph structure, and then the prediction of the corresponding ship position in the next time interval is performed.

[0059] Specifically, for the case where the change frequency is less than the preset frequency, and / or the change trend belongs to a linear change trend, short-term prediction (for example, predicting data within 5 minutes in the future) is adopted, and the time series analysis model is used for prediction. For the case where the change frequency is greater than or equal to the preset frequency, and / or the change trend does not belong to a linear change trend (non-linear change trend), long-term prediction (for example, predicting data within 10 minutes in the future) is adopted, and the deep learning model is used for prediction.

[0060] The following model training method is used for training of the deep learning model:

[0061] The loss function is defined: taking the prediction position deviation as the consideration, the mean square error (MSE) can be used as the loss function, and the prediction error of the speed and the heading can be considered by weighting as needed.

[0062] Optimization algorithm: trained with Adam adaptive optimizer, using iterative parameter updates to converge to global or local optimal solution.

[0063] Early stopping and regularization: monitor the error on the validation set during training, if the validation error does not decrease after a certain number of iterations, stop training (early stopping strategy); prevent model overfitting through L2 regularization or dropout strategy.

[0064] Hyperparameter search: use Bayesian optimization to automatically tune hyperparameters (such as learning rate, number of hidden layers, number of neurons) to improve model performance.

[0065] Specifically, the training sample data of the target prediction model needs to be preprocessed, as follows:

[0066] Remove duplicate values: for records with the same timestamp appearing multiple times, only keep the first valid record to avoid data redundancy.

[0067] Remove outliers: remove records that deviate from the reasonable range through statistical analysis and empirical threshold, such as records with speed significantly exceeding the normal range (e.g. more than 50 knots) or abnormal changes in latitude and longitude.

[0068] Interpolate missing values: linear interpolation of missing data to ensure data continuity and integrity.

[0069] Furthermore, based on the ship type, the training sample data is divided to build corresponding trajectory prediction models for different types of ships, thereby reducing the adverse effects of characteristic differences between different types on model prediction accuracy. Moreover, each ship type corresponds to three sub-prediction models.

[0070] Among them, the ship types include oil tankers, container ships, bulk carriers and passenger ships, etc.

[0071] For each ship type, extract time series features (features corresponding to the time period) from the training sample data, including historical trajectory points (longitude and latitude), speed, heading, and environmental data, etc., to construct a multi-dimensional feature space. Then, train the model based on this to predict the ship's position (longitude and latitude) time series within a certain time period.

[0072] Among them, the output results can be presented in a standardized data interface (such as JSON format) or a visual interface to facilitate the maritime department, ship drivers or third-party applications to query, analyze and decision support for the prediction results.

[0073] The prediction results can be output as a JSON format data file, containing the following fields:

[0074] {

[0075] "timestamp_predict_start": "2024-01-01T10:00:00Z",

[0076] "predicted_points": [

[0077] {"timestamp": "2024-01-01T10:10:00Z", "latitude": Latitude XX.XXXX, "longitude": Longitude YY.YYYY},

[0078] {"timestamp": "2024-01-01T10:25:00Z", "latitude": Latitude XX.XXXX, "longitude": Longitude YY.YYYY}

[0079] ],

[0080] "latitude prediction_error_estimate": 0.00002 / / unit in °

[0081] "longitude prediction_error_estimate": 0.00003 / / unit in °

[0082] "trajectory prediction_error_estimate": 0.2 / / unit in nautical miles, converted from latitude and longitude prediction error

[0083] }

[0084] Specifically, after obtaining the prediction results, the real ship historical track data is used to evaluate the accuracy of the prediction results. The root mean square error (RMSE), mean absolute error (MAE), and average displacement deviation (ADE) are used to quantify the performance of the model. According to the evaluation results, the model is continuously optimized, including introducing more environmental factors, improving the network structure or adjusting the parameters, etc. to continuously improve the prediction accuracy and robustness.

[0085] Finally, during the use of the trajectory prediction model, when the sailing sea area changes significantly, the model can be incrementally trained or fine-tuned using newly collected AIS data on a regular basis (such as every week), so that the model can continuously adapt to the dynamically changing shipping conditions, thereby ensuring long-term prediction accuracy.

[0086] In one specific embodiment, based on the sailing data, other ship information and channel information, the specific implementation of the target prediction model is determined from the three sub-prediction models, which includes:

[0087] The correspondence between the navigation data, other ship information, and channel information and the sub-prediction models is created in advance, and the target prediction model corresponding to the navigation data, other ship information, and channel information is obtained based on the correspondence.

[0088] In one embodiment, after the target prediction model is determined from the three sub-prediction models, the time period corresponding to the target prediction model is obtained; the corresponding navigation data and environmental data in the time period are extracted from the navigation data and environmental data in the current time interval; and the corresponding ship position in the next time period output by the target prediction model is obtained by inputting the corresponding navigation data and environmental data in the time period into the target prediction model.

[0089] The time interval is greater than the time period.

[0090] In one embodiment, the time periods corresponding to the three sub-prediction models are different.

[0091] The data obtained in the time interval is used for determination of the target prediction model, and the time periods of different target prediction models are different, which are the prediction periods (or time windows) of the models. Of course, the size of the period can be set by the user according to the actual needs, and the present application does not make any limit.

[0092] In one embodiment, the specific implementation of inputting the navigation data and environmental data into the target prediction model to obtain the corresponding ship position in the next time interval output by the target prediction model includes:

[0093] The navigation data and environmental data are input into the target prediction model, the ship motion characteristics and ship operation characteristics corresponding to the ship type are extracted based on the navigation data by the target prediction model, and the corresponding ship position in the next time interval is predicted and output based on the ship motion characteristics, ship operation characteristics, and environmental data.

[0094] For example, the hull of an oil tanker is generally large, has high load capacity, large inertia, and slow acceleration and deceleration response. The sailing speed is moderate (usually 10-15 knots), the sailing trajectory is relatively stable, and the turning action is slow. In the process of sailing operation, in order to reduce the influence of oil liquid sloshing on the stability of the ship, emergency turning or sudden stopping is avoided as much as possible. For example, the sailing speed of a container ship is relatively high (usually 18-25 knots), and the maneuverability is strong, so it often needs to turn quickly or change speed in a crowded channel to avoid other ships.

[0095] The application can significantly improve the reliability of model input data by performing preprocessing operations on training sample data during model training. According to independent modeling based on the differences between ship types, the trajectory prediction accuracy and adaptability of multiple types of ships are significantly improved. Moreover, the environment factors are incorporated into the model, which improves the stability and practicality of the prediction results under complex sea conditions. In addition, different scenarios correspond to different models, which can be applied to various scenarios, and the prediction accuracy and robustness can still be maintained in complex navigation scenarios.

[0096] The embodiment of the application also provides a ship type-based navigation trajectory prediction device. The specific implementation of the device can refer to the description of the ship type-based navigation trajectory prediction method, and repeated parts will not be described here again, such as Figure 2 As shown in the figure, the device comprises:

[0097] The acquisition module 201 is configured to acquire navigation data of a ship in a current time interval, environment data, other ship information and channel information in a preset range corresponding to a position of the ship, and a ship type corresponding navigation trajectory prediction model of the ship, wherein the navigation trajectory prediction model comprises three sub-prediction models, and the three sub-prediction models comprise a time series analysis model, a deep learning model and a graph neural network model. The environment data comprises a wind speed, a water flow rate at the position of the ship and a wave parameter at the position of the ship. The navigation data comprises a ship position, a speed and a heading.

[0098] The determination module 202 is configured to determine a target prediction model from the three sub-prediction models based on the navigation data, the other ship information and the channel information.

[0099] The prediction module 203 is configured to input the navigation data and the environment data into the target prediction model to obtain a corresponding ship position in a next time interval output by the target prediction model, wherein the navigation trajectory prediction model is trained based on a navigation data sample, an environment data sample and a next ship position sample.

[0100] In one specific embodiment, the determination module 202 is configured to determine a navigation difficulty of the ship based on the other ship information and the channel information. In a case where the navigation difficulty is greater than a preset difficulty, the graph neural network model is determined as the target prediction model. In a case where the navigation difficulty is less than or equal to the preset difficulty, a navigation change of the ship is determined based on the navigation data, and the target prediction model is determined from the time series analysis model and the deep learning model based on the navigation change.

[0101] In one specific embodiment, the determining module 202 is used to determine the frequency of change, the trend of change, and whether the trend of change is a linear trend of change of navigation data; if the frequency of change is less than a preset frequency and / or the trend of change is a linear trend of change, the time series analysis model is determined as the target prediction model; if the frequency of change is greater than or equal to the preset frequency and / or the trend of change is not a linear trend of change, the deep learning model is determined as the target prediction model.

[0102] In one specific embodiment, the determining module 202 is further configured to obtain the time period corresponding to the target prediction model; extract the navigation data and environmental data corresponding to the time period from the navigation data and environmental data within the current time interval, wherein the time interval is greater than the time period; the prediction module 203 is configured to input the navigation data and environmental data corresponding to the time period into the target prediction model to obtain the ship position corresponding to the next time period output by the target prediction model.

[0103] In one specific embodiment, the prediction module 203 is used to input navigation data and environmental data into the target prediction model, extract ship motion features and ship operation features corresponding to the ship type based on the navigation data, and predict and output the corresponding ship position in the next time interval based on the ship motion features, ship operation features and environmental data.

[0104] In one specific embodiment, the three sub-prediction models correspond to different time periods.

[0105] In one specific embodiment, the time series analysis model includes: an ARIMA model; the deep learning model includes: either an LSTM network model or a GRU network model.

[0106] Figure 3 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 3 As shown, the electronic device may include a processor 301, a communications interface 302, a memory 303, and a communication bus 304. The processor 301, communications interface 302, and memory 303 communicate with each other via the communication bus 304. The processor 301 can call logical instructions from the memory 303 to execute a navigation trajectory prediction method based on ship type.

[0107] In addition, the logic instructions in the memory 303 described above can be implemented in the form of a software function unit and sold or used as an independent product, and can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the parts that contribute to the prior art or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0108] In another aspect, the present application also provides a computer program product, which comprises a computer program stored on a non-transitory computer readable storage medium, and the computer program comprises program instructions, and when the program instructions are executed by a computer, the computer can execute the ship type based voyage trajectory prediction method provided by the above method.

[0109] In another aspect, the present application also provides a non-transitory computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the ship type based voyage trajectory prediction method provided by the above embodiments.

[0110] The device embodiments described above are illustrative, wherein the units illustrated as separate components can or can not be physically separated, and the components illustrated as units can or can not be physical units, i.e., they can be located in one place, or distributed on multiple network units. Part or all of the modules can be selected to achieve the purpose of the embodiment according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0111] From the description of the above embodiments, those skilled in the art can clearly understand that the embodiments can be implemented by means of software and the necessary general hardware platform, and of course, can also be implemented by hardware. Based on such understanding, the technical solutions described above essentially or the parts that contribute to the prior art can be embodied in the form of a software product. The computer software product can be stored in a computer readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments.

[0112] Finally, it should be noted that the above is only the preferred embodiment of the present application, and the present application is not limited to the above embodiments. Other improvements and changes that can be directly derived or thought of by those skilled in the art without departing from the spirit and concept of the present application should be considered to be included in the protection scope of the present application.

Claims

1. A method for predicting navigation trajectories based on ship type, characterized in that, The method includes: The system acquires the ship's navigation data, environmental data, information on other ships and waterways within a preset range corresponding to the ship's location, and a navigation trajectory prediction model corresponding to the ship's type. The navigation trajectory prediction model includes three sub-prediction models: a time series analysis model, a deep learning model, and a graph neural network model. The environmental data includes wind speed, water flow speed at the ship's location, and wave parameters at the ship's location. The navigation data includes the ship's position, speed, and heading. Based on the navigation data, the other vessel information, and the waterway information, a target prediction model is determined from the three sub-prediction models; The navigation data and environmental data are input into the target prediction model to obtain the ship position in the next time interval output by the target prediction model. The navigation trajectory prediction model is trained based on navigation data samples, environmental data samples and next ship position samples. Specifically, based on the navigation data, the other vessel information, and the waterway information, a target prediction model is determined from three sub-prediction models, including: Based on the other ship information and the waterway information, the navigation difficulty of the ship is determined; If the navigation difficulty is determined to be greater than the preset difficulty, the graph neural network model is selected as the target prediction model. If the navigation difficulty is determined to be less than or equal to the preset difficulty, the navigation changes of the ship are determined based on the navigation data, and the target prediction model is determined from the time series analysis model and the deep learning model based on the navigation changes.

2. The navigation trajectory prediction method based on ship type according to claim 1, characterized in that, Based on the navigation data, the navigation changes of the vessel are determined, and based on the navigation changes, the target prediction model is determined from the time series analysis model and the deep learning model, including: Determine the frequency of change, the trend of change, and whether the trend of change is linear in the navigation data; If it is determined that the frequency of change is less than the preset frequency, and / or the trend of change belongs to the linear trend of change, the time series analysis model is determined as the target prediction model; If the frequency of change is greater than or equal to the preset frequency, and / or the trend of change does not belong to the linear trend of change, the deep learning model is determined as the target prediction model.

3. The navigation trajectory prediction method based on ship type according to any one of claims 1-2, characterized in that, After determining the target prediction model from the three sub-prediction models, the following steps are also included: Obtain the time period corresponding to the target prediction model; From the navigation data and environmental data within the current time interval, extract the corresponding navigation data and environmental data within the time period, wherein the time interval is greater than the time period.

4. The navigation trajectory prediction method based on ship type according to any one of claims 1-2, characterized in that, By inputting navigation data and environmental data into the target prediction model, the ship's position in the next time interval output by the target prediction model is obtained, including: The navigation data and environmental data are input into the target prediction model. The target prediction model extracts ship motion features and ship operation features corresponding to the ship type based on the navigation data. Based on the ship motion features, the ship operation features, and the environmental data, the target prediction model predicts and outputs the corresponding ship position in the next time interval.

5. The navigation trajectory prediction method based on ship type according to any one of claims 1-2, characterized in that, The three sub-prediction models correspond to different time periods.

6. The navigation trajectory prediction method based on ship type according to any one of claims 1-2, characterized in that, The time series analysis model includes: the ARIMA model; The deep learning model includes either the LSTM network model or the GRU network model.

7. A navigation trajectory prediction device based on ship type, characterized in that, The device includes: The acquisition module is used to acquire the ship's navigation data, environmental data, information on other ships and waterways within a preset range corresponding to the ship's location, and a navigation trajectory prediction model corresponding to the ship's type. The navigation trajectory prediction model includes three sub-prediction models: a time series analysis model, a deep learning model, and a graph neural network model. The environmental data includes wind speed, water flow speed at the ship's location, and wave parameters at the ship's location. The navigation data includes the ship's position, speed, and heading. The determination module is used to determine the target prediction model from the three sub-prediction models based on the navigation data, the other ship information and the waterway information; The prediction module is used to input navigation data and environmental data into the target prediction model to obtain the ship position in the next time interval output by the target prediction model. The navigation trajectory prediction model is trained based on navigation data samples, environmental data samples and next ship position samples. The determination module is used to determine the navigation difficulty of the vessel based on the other vessel information and the waterway information; if the navigation difficulty is determined to be greater than a preset difficulty, the graph neural network model is determined as the target prediction model; if the navigation difficulty is determined to be less than or equal to the preset difficulty, the navigation change of the vessel is determined based on the navigation data, and the target prediction model is determined from the time series analysis model and the deep learning model based on the navigation change.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the navigation trajectory prediction method based on ship type as described in any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the navigation trajectory prediction method based on ship type as described in any one of claims 1 to 6.

Images