Ship navigation time estimation method, system, equipment, medium and product
By constructing a VDTTE model, deep learning technology is used to capture the spatial and temporal correlation and time dependence in the ship trajectory data, and comprehensively considering a variety of environmental factors, the problem of low accuracy of ship navigation time estimation in the existing technology is solved, and higher estimation accuracy and transportation efficiency are achieved.
Patent Information
- Application Number
- CN202411884625.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-20
- Publication Date
- 2025-04-11
AI Technical Summary
The existing ship navigation time estimation methods have low accuracy, fail to fully utilize the spatiotemporal correlation and time dependence in trajectory data, and do not consider the impact of multiple environmental factors on navigation time.
By acquiring the ship's trajectory data, geometric features and meteorological data, a ship's depth navigation time estimation (VDTTE) model including spatiotemporal modules, attribute modules and multi-task modules is constructed, and a deep learning technology is used to capture spatiotemporal correlations and time dependencies are comprehensively considered.
It significantly improves the accuracy of ship navigation time estimation, can predict navigation time more accurately, reduce transportation costs, and improve transportation efficiency and customer satisfaction.
Smart Images
Figure CN119807709A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of shipping, and in particular to a method, system, device, medium and product for estimating ship sailing time. Background Art
[0002] Ocean transportation plays an important role in global cargo transportation. As the core mode of international logistics, its efficiency and reliability are crucial to the smooth operation of the global economy. Accurate ship travel time estimation (TTE) can not only optimize port scheduling and management, but also effectively reduce transportation costs, improve transportation efficiency and customer satisfaction.
[0003] In response to the TTE problem of ships, Alessandrini et al. optimized the path-finding algorithm to estimate the arrival time based on historical ship trajectory data; Park et al. used reinforcement learning to find the possible trajectory of the ship, replacing the traditional path-finding algorithm and improving the prediction accuracy of the arrival time; Ogura et al. predicted the arrival time of the ship more accurately by analyzing the future weather conditions that the ship may encounter during the voyage; Kwun et al. not only considered the weather, but also evaluated the impact of sea conditions based on the geographical location, further improving the accuracy of the prediction.
[0004] In general, most of the existing TTE methods for ships are isolated studies, or rely on pathfinding algorithms to assume a trajectory. However, pathfinding algorithms often have huge computational complexity, and the path found may still deviate from the actual situation, resulting in large errors in time navigation estimation and low accuracy. In addition, the current TTE methods for ships with determined trajectories still mainly rely on traditional methods, without fully combining deep learning technology, thus failing to fully utilize the spatiotemporal correlation and time dependence in the trajectory data, and failing to consider the impact of various environmental factors on the ship's sailing time, which also leads to low accuracy in time navigation estimation. Summary of the invention
[0005] The purpose of this application is to provide a ship sailing time estimation method, system, device, medium and product to solve the problem of low accuracy of ship sailing time evaluated by the existing ship TTE method.
[0006] To achieve the above objectives, this application provides the following solutions:
[0007] In a first aspect, the present application provides a method for estimating ship sailing time, comprising:
[0008] Acquiring trajectory data, ship geometry and meteorological data; the trajectory data is obtained based on the ship AIS data; the ship geometry includes the length, width, maximum draft and cargo type of the ship; the meteorological data includes wind speed, wind direction, wave height and visibility;
[0009] Constructing a VDTTE model; the VDTTE model includes a spatiotemporal module, an attribute module and a multi-task module;
[0010] Inputting the trajectory data into the spatiotemporal module, and inputting the ship geometry and the meteorological data into the attribute module;
[0011] The output data of the spatiotemporal module and the attribute module are transmitted to the multi-task module to determine the estimation result of the ship's sailing time.
[0012] In a second aspect, the present application provides a ship sailing time estimation system, comprising:
[0013] A data acquisition module, used to obtain trajectory data, ship geometry and meteorological data; the trajectory data is obtained based on the ship AIS data; the ship geometry includes the length, width, maximum draft and cargo type of the ship; the meteorological data includes wind speed, wind direction, wave height and visibility;
[0014] A model building module, used to build a VDTTE model; the VDTTE model includes a spatiotemporal module, an attribute module and a multi-task module;
[0015] The result output module is used to input the trajectory data into the space-time module, input the ship's geometric characteristics and the meteorological data into the attribute module, transmit the output data of the space-time module and the attribute module to the multi-task module, and determine the ship's sailing time estimation result.
[0016] In a third aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement any of the ship sailing time estimation methods described above.
[0017] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-described methods for estimating the ship's voyage time.
[0018] In a fifth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements any of the above-mentioned ship navigation time estimation methods.
[0019] According to the specific embodiments provided in this application, this application discloses the following technical effects:
[0020] This application obtains trajectory data based on the ship's Automatic Identification System (AIS) data, and obtains the ship's geometric characteristics and meteorological data corresponding to each trajectory data, so as to fully consider the impact of various environmental factors on the ship's navigation time. At the same time, by constructing a vessel deep travel time estimation (VDTTE) model including a spatiotemporal module, an attribute module and a multi-task module, and inputting the above-collected data into different modules of the VDTTE model, the spatiotemporal correlation and time dependence in the trajectory data are fully utilized, and the accuracy of time navigation estimation is further improved. BRIEF DESCRIPTION OF THE DRAWINGS
[0021] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0022] Figure 1 A flow chart of a method for estimating ship sailing time provided in one embodiment of the present application;
[0023] Figure 2 A schematic diagram of a VDTTE model provided in an embodiment of the present application;
[0024] Figure 3 A schematic diagram of the structure of a ship sailing time estimation system provided in one embodiment of the present application;
[0025] Figure 4 A schematic diagram of a computer device provided in accordance with an embodiment of the present application. DETAILED DESCRIPTION
[0026] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0027] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.
[0028] The present application embodiment provides a method for estimating the sailing time of a ship, which is executed by a computer device, and can be executed by a computer device such as a terminal or a server alone, or can be executed by a terminal and a server together. In the present application embodiment, Figure 1 As shown, the method includes the following steps.
[0029] A method for estimating a ship's sailing time, the method comprising:
[0030] S1: Acquire trajectory data, ship geometry and meteorological data; the trajectory data is obtained based on the ship AIS data; the ship geometry includes the length, width, maximum draft and cargo type of the ship; the meteorological data includes wind speed, wind direction, wave height and visibility.
[0031] S2: Constructing a VDTTE model; the VDTTE model includes a spatiotemporal module, an attribute module and a multi-task module.
[0032] S3: Input the trajectory data into the space-time module, and input the ship geometric features and the meteorological data into the attribute module.
[0033] S4: The output data of the spatiotemporal module and the attribute module are transmitted to the multi-task module to determine the estimation result of the ship's sailing time.
[0034] In an exemplary embodiment, unlike land transportation, which can directly obtain a large amount of GPS trajectory data, ship AIS data is often provided in the form of discrete points and needs to be processed to be converted into trajectory data. In addition, in order to better predict the sailing time of the ship, the trajectory data needs to be combined with other data, such as sea conditions (weather data), ship information (ship geometry characteristics), etc.
[0035] S1 can be replaced by the following steps.
[0036] S11: classify the ship AIS data according to the Maritime Mobile Service Identity (MMSI) code, and then sort the classified ship AIS data according to the timestamp to determine the AIS data point sequence sorted according to the timestamp; one ship AIS data is one ship AIS data point.
[0037] S12: Based on the AIS data point sequence, determine the ship status according to the Status field of each ship AIS data point, and remove AIS data points with a ship status of "not underway".
[0038] S13: Segment the discontinuous point sequence to obtain multiple trajectory data, and for each trajectory data, calculate the shortest surface distance between two ship AIS data points, and record the ship geometric features and meteorological data corresponding to each trajectory data.
[0039] Furthermore, the steps to obtain trajectory data are as follows:
[0040] First, define a trajectory T as a sequence of points, that is, T = {p0, p1, ..., p n}, where each point p i (i=0,1,...,n) includes at least three attributes: longitude p i .lng, latitude p i .lat and timestamp p i .ts.
[0041] By classifying the AIS raw data according to the MMSI code and then sorting them by timestamp, we can get a sequence of AIS data points for each ship sorted by timestamp. However, this point sequence cannot be used directly as a ship trajectory, because a ship will not always be in a sailing state. It is necessary to judge the ship status based on the Status field of the AIS data, remove the AIS data points whose ship status is not sailing, and split the discontinuous point sequence to obtain multiple trajectory data.
[0042] For each trajectory data, the Haversine method is used to calculate the shortest surface distance between two points and save it. The distance between the two points will be used as an important spatial feature during the training process.
[0043] In this embodiment, the input (trajectory data) of the spatiotemporal module includes the timestamp, longitude and latitude information of the AIS data point in flight, and also includes the calculated shortest distance on the curved surface.
[0044] In addition, non-temporal and spatial factors that may affect the ship's sailing time in each trajectory must be recorded, including the ship's geometric characteristics (including the ship's length, width, maximum draft, ship's cargo type and other factors that will affect the ship's navigation capability), weather conditions in the sea area (including wind speed, wind direction, wave height, visibility and other sea conditions information), etc.
[0045] The properties of the ship's geometric factors determine the ship's sailing ability: the type of cargo determines the maximum load, the length and width affect the sailing stability, and the draft determines the range of waters in which the ship can sail. Weather conditions directly affect the ship's sailing speed, wind speed and direction determine the sailing direction, wave height and visibility affect sailing stability, and extreme weather may even cause the ship to change its sailing plan. Ultimately, these factors will affect the total sailing time of the ship, so this application conducts a more in-depth analysis of these complex factors.
[0046] Furthermore, geometric feature data can be directly obtained from AIS data, while sea state data needs to be obtained from other meteorological data sources.
[0047] In an exemplary embodiment, Figure 2 As shown, the spatiotemporal module includes a geographic convolutional layer and a long short-term memory layer connected in sequence.
[0048] The attribute module includes a multi-layer perceptron and a first connection layer connected in sequence; the first connection layer is connected to the long short-term memory layer.
[0049] The multi-layer perceptron is:
[0050] h1=σ(W1x v +b1),h2=σ(W2h1+b2),…,h n-1 =σ(W n-1 h n-2 +b h-1 )
[0051] e v =W n h n-1 +b n
[0052] Among them, x v is the category feature, W i is the weight matrix of each i layer, b i is the bias term for each i-th layer, i=1,2,3...n, σ(·) is the activation function, e v is the final embedding representation of category v.
[0053] The multi-task module includes an attention mechanism layer, a second connection layer, a residual connection layer and a fully connected layer; the attention mechanism layer is connected to the long short-term memory layer and the first connection layer respectively; the second connection layer is also connected to the first connection layer; the fully connected layer and the residual connection layer jointly output the ship sailing time estimation result.
[0054] In practical applications, trajectory data passes through the geographic convolution layer and long short-term memory layer in the spatiotemporal module in turn to capture the spatiotemporal correlation and time dependency in the trajectory data. The geographic convolution layer is a graph convolutional neural network (GCN), and the long short-term memory layer is a long short-term memory network (LSTM).
[0055] The geographic convolution layer performs convolution operations on the input trajectory data (vector) to obtain a feature map containing the spatial features of each local path (the path from one point to the next point is a local path). In fact, it is a vector, each element of which represents the spatial information of one of the local paths on the entire trajectory. This vector is then input to the LSTM layer. The LSTM layer consists of two groups of LSTM units. Each LSTM receives the output of the previous LSTM unit, the attribute vector, and an element in the spatial information vector (the spatial information of a local path), and calculates them with the learnable weights respectively to obtain a vector containing spatiotemporal information as the final output.
[0056] The ship's geometric features and meteorological data are input into the Multilayer Perceptron (MLP) in the attribute module. MLP can significantly improve the expression ability of category features.
[0057] MLP can be seen as a nonlinear function that transforms input features layer by layer through multiple fully connected layers and activation functions. Assume that the input is a "one-hot" encoded category feature x v , the mathematical representation of MLP is as follows:
[0058] h1=σ(W1x v +b1),h2=σ(W2h1+b2),…,h n-1 =σ(W n-1 h n-2 +b n-1 )
[0059] e v =W n h n-1 +b n
[0060] in, represents the weight matrix of each layer, is d i ×d i-1 The set of real number matrices, that is, the weight matrix W is d i Line d i-1 The elements are all real numbers, di is the number of neurons in the i-th layer of MLP, and b iis the bias term, σ(·) is the activation function, such as the ReLU activation function, e v is the final embedding representation of category v. Through multiple layers of linear transformation and nonlinear activation, MLP is able to learn complex feature interactions and higher-order relationships.
[0061] MLP introduces nonlinearity through the activation function σ, which can model the complex distribution of category features, so that category features are not limited to linear mapping, but can generate richer nonlinear representations. Through multi-layer stacking, MLP can capture high-order category feature interactions, which has significant advantages in processing complex category features.
[0062] In the attribute module, different sea conditions such as wind speed, waves, etc. and the geometric features of the ship are encoded as vectors to represent the sea conditions and the geometric features of the ship. After that, the vector is used as input, and the MLP extracts its features. The output is still a vector, which is output to the connection layer. The connection layer fuses the vectors of the extracted sea conditions and the geometric features of the ship to obtain a vector that can represent the characteristics of non-spatial factors in navigation as the output of the attribute module.
[0063] Finally, the multi-task module of VDTTE uses the attention mechanism to selectively focus on different parts of the input sequence according to the input sequence (output of the spatiotemporal module) and the relevance of the current prediction task (output of the attribute module), and then uses a connection layer to combine the output of the attribute module with the output of the attention mechanism, and uses a residual fully connected block to obtain the result.
[0064] The role of the attention mechanism in the multi-task module is to pay more attention to the key local paths. Finally, the fully connected layer will generate direct prediction results for each local path, and the residual connection layer will generate the final prediction results for the entire trajectory. In this embodiment, the only data indicator of interest is the time prediction result of the entire trajectory output by the residual connection layer; however, in actual training, it is necessary to know the time prediction result of each local path for optimization, so the fully connected layer is still retained in the framework.
[0065] The multi-task module comprehensively considers the outputs of the spatiotemporal module and the attribute module to obtain the final navigation time estimation result.
[0066] This embodiment also conducts an experimental evaluation on the ship sailing time estimation method proposed in this application, as follows:
[0067] A. Experimental setup.
[0068] 1) Experimental environment: Table 1 is the experimental environment table. The hardware and software environment used in this experiment are shown in Table 1.
[0069] Table 1
[0070] Components Details CPU AMDEPYC740224-CoreProcessor GPU Nvidia GeForce RTX 4090 OS Ubuntu 22.04LTS GPUDriver 525.105.17 CUDA 11.8 cuDNN 8.6.0 PyTorch 2.4.1
[0071] 2) Dataset: The AIS dataset used in the experiment comes from MarineCadastre. This dataset provides AIS data of ships in coastal waters of the United States since 2009, and since 2015, the dataset divides the data by date, and the data of each day is stored in a separate CSV file, and the possible duplicate AIS data has been filtered according to the minimum time interval of ten minutes.
[0072] This experiment used the data from January 1, 2023 to December 31, 2023 as the training data set, a total of 365 CSV files, with 400 million AIS data from about 20,000 ships; the data from January 1, 2024 to January 31, 2024 in the data set was also used as the evaluation data set, a total of 31 CSV files, with 30 million AIS data from about 12,000 ships. In addition to the ship's location information such as longitude and latitude, the data set also provides information such as the ship's SOG speed, heading, length and width of the ship, draft depth, and cargo type that may affect the ship's sailing time. In addition, this experiment also collected sea condition data for the entire year of 2023 off the coast of the United States, including wind speed, wind direction, wave height, visibility, and other information. These data are processed to obtain trajectory data, which are ultimately used for model training and evaluation.
[0073] 3) Evaluation index: Let the navigation start time of trajectory j be S j , assuming the actual end time of the voyage is A j , the estimated end time is E j This example uses the following parameters to evaluate the performance of the model:
[0074]
[0075] B. Experimental methods.
[0076] 1) Baseline method: Due to the lack of a typical baseline method, this embodiment uses a traditional and simple TTE calculation method as the baseline method: that is, it is assumed that the ship travels in a straight line, and the distance between two points is the shortest distance on the surface calculated by the Haversine method; the speed is the average speed of the historical data of the geographical location; and finally the sum of the estimated sailing time of multiple straight line segments is calculated as the estimated sailing time of the entire trajectory.
[0077] 2) Spatio-temporal method: Using the spatio-temporal module of the VDTTE model alone, without considering the impact of external factors on the flight time, only uses deep learning technology to analyze the spatio-temporal dependency of the trajectory and learns the estimated flight time of the entire trajectory. This is called a relatively simple spatio-temporal (ST) method.
[0078] 3) Spatio-Temporal Embedding Method: In order to verify the advantages of the MLP method over the embedding layer method, the MLP of the attribute module in VDTTE is replaced by the embedding layer method, and the same analysis is performed on the external factors, which is called the Spatio-Temporal Embedding (STE) method.
[0079] C. Experimental results.
[0080] 1) Method comparison: Table 2 is a performance comparison table of different methods. It is not difficult to see from the experimental results in Table 2 that the result obtained by simply using the historical average speed and the straight-line distance between two points to estimate the ship's sailing time is very inaccurate. The VDTTE method performs well on this dataset. Compared with the baseline method, both MAE and RMSE have been significantly improved, and MAPE has also been greatly reduced. This shows that the deep learning method VDTTE can better learn the spatiotemporal dependence of trajectory data and more comprehensively consider other factors that affect the ship's sailing time, thereby more accurately estimating the ship's sailing time.
[0081] 2) Ablation experiment: In order to verify the improvement that the MLP method can bring in VDTTE, we conducted an ablation experiment to compare the performance of VDTTE and STE methods. The results in Table 2 show that the STE method is significantly better than the ST method, and the VDTTE method using the MLP method is significantly better than the STE method, which confirms the advantage of the MLP method in dealing with complex external factors in the attribute module.
[0082] The results show that the VDTTE model proposed in this application achieves higher estimation accuracy and is superior to other existing methods.
[0083] Table 2
[0084] Model MAPE(%) RMSE (hours) MAE(hours) Baseline 33.12 21.39 15.28 ST 21.93 13.97 10.12 STE 16.10 11.30 7.43 VDTTE 14.72 10.88 6.80
[0085] Based on the same inventive concept, the embodiment of the present application also provides a ship sailing time estimation system for implementing the ship sailing time estimation method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more ship sailing time estimation system embodiments provided below can refer to the limitations of the ship sailing time estimation method above, and will not be repeated here.
[0086] In an exemplary embodiment, Figure 3 As shown, a ship sailing time estimation system is provided, comprising:
[0087] The data acquisition module is used to obtain trajectory data, ship geometry and meteorological data; the trajectory data is obtained based on the ship AIS data; the ship geometry includes the length, width, maximum draft and cargo type of the ship; the meteorological data includes wind speed, wind direction, wave height and visibility.
[0088] The model building module is used to build a VDTTE model; the VDTTE model includes a spatiotemporal module, an attribute module and a multi-task module.
[0089] The result output module is used to input the trajectory data into the space-time module, input the ship's geometric characteristics and the meteorological data into the attribute module, transmit the output data of the space-time module and the attribute module to the multi-task module, and determine the ship's sailing time estimation result.
[0090] In an exemplary embodiment, the data acquisition module includes a geometric feature acquisition unit, a meteorological data collection unit and a trajectory data acquisition unit; wherein the trajectory data acquisition unit includes:
[0091] The AIS data point sequence acquisition subunit is used to classify the ship AIS data according to the MMSI code, and then sort them according to the timestamp to obtain the AIS data point sequence sorted according to the timestamp.
[0092] The AIS data point processing subunit is used to determine the ship status according to the Status field of the ship's AIS data, remove the AIS data points where the ship status is not underway, and then segment the discontinuous point sequence to obtain multiple trajectory data.
[0093] The distance calculation subunit is used to calculate the shortest distance between two points on the surface for each trajectory data.
[0094] The present application obtains trajectory data, ship geometry and meteorological data; constructs a VDTTE model, which includes a spatiotemporal module, an attribute module and a multi-task module; the attribute module includes a multi-layer perceptron and a connection layer connected in sequence; the trajectory data is input into the spatiotemporal module, the ship geometry and meteorological data are input into the attribute module, the outputs of the spatiotemporal module and the attribute module are transmitted to the multi-task module, and the ship sailing time estimation result is obtained based on the multi-task module. The VDTTE model proposed in the present application can effectively capture the spatial and temporal dependencies in a given path, and comprehensively considers various factors affecting the ship sailing time to improve the prediction accuracy.
[0095] In an exemplary embodiment, a computer device is provided, such as Figure 4As shown, the computer device can be a server or a terminal. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store ship sailing time estimation data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a ship sailing time estimation method is implemented.
[0096] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the above method is implemented when the processor executes the computer program.
[0097] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and the computer program implements the above method when executed by a processor.
[0098] In an exemplary embodiment, a computer program product is provided, including a computer program, which implements the above method when executed by a processor.
[0099] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ReadOnlyMemory, ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (Magnetoresistive RandomAccess Memory, MRAM), ferroelectric random access memory (Ferroelectric RandomAccess Memory, FRAM), phase change memory (Phase Change Memory, PCM), graphene memory, etc. Volatile memory can include random access memory (RandomAccess Memory, RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).
[0100] In this application, all actions to obtain signals, information or data are carried out in compliance with the relevant data protection laws and policies of the country where they are located and with the authorization given by the owner of the corresponding device.
[0101] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.
[0102] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0103] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.
Claims
1. A method for estimating ship sailing time, characterized in that: The ship sailing time estimation method comprises: Acquiring trajectory data, ship geometry and meteorological data; the trajectory data is obtained based on the ship AIS data; the ship geometry includes the length, width, maximum draft and cargo type of the ship; the meteorological data includes wind speed, wind direction, wave height and visibility; Constructing a VDTTE model; the VDTTE model includes a spatiotemporal module, an attribute module and a multi-task module; Inputting the trajectory data into the spatiotemporal module, and inputting the ship geometry and the meteorological data into the attribute module; The output data of the spatiotemporal module and the attribute module are transmitted to the multi-task module to determine the estimation result of the ship's sailing time.
2. The method for estimating ship sailing time according to claim 1, characterized in that: Obtain trajectory data, vessel geometry and meteorological data, including: Classifying the ship AIS data according to the MMSI code, and then sorting the classified ship AIS data according to the timestamp, to determine the AIS data point sequence sorted according to the timestamp; one ship AIS data is one ship AIS data point; Based on the AIS data point sequence, judging the ship status according to the Status field of each ship AIS data point, and removing the AIS data points whose ship status is not underway; The discontinuous point sequence is segmented to obtain multiple trajectory data. For each trajectory data, the shortest surface distance between two ship AIS data points is calculated, and the ship geometric characteristics and meteorological data corresponding to each trajectory data are recorded.
3. The method for estimating ship sailing time according to claim 1, characterized in that: The spatiotemporal module includes a geographic convolutional layer and a long short-term memory layer connected in sequence.
4. The method for estimating ship sailing time according to claim 3, characterized in that: The attribute module includes a multi-layer perceptron and a first connection layer connected in sequence; The first connection layer is connected to the long short-term memory layer.
5. The method for estimating ship sailing time according to claim 4, characterized in that: The multi-layer perceptron is: h1=σ(W1x v +b1),h2=σ(W2h1+b2),…,h n-1 =σ(W n-1 h n-2 +b n-1 ) e v =W n h n-1 +b n Among them, x v is the category feature, W i is the weight matrix of each i layer, b i is the bias term for each i-th layer, i=1,2,3...n, σ(·) is the activation function, e v is the final embedding representation of category v.
6. The method for estimating the sailing time of a ship according to claim 5, characterized in that: The multi-task module includes an attention mechanism layer, a second connection layer, a residual connection layer and a fully connected layer; The attention mechanism layer is connected to the long short-term memory layer and the first connection layer respectively; The second connection layer is also connected to the first connection layer; The fully connected layer and the residual connection layer jointly output the ship sailing time estimation result.
7. A ship sailing time estimation system, characterized in that: The ship sailing time estimation system comprises: A data acquisition module, used to obtain trajectory data, ship geometry and meteorological data; the trajectory data is obtained based on the ship AIS data; the ship geometry includes the length, width, maximum draft and cargo type of the ship; the meteorological data includes wind speed, wind direction, wave height and visibility; A model building module, used to build a VDTTE model; the VDTTE model includes a spatiotemporal module, an attribute module and a multi-task module; The result output module is used to input the trajectory data into the space-time module, input the ship's geometric characteristics and the meteorological data into the attribute module, transmit the output data of the space-time module and the attribute module to the multi-task module, and determine the ship's sailing time estimation result.
8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the ship sailing time estimation method according to any one of claims 1 to 6.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the ship sailing time estimation method according to any one of claims 1 to 6 is implemented.
10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the ship sailing time estimation method according to any one of claims 1 to 6 is implemented.