Ultra-large oil tanker route intelligent analysis system and method
By designing an intelligent analysis system for routes of super-large oil tankers, using data refining and deep learning algorithms (Transformer models) to process AIS data, the problems of high noise, high complexity and difficult route prediction are solved, and efficient oil tanker information retrieval and accurate route prediction are achieved.
Patent Information
- Application Number
- CN202411707269.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-27
- Publication Date
- 2025-04-25
AI Technical Summary
The prior art has problems such as high noise, high complexity, and difficulty in predicting long-distance routes when analyzing ship behavior using AIS data.
An intelligent analysis system for routes of super-large oil tankers is designed, including ship information query module and route analysis and processing module. The target route is generated through data refining and deep learning algorithms (Transformer model), and the exception handling and route extraction of route data are realized.
It improves the value density of track data, reduces data storage needs, improves the efficiency and accuracy of oil tanker information retrieval, can effectively analyze the behavior of maritime ships and generate accurate route predictions.
Smart Images
Figure CN119884506A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of shipping data processing, and in particular to a system and method for intelligently analyzing routes of very large oil tankers. Background Art
[0002] The data provided by the AIS system includes static data, dynamic data and voyage data. Among them, static data include: ship call sign, ship name, length and width, ship type, location of fixed antenna on board (behind the bow and on the left and right of the center line), AIS identification code, etc. Dynamic data include: ship position with accurate display and complete status, world standard time, course to ground, speed to ground, heading, navigation status, turning rate (if any), etc. Navigation data include: ship status, ship draft, type of cargo loaded on the ship, type of dangerous goods on the ship, the port of destination to be reached by the ship, the estimated time of arrival at the port of destination, and the route plan selected by the ship (voyage reference point). AIS data analysis of maritime ship behavior has become a new development direction in the shipping field, but the AIS data obtained in reality is redundant, complex and contains a lot of noise, which seriously hinders researchers from efficiently retrieving oil tanker information and cannot meet the business needs of complex behavior analysis of oil tankers. In addition, due to the lack of effective long-distance route generation methods, the prediction of complete routes has become particularly difficult. Therefore, there is an urgent need for a data mining method to timely and effectively analyze the behavior of marine vessels. Summary of the invention
[0003] To this end, the present invention provides a very large oil tanker route intelligent analysis system and method to solve the problems of large noise, high complexity, and difficulty in complete route prediction in the existing use of AIS data to analyze ship behavior.
[0004] According to the design scheme provided by the present invention, on the one hand, a very large oil tanker route intelligent analysis system is provided, comprising: a ship information query module and a route analysis and processing module, wherein:
[0005] A ship information query module, used to use a database to store and align ship information data, and to implement a ship information query function based on a key ship information table in the database, wherein the key ship information table includes a static data refinement data table, a historical ship information data table, and a crawler data table, wherein the static data refinement data table is used to store the AIS system static data after data refinement, the historical ship information data table is used to store historical ship information, and the crawler data table is used to align and store the original crawler ship data obtained according to the crawler rules with the historical ship information in the historical ship information data table;
[0006] The route analysis and processing module is used to perform exception processing and route extraction on route data based on the dynamic data of the AIS system and generate a target route using a deep learning algorithm. The deep learning algorithm uses a Transformer model to generate route predictions.
[0007] As the ultra-large oil tanker route intelligent analysis system of the present invention, further, the ship information query module includes: a static information management unit and a static information query unit, wherein:
[0008] A static information management unit, used to receive the imported AIS system static data, obtain the ship information data in the AIS system static data by using a data refinement method, and align and store them in a database, wherein the data refinement method is used to eliminate redundant data in the AIS system static data;
[0009] The static information query unit is used to match the input ship data to be queried with the ship information data stored in the database, and obtain the ship information to be queried based on the matching result.
[0010] As the VLCC route intelligent analysis system of the present invention, further, the route analysis processing module includes: a route processing unit, a route extraction unit, a model training unit and a route generation unit, wherein:
[0011] A route processing unit, used for receiving the imported AIS system dynamic data and preprocessing the dynamic data, wherein the preprocessing at least includes: data deduplication processing, data outlier processing and data filtering processing, wherein the data outlier includes track data attribute outlier and space outlier;
[0012] A route extraction unit, for extracting route data between designated starting points based on a regional geom data table, wherein the geom data table is used to store map elements in a geographic area based on pre-processed dynamic data;
[0013] A model training unit, used to identify route disruption data according to specified latitude and longitude parameters and construct a model training data set, and use the training data set to train the route identification model;
[0014] The route generation unit is used to input the route data to be predicted into the trained route recognition model, and generate the corresponding route using the route recognition model, wherein the route data to be predicted includes route starting point parameters and route point number parameters between route starting points.
[0015] In another aspect, the present invention further provides a method for querying information of a very large oil tanker, which is implemented based on the above system, and the implementation process includes:
[0016] Receive a ship keyword to be queried by the user, where the ship keyword to be queried is any one of the ship's IMO number, MMSI number and ship name;
[0017] The ship keyword to be queried is matched with the historical ship information stored in the historical ship information data table to obtain the ship information most similar to the ship keyword to be queried and recommend it to the user.
[0018] As the ultra-large oil tanker information query method of the present invention, it further includes: extracting the ship category field in the historical ship information data table, and generating a category dictionary data table to match and store the ship category in the ship data set collected by the user with the category dictionary data table.
[0019] As a very large oil tanker information query method of the present invention, further, if the keyword of the ship to be queried is a number, the number is identified as the ship's IMO number or MMSI number based on the number of digits, so as to match the most similar ship information in the historical ship information data table based on the ship's IMO number or MMSI number.
[0020] In another aspect, the present invention further provides a method for extracting a route from a starting point, which is implemented based on the above system, and the implementation process includes:
[0021] Perform data preprocessing on the dynamic data of the AIS system, and visualize each route after data preprocessing on the map, wherein the data preprocessing includes outlier processing and track data filtering;
[0022] According to the starting point and the maximum duration of the track specified by the user, the corresponding routes between the starting points are extracted from the route data after data preprocessing.
[0023] As the starting point route extraction method of the present invention, further, the data preprocessing process includes:
[0024] According to the specified route attributes and the specified longitude and latitude thresholds, the dynamic data is filtered for attribute outliers and spatial outliers respectively;
[0025] And according to the minimum number of track points set in the route and the specified longitude and latitude threshold, abnormal routes in the dynamic data are eliminated, and the abnormal routes include: wandering routes and short routes with a length less than the threshold.
[0026] On the other hand, the present invention also provides a method for predicting a route of a very large oil tanker based on a Transformer model, which is implemented based on the above system, and the implementation process includes:
[0027] The starting point and route points input by the user are used as model input data, and the trained route generation model is used to generate routes with corresponding route points between the starting points;
[0028] Among them, the route generation model is built based on the Transformer model, and a model training data set is constructed according to the longitude and latitude thresholds between the starting points, so as to use the training data set to train the route generation model.
[0029] As a very large oil tanker route prediction method based on the Transformer model of the present invention, further, a model training data set is constructed according to the latitude and longitude thresholds between the starting points, including:
[0030] Set longitude and latitude threshold parameters, and use the longitude and latitude threshold parameters to identify and determine the interrupted portion of the designated route;
[0031] The interrupted part of the designated route is used to construct the corresponding data set for model training.
[0032] Beneficial effects of the present invention:
[0033] According to the actual needs of business units, the present invention utilizes static information processing strategies to retain and expand the original static information, thereby increasing the value density of track data, compressing data storage requirements, and improving the efficiency and accuracy of business personnel in tanker information retrieval; deep learning technology is applied to route generation tasks, combined with trajectory prediction and route generation, to expand the pre-trajectory measurement, and experimentally verifies the effectiveness and feasibility of the scheme, achieving relatively satisfactory results; the route intelligent analysis system is used to modularly design and manage the functions required by business units, combined with a friendly interactive interface to achieve the integration of data visualization and intelligent algorithms, promote human-computer collaborative analysis, help users quickly understand relevant routes and make decisions, and provide convenience for business units. BRIEF DESCRIPTION OF THE DRAWINGS
[0034] Figure 1 This is a schematic diagram of the architecture of the very large oil tanker route intelligent analysis system in the embodiment;
[0035] Figure 2 This is a schematic diagram of the workflow of the intelligent analysis system in the embodiment;
[0036] Figure 3 This is a schematic diagram of the static information management process in the embodiment;
[0037] Figure 4 This is a schematic diagram of the static information query process in the embodiment;
[0038] Figure 5 This is a schematic diagram of the route data preprocessing process in the embodiment;
[0039] Figure 6 This is a schematic diagram of the route extraction process in the embodiment;
[0040] Figure 7 This is a schematic diagram of the route generation process in the embodiment;
[0041] Figure 8 The following is a schematic diagram of the training process of two deep learning models in the embodiment;
[0042] Fig. 9 This is a schematic diagram of the data import UI interface in the embodiment;
[0043] Fig.10 This is a schematic diagram of a data refining operation log file in the embodiment;
[0044] Fig.11 This is a schematic diagram of the dynamic data preprocessing interface in the embodiment;
[0045] Fig.12 This is a schematic diagram of the route generation interface in the embodiment. DETAILED DESCRIPTION
[0046] In order to make the purpose, technical solutions and advantages of the present invention clearer and more understandable, the present invention is further described in detail below in conjunction with the accompanying drawings and technical solutions.
[0047] In view of the problems of large noise and complexity in using AIS data for track data analysis, an embodiment of the present invention provides a very large oil tanker route intelligent analysis system, including: a ship information query module and a route analysis and processing module, wherein:
[0048] A ship information query module, used to use a database to store and align ship information data, and to implement a ship information query function based on a key ship information table in the database, wherein the key ship information table includes a static data refinement data table, a historical ship information data table, and a crawler data table, wherein the static data refinement data table is used to store the AIS system static data after data refinement, the historical ship information data table is used to store historical ship information, and the crawler data table is used to align and store the original crawler ship data obtained according to the crawler rules with the historical ship information in the historical ship information data table;
[0049] The route analysis and processing module is used to perform exception processing and route extraction on route data based on the dynamic data of the AIS system and generate a target route using a deep learning algorithm. The deep learning algorithm uses a Transformer model to generate route predictions.
[0050] Based on the actual application needs, in response to the challenges faced by the stability of current energy routes, in this case implementation, visual analysis is combined with intelligent algorithms to collaborate with humans and machines to improve the ability to detect suspicious oil tankers and the early warning ability of energy routes, providing scientific and reasonable reference support for shipping data mining and overall strategic control of routes, strategic decision-making, etc.
[0051] The ship information query module includes: a static information management unit and a static information query unit, wherein:
[0052] A static information management unit, used to receive the imported AIS system static data, obtain the ship information data in the AIS system static data by using a data refinement method, and align and store them in a database, wherein the data refinement method is used to eliminate redundant data in the AIS system static data;
[0053] The static information query unit is used to match the input ship data to be queried with the ship information data stored in the database, and obtain the ship information to be queried based on the matching result.
[0054] The route analysis and processing module includes: a route processing unit, a route extraction unit, a model training unit and a route generation unit, wherein:
[0055] A route processing unit, used for receiving the imported AIS system dynamic data and preprocessing the dynamic data, wherein the preprocessing at least includes: data deduplication processing, data outlier processing and data filtering processing, wherein the data outlier includes track data attribute outlier and space outlier;
[0056] A route extraction unit, for extracting route data between designated starting points based on a regional geom data table, wherein the geom data table is used to store map elements in a geographic area based on pre-processed dynamic data;
[0057] A model training unit, used to identify route disruption data according to specified latitude and longitude parameters and construct a model training data set, and use the training data set to train the route identification model;
[0058] The route generation unit is used to input the route data to be predicted into the trained route recognition model, and generate the corresponding route using the route recognition model, wherein the route data to be predicted includes route starting point parameters and route point number parameters between route starting points.
[0059] like Figure 1 As shown in the figure, B / S architecture and Web development technology can be used to realize the modular management of static information management and query, dynamic data preprocessing, route extraction and route generation in the route intelligent analysis system. Figure 2 The system business process shown in the figure classifies AIS data, realizes static information query of ships based on static data, obtains the target route situation based on dynamic data and generates corresponding routes.
[0060] Furthermore, based on the above system, an embodiment of the present invention also provides a method for querying information of a very large oil tanker, comprising:
[0061] Receive a ship keyword to be queried by the user, where the ship keyword to be queried is any one of the ship's IMO number, MMSI number and ship name;
[0062] The ship keyword to be queried is matched with the historical ship information stored in the historical ship information data table to obtain the ship information most similar to the ship keyword to be queried and recommend it to the user.
[0063] The ship category fields in the historical ship information data table can be extracted in advance, and a category dictionary data table can be generated to match and align the ship categories in the ship data set collected by the user with the category dictionary data table.
[0064] like Figure 3 and 4 As shown, the ship information data obtained through legal means such as web crawlers can be imported from CSV files into the database and aligned with the contents of the ship information data table in historical records. At the same time, it supports the normal operation of static data refinement and realizes the total query of ship static information based on the key tables of the ship information database.
[0065] If the keyword of the ship to be queried is a number, the number is identified as the ship's IMO number or MMSI number based on the number of digits, so as to match the most similar ship information in the historical ship information data table based on the ship's IMO number or MMSI number.
[0066] Furthermore, based on the above system, an embodiment of the present invention also provides a method for extracting a route from a starting point, comprising:
[0067] Perform data preprocessing on the dynamic data of the AIS system, and visualize each route after data preprocessing on the map, wherein the data preprocessing includes outlier processing and track data filtering;
[0068] According to the starting point and the maximum duration of the track specified by the user, the corresponding routes between the starting points are extracted from the route data after data preprocessing.
[0069] The data preprocessing process includes:
[0070] According to the specified route attributes and the specified longitude and latitude thresholds, the dynamic data is filtered for attribute outliers and spatial outliers respectively;
[0071] And according to the minimum number of track points set in the route and the specified longitude and latitude threshold, abnormal routes in the dynamic data are eliminated, and the abnormal routes include: wandering routes and short routes with a length less than the threshold.
[0072] like Figure 5 and 6As shown in the figure, by preprocessing the AIS dynamic data, simplifying the track data, processing outliers and filtering the tracks, the accuracy and availability of the data before model training are ensured, and the automatic extraction of the route between two points is realized based on the regional geom data table, where geom data is used to describe and draw geographic areas and maps, providing various map elements such as lines, polygons and points, as well as map symbols and labels, etc., and using the geom data table to store the corresponding geographic area map elements.
[0073] Furthermore, based on the above system, an embodiment of the present invention also provides a method for predicting a route of a very large oil tanker based on a Transformer model, comprising:
[0074] The starting point and route points input by the user are used as model input data, and the trained route generation model is used to generate routes with corresponding route points between the starting points;
[0075] Among them, the route generation model is built based on the Transformer model, and a model training data set is constructed according to the longitude and latitude thresholds between the starting points, so as to use the training data set to train the route generation model.
[0076] From the perspective of route prediction technology, route trajectory data usually exhibits characteristics such as massiveness, trend, periodicity, volatility, and stability due to the existence of multiple potential connections between its variables. In existing research, although deep learning technology has been applied to the analysis of trajectory data, it mainly focuses on short-distance trajectory prediction problems, while long-distance route generation tasks are mostly implemented using path planning and clustering algorithms. In the present embodiment, Figure 7 As shown in the figure, deep learning is combined with route generation to realize route generation based on global OD (Origin-Destination). First, the route extraction algorithm based on OD is used, combined with the buffer technology, to accurately extract the route information from the origin to the destination from the large-scale AIS data set; then, the extracted route data is enhanced through resampling and inverse processing to expand the effective coverage of the data set. Finally, as Figure 8 As shown in the figure, two Transformer-based deep learning models can be used for training, recursive prediction, and a complete route sequence can be derived. In the deep learning model framework, the Encoder-LSTM architecture is constructed, combining the Transformer model and the long short-term memory network model; the Transformer model is used to process the long-distance dependency of time series data and extract route features; the LSTM model is used to predict routes, using its time series prediction capabilities.
[0077] Among them, the model training data set is constructed based on the latitude and longitude thresholds between the starting points, including:
[0078] Set longitude and latitude threshold parameters, and use the longitude and latitude threshold parameters to identify and determine the interrupted portion of the designated route;
[0079] The interrupted part of the designated route is used to construct the corresponding data set for model training.
[0080] like Figure 7 As shown, by extracting the track data, the model training data set can be constructed by performing data enhancement operations such as data resampling and route inversion processing on the track data, so as to train the model using the training data set. The target track is generated based on the trained model.
[0081] In order to verify the effectiveness of this solution, the following is a further explanation of the solution in combination with the system platform:
[0082] In static information management, users can activate the UI interface file through the data import button in the platform. In the UI window, users can click "Read Data" and select a local folder to read all CSV format data files in the folder. Subsequently, the program will automatically match the category field in the data with the category number in the dictionary data table, and return the matching results in the list at the bottom of the window. Users have the opportunity to manually review and modify the automatic matching results to ensure the accuracy of the categories. The modified corresponding results can be saved through the "Save Results" button. The saved files can be viewed by clicking the "Existing Files" button in the upper left corner of the window, such as Fig. 9 Finally, after the user confirms that the corresponding result is correct, click "Confirm Input", the program will replace the category field in the original data with the corresponding category number and import the data into the database. In the back-end system, the refining of static data can be deployed on a regular basis by using the scheduled task function of the Django framework. Fig.10 As shown in the figure, the system uses Django's log function to externally mount the program's running status, making it easier for users to monitor the execution of the program. This improves the efficiency and transparency of data processing.
[0083] In static query, users can query ship information by searching for any one of the IMO number, MMSI number or ship name in the input box. The system first recommends to users the complete ship information that is most similar to the input through an automatic matching mechanism, including the IMO number, MMSI number and ship name. Users can choose to click on the recommended results to obtain specific ship information for query according to their needs, or directly use the entered number for query. If the query is directly based on the entered number, the system will automatically identify the 7-digit number as the IMO number and the 9-digit number as the MMSI number. Furthermore, click the "Query" button, and the system will conduct a comprehensive query in the three key data tables of the ship information database based on the user input. The query returns the life cycle information and physical property information about the MMSI number and IMO number, and highlights the most matching query results in red.
[0084] During route generation, in the route analysis interface, open the "Dynamic Data Preprocessing" window. The user first clicks the "Click to Upload" button and selects the CSV data file of the original route to be analyzed for upload. After successful upload, the original data points will be displayed on the base map. At the same time, the "Outlier Processing" and "Track Data Filtering" function buttons at the bottom of the window are activated. Fig.11 As shown. Next, users can select the specific attributes they want to process for attribute outliers, set longitude and latitude thresholds to determine route interruptions, and click "Process Outliers". The system will filter attribute outliers and spatial outliers based on the parameters. In addition, users can also set the minimum number of track points and longitude and latitude thresholds for each route to identify and eliminate wandering routes or short routes. Users can adjust these parameters multiple times to obtain the best results.
[0085] Open the "OD-based route extraction" window, and the user enters the names of the two locations into the input boxes on both sides of the double-headed arrow. During the input process, the system will automatically display a drop-down box to provide matching options for the location name for the user to select. After completing the location input, the user needs to set the maximum duration of each track. After the settings are completed, click the "Extract route" button, and the system will extract the route between the two points from the pre-processed route data. The user can adjust the parameters multiple times to obtain the best result.
[0086] Open the "Transformer-LSTM Model Training" window and set the latitude and longitude threshold parameters in the window. These parameters will be used to identify and judge the interrupted parts of the route to build a suitable data set. After the settings are completed, click the "Train Model" button to start training. The progress bar at the bottom of the window will show the progress percentage of the current training cycle (epoch). After the training is completed, the system will pop up a message box (messagebox) to inform the user that the training is complete. Then, the user enters the save file name of the model in the specified input box and clicks the "Save Model" button. The model will be automatically saved to the folder specified by the backend.
[0087] Open the "Route Generation" window, the user first clicks the "Click to Upload" button to upload the short route csv file to be predicted, and the base map will automatically visualize the short routes in the file. Next, the user needs to enter the estimated number of route points and select the required model file from the list at the bottom of the window. This list monitors the model folder in the backend in real time and presents all the model files in it. Fig.12 As shown in the figure, after clicking the "Generate Route" button, the system will display the generated route on the base map, where the yellow part is the known part and the red part is the generated part. Users can try different route generation results by constantly changing the generated route point parameters and model files. Finally, users can click the "Export Results" button to export the generated route data.
[0088] The processing process and results of the track data through the above system platform interface can further illustrate that the solution in this case can realize the management and query of AIS static information, the processing of route dynamic data, and the extraction and generation of routes, which can improve the implementation and reliability of route prediction, improve the retrieval efficiency and mining depth of ship information such as oil tankers, and has good application prospects in the field of route control.
[0089] Unless otherwise specifically stated, the relative steps, numerical expressions and values of the components and steps set forth in these embodiments do not limit the scope of the present invention.
[0090] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.
[0091] The units and method steps of each example described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. A person of ordinary skill in the art may use different methods to implement the described functions for each specific application, but such implementation is not considered to be beyond the scope of the present invention.
[0092] Those skilled in the art will appreciate that all or part of the steps in the above method can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, such as a read-only memory, a disk or an optical disk. Optionally, all or part of the steps in the above embodiment can also be implemented using one or more integrated circuits, and accordingly, each module / unit in the above embodiment can be implemented in the form of hardware or in the form of software function modules. The present invention is not limited to any specific form of combination of hardware and software.
[0093] Finally, it should be noted that the above-described embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above-described embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-described embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims
1. An intelligent analysis system for very large oil tanker routes, characterized in that: Contains: ship information query module and route analysis and processing module, among which, A ship information query module, used to use a database to store and align ship information data, and to implement a ship information query function based on a key ship information table in the database, wherein the key ship information table includes a static data refinement data table, a historical ship information data table, and a crawler data table, wherein the static data refinement data table is used to store the AIS system static data after data refinement, the historical ship information data table is used to store historical ship information, and the crawler data table is used to align and store the original crawler ship data obtained according to the crawler rules with the historical ship information in the historical ship information data table; The route analysis and processing module is used to perform exception processing and route extraction on route data based on the dynamic data of the AIS system and generate a target route using a deep learning algorithm. The deep learning algorithm uses a Transformer model to generate route predictions.
2. The VLCC route intelligent analysis system according to claim 1 is characterized in that: The ship information query module includes: a static information management unit and a static information query unit, wherein: A static information management unit, used to receive the imported AIS system static data, obtain the ship information data in the AIS system static data by using a data refinement method, and align and store them in a database, wherein the data refinement method is used to eliminate redundant data in the AIS system static data; The static information query unit is used to match the input ship data to be queried with the ship information data stored in the database, and obtain the ship information to be queried based on the matching result.
3. The intelligent analysis system for VLCC routes according to claim 1 is characterized in that: The route analysis and processing module includes: a route processing unit, a route extraction unit, a model training unit and a route generation unit, wherein: A route processing unit, used for receiving the imported AIS system dynamic data and preprocessing the dynamic data, wherein the preprocessing at least includes: data deduplication processing, data outlier processing and data filtering processing, wherein the data outlier includes track data attribute outlier and space outlier; A route extraction unit, for extracting route data between designated starting points based on a regional geom data table, wherein the geom data table is used to store map elements in a geographic area based on pre-processed dynamic data; A model training unit, used to identify route disruption data according to specified latitude and longitude parameters and construct a model training data set, and use the training data set to train the route identification model; The route generation unit is used to input the route data to be predicted into the trained route recognition model, and generate the corresponding route using the route recognition model, wherein the route data to be predicted includes route starting point parameters and route point number parameters between route starting points.
4. A method for querying information of a very large oil tanker, characterized in that: Based on the system implementation of claim 1, the implementation process includes: Receive a ship keyword to be queried by the user, where the ship keyword to be queried is any one of the ship's IMO number, MMSI number and ship name; The ship keyword to be queried is matched with the historical ship information stored in the historical ship information data table to obtain the ship information most similar to the ship keyword to be queried and recommend it to the user.
5. The method for querying information of a very large oil tanker according to claim 4, characterized in that: It also includes: extracting the ship category field in the historical ship information data table and generating a category dictionary data table to match and store the ship categories in the ship data set collected by the user with the category dictionary data table.
6. The method for querying information of a very large oil tanker according to claim 4, characterized in that: If the keyword of the ship to be queried is a number, the number is identified as the ship's IMO number or MMSI number based on the number of digits, so as to match the most similar ship information in the historical ship information data table based on the ship's IMO number or MMSI number.
7. A method for extracting a route from a starting point, characterized in that: Based on the system implementation of claim 1, the implementation process includes: Perform data preprocessing on the dynamic data of the AIS system, and visualize each route after data preprocessing on the map, wherein the data preprocessing includes outlier processing and track data filtering; According to the starting point and the maximum duration of the track specified by the user, the corresponding routes between the starting points are extracted from the route data after data preprocessing.
8. The method for extracting the starting point route according to claim 7, characterized in that: The data preprocessing process includes: According to the specified route attributes and the specified longitude and latitude thresholds, the dynamic data is filtered for attribute outliers and spatial outliers respectively; And according to the minimum number of track points set in the route and the specified longitude and latitude threshold, abnormal routes in the dynamic data are eliminated, and the abnormal routes include: wandering routes and short routes with a length less than the threshold.
9. A method for predicting the route of a very large oil tanker based on a Transformer model, characterized in that: Based on the system implementation of claim 1, the implementation process includes: The starting point and route points input by the user are used as model input data, and the trained route generation model is used to generate routes with corresponding route points between the starting points; Among them, the route generation model is built based on the Transformer model, and a model training data set is constructed according to the longitude and latitude thresholds between the starting points, so as to use the training data set to train the route generation model.
10. The method for predicting the route of a very large oil tanker based on a Transformer model according to claim 9, characterized in that: The model training dataset is constructed based on the latitude and longitude thresholds between the starting points, including: Set longitude and latitude threshold parameters, and use the longitude and latitude threshold parameters to identify and determine the interrupted portion of the designated route; The interrupted part of the designated route is used to construct the corresponding data set for model training.