A method for analyzing maritime traffic flow based on big data

Abstract

This invention discloses a big data-based maritime traffic flow analysis method, relating to the field of information technology. The method includes: collecting maritime traffic-related data and preprocessing it; using the processed data to adaptively and dynamically divide the sea area, generating real-time updated water boundaries; analyzing ship behavior patterns based on the adaptive division results to obtain anomaly detection results; analyzing the interaction between ship groups based on the anomaly detection results to predict potential congestion and conflicts; optimizing ship path planning based on the group behavior prediction results to generate optimal route suggestions; accurately predicting potential navigation conflicts by constructing a spatiotemporal graph convolutional network model that comprehensively considers the relative position, speed, and meeting frequency of ships; and providing ships with optimal route suggestions based on multi-objective optimization through optimized path planning, thereby maximizing navigation efficiency and economy while ensuring safety.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a method for analyzing marine traffic flow based on big data. Background Art

[0002] Traditional methods of marine traffic management mainly rely on experience and a small amount of real-time data, which makes it difficult to cope with complex navigation environments and frequent dynamic changes of ships. With the advancement of information technology and data science, big data analysis has gradually become an important tool for marine traffic flow management. By collecting and processing massive amounts of marine traffic data, researchers can use advanced data analysis and machine learning techniques to extract valuable information, thereby achieving more accurate monitoring and management of marine traffic flows. For example, using satellite communications and AIS (Automatic Identification System) data, the movement status of ships can be tracked in real time, greatly improving the transparency and predictability of marine traffic flows.

[0003] However, existing technologies still have several shortcomings in maritime traffic flow analysis. First, many existing methods fail to fully consider the diversity and complexity of data in the data preprocessing stage, which affects the accuracy of the analysis results. Secondly, although some technologies can achieve dynamic division of sea areas, they often lack adaptive capabilities and cannot respond to environmental changes and dynamic adjustments of ship behavior patterns in real time. In addition, most existing anomaly detection methods rely on static models, which are difficult to adapt to the complex marine environment and ship interactions, resulting in the inability to effectively predict potential congestion and collision risks. These shortcomings make traditional marine traffic management methods unable to meet the requirements of modern marine transportation, and there is an urgent need to develop new methods to improve the management efficiency and safety of maritime traffic flows. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a maritime traffic flow analysis method based on big data to solve the problems of insufficient data processing, inaccurate analysis of ship behavior patterns and poor path planning optimization effect in maritime traffic flow management.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, an embodiment of the present invention provides a method for analyzing maritime traffic flow based on big data, which includes collecting data related to maritime traffic and performing preprocessing;

[0008] Use the processed data to adaptively and dynamically divide the sea area and generate water boundaries that are updated in real time;

[0009] Based on the adaptive segmentation results, the ship behavior patterns are analyzed to obtain anomaly detection results;

[0010] Analyze the interactions between groups of ships based on anomaly detection results to predict potential congestion and conflicts;

[0011] Based on the group behavior prediction results, ship path planning is optimized and optimal route recommendations are generated.

[0012] As a preferred solution of the marine traffic flow analysis method based on big data of the present invention, the marine traffic-related data includes AIS data, meteorological data and marine environment data; the AIS data refers to providing real-time ship position information, speed and heading by receiving signals transmitted by ships; the meteorological data refers to wind speed, wind direction and visibility obtained by meteorological stations and satellites; the marine environment data refers to tides and wave heights monitored by buoys and satellites;

[0013] The preprocessing includes data format standardization, deduplication, error record cleaning and missing value processing.

[0014] As a preferred solution of the marine traffic flow analysis method based on big data described in the present invention, wherein: using the pre-processed data to adaptively and dynamically divide the sea area to generate a real-time updated water area boundary includes the following steps:

[0015] Extract the position information, speed and heading of each ship from the preprocessed data;

[0016] According to the extracted ship position information, determine the grid area where each ship is located;

[0017] Using the ship’s speed and heading data, predict the ship’s position at a future point in time;

[0018] Map the predicted ship positions to grid areas and count the actual and predicted number of ships in each area;

[0019] Calculate current and predicted ship density based on statistical results;

[0020] Apply K-means clustering algorithm to perform cluster analysis on current density and predicted density to identify the ship flow and congestion in each area;

[0021] Based on the results of K-means cluster analysis, combined with real-time ship density, flow direction, speed, and meteorological and ocean environmental data, a real-time updated water boundary is formed.

[0022] As a preferred solution of the marine traffic flow analysis method based on big data described in the present invention, wherein: based on the adaptive division result, analyzing the ship behavior pattern and obtaining the abnormal detection result includes the following steps:

[0023] Extracting real-time behavior data of ships from the adaptive segmentation results;

[0024] Define the characteristics of normal behavior models based on historical data and domain knowledge;

[0025] Use machine learning algorithms to model normal behavior;

[0026] Use historical normal behavior data to train models and identify normal speed, course changes, and ship spacing;

[0027] Compare the real-time collected ship behavior data with the normal behavior model, and calculate the deviation of each ship in the real-time data and the normal behavior model identification results;

[0028] Set a deviation threshold. When the deviation exceeds the set threshold, the behavior is automatically marked as abnormal.

[0029] Based on the deviation results, abnormal behavior reports are automatically generated.

[0030] As a preferred solution of the marine traffic flow analysis method based on big data of the present invention, wherein: analyzing the interaction between ship groups and predicting potential congestion and conflict according to the abnormal detection results includes the following steps:

[0031] Collect historical data and historical voyage records related to abnormal vessel behavior;

[0032] Select key features based on the ship's behavior patterns and environmental conditions, and perform feature enhancement;

[0033] Apply natural language processing technology to identify ship behavior patterns and related keywords in specific situations in historical voyage record texts, and convert related keywords into numerical features;

[0034] The features extracted from the text analysis are combined with the key features to obtain the characteristics of the ship and form a feature vector;

[0035] Consider each ship as a node in the graph;

[0036] The relative positions and distances between ships are used as edges;

[0037] Define a distance threshold, and the ships with distances less than the distance threshold are considered as interconnected edges;

[0038] Calculate the base weight of the edge based on the ship's speed and heading;

[0039] Count the frequency of each pair of ships passing each other in the past period of time, analyze historical data, and calculate the degree of interference between ships in a congested state;

[0040] Combine the intersection frequency and congestion interference degree to calculate the comprehensive impact;

[0041] Combine the comprehensive influence with the basic edge weight to update the final weight of each edge;

[0042] Connect all ship nodes with their corresponding dynamically updated edges to form a comprehensive spatiotemporal graph convolutional network model;

[0043] Use the ship’s feature vectors for model training;

[0044] The trained spatiotemporal graph convolutional network model is used to predict real-time data and obtain potential congestion and conflict situations.

[0045] As a preferred solution of the marine traffic flow analysis method based on big data described in the present invention, wherein: based on the group behavior prediction results, optimizing the ship path planning and generating the optimal route suggestion include the following steps:

[0046] Treat each ship as an independent player in the game theory and define clear goals for each ship;

[0047] Construct a payoff matrix to quantify the gains and losses of each ship under different strategies and define the optional strategies of the ships;

[0048] Use an iterative algorithm to calculate the Nash equilibrium so that in the equilibrium state, no ship can obtain better benefits by unilaterally changing its strategy;

[0049] Based on the Nash equilibrium results, the path optimization algorithm is applied to calculate the optimal route selection;

[0050] Use multi-objective optimization algorithms to combine multiple well-defined objectives, perform multi-objective optimization, and generate optimal route recommendations.

[0051] In a second aspect, an embodiment of the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the maritime traffic flow analysis method based on big data as described in the first aspect of the present invention is implemented.

[0052] In a third aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the maritime traffic flow analysis method based on big data as described in the first aspect of the present invention is implemented.

[0053] The beneficial effects of the present invention are as follows: by collecting and preprocessing maritime traffic-related data, the accuracy and consistency of the data are ensured, and erroneous analysis caused by data noise is reduced; the preprocessed data is used for adaptive dynamic partitioning, combined with the K-means clustering algorithm, so that changes in ship density and flow can be reflected in real time; ship behavior patterns are analyzed based on adaptive partitioning, and abnormality detection is performed using normal behavior models, so that abnormal behaviors can be quickly identified and reports can be generated; by analyzing the interaction between ship groups, combining historical data and natural language processing technology, a spatiotemporal graph convolutional network model is constructed to predict potential navigation conflicts; by optimizing path planning, optimal route recommendations based on multi-objective optimization are generated, which further improves navigation efficiency and economy. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0055] Figure 1 This is a flow chart of the maritime traffic flow analysis method based on big data in Example 1.

[0056] Figure 2 This is a flow chart for generating a real-time updated water area boundary in Example 1. DETAILED DESCRIPTION

[0057] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0058] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0059] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0060] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, and provides a method for analyzing marine traffic flow based on big data, comprising the following steps:

[0061] S1. Collect and pre-process data related to maritime traffic.

[0062] S1.1. Data related to maritime traffic include AIS (Automatic Identification System) data, meteorological data and marine environment data.

[0063] S1.1.1. AIS data refers to the real-time information on ship position, speed and heading provided by receiving signals transmitted by ships.

[0064] Specifically, the ship's location information includes longitude and latitude, which are used to determine the ship's current position and support subsequent trajectory analysis and collision avoidance; speed indicates the speed of the ship's movement in the water, usually in knots, which is crucial for evaluating the ship's driving status; heading indicates the direction of the ship's travel, usually expressed in degrees, and is used to analyze the selection and adjustment of the route.

[0065] S1.1.2 Meteorological data refers to wind speed, wind direction and visibility obtained from weather stations and satellites.

[0066] Specifically, real-time wind speed data is used to help judge sea conditions, which affect the ship's maneuvering and navigation safety; wind direction information helps ships adjust their navigation strategies to reduce inefficiencies in navigation due to headwinds; visibility data provides a safety assessment of navigation under different weather conditions, and low visibility may increase the risk of collision.

[0067] S1.1.3 Marine environmental data refers to tide and wave heights monitored by buoys and satellites.

[0068] Specifically, changes in tide levels directly affect the draft of ships and the feasibility of entering and leaving ports. Tidal data can help ships arrange berthing and navigation reasonably. Wave height information can be used to assess the complexity of ocean conditions. Excessively high waves may affect the stability and navigation safety of ships.

[0069] S1.2. Preprocessing includes data format standardization, deduplication, error record cleaning and missing value processing.

[0070] Specifically, data format standardization includes timestamp standardization and unit standardization.

[0071] Furthermore, timestamp standardization means that the system automatically identifies the format of timestamps in various data sources. By setting unified conversion rules, the system converts all timestamps into Coordinated Universal Time to ensure that all timestamps are compared in the same reference frame; unit standardization means that the system uniformly converts the ship's speed into knots or kilometers per hour (km / h), depending on the analysis needs. During the conversion process, the system will automatically identify the speed unit in the data and convert it according to the set rules. For example, if a data source uses kilometers per hour, the system will convert it into knots, and 1 knot is equal to approximately 1.852 kilometers per hour; other fields such as the representation of location information (such as the degree format of longitude and latitude) will also be unified to ensure that all data uses the same format.

[0072] Specifically, the deduplication process refers to traversing all data sources and checking each record one by one. During the inspection process, the system will compare the ship's unique identifier (such as ship ID) and timestamp. If it is found that there are multiple records for the same ship at the same time point, the system will further analyze these records; sort them according to the timestamp of the records, retain the most recent record in time, and ensure that the retained data reflects the latest status of the ship; records judged to be redundant will be marked as invalid and deleted from the data set after the cleaning is completed, ensuring that there is no duplicate information in the data set, thereby improving the availability of the data.

[0073] Specifically, error record cleaning refers to verifying the key fields in each record, identifying invalid speed values ​​and heading data, and marking them; for detected error records, the system will try to correct them according to preset rules. For example, if the ship speed is a negative value, the system can correct it to 0, indicating that the ship is stationary. If the heading is out of a reasonable range, the system will adjust it to a reasonable value.

[0074] Specifically, missing value processing refers to checking the missing values ​​in the records during the cleaning process and taking appropriate interpolation methods to fill them in, ensuring that each record contains complete information and improving the overall quality of the data.

[0075] S2. Use the processed data to adaptively and dynamically divide the sea area and generate real-time updated water boundaries.

[0076] S2.1. Extract the position information, speed and heading of each ship from the preprocessed data.

[0077] It should be noted that accurate position information, speed and heading are the basis for analyzing ship behavior, predicting future positions and evaluating navigation safety. Inaccurate data may lead to wrong decisions and increase navigation risks. By extracting these key parameters, the system can monitor ship dynamics in real time, promptly identify abnormal behavior or potential collision risks, and take corresponding preventive measures. With accurate ship data, routes can be optimized and dynamically adjusted more effectively, improving navigation efficiency and reducing unnecessary time and fuel consumption.

[0078] S2.2. Determine the grid area where each ship is located based on the extracted ship position information.

[0079] Furthermore, the specific steps for determining the grid area where each ship is located are as follows:

[0080] According to the geographical characteristics and navigation needs of the sea area, the sea area is divided into multiple grid areas. Each grid area should correspond to a specific coordinate range, usually using a rectangular or hexagonal grid structure; in areas with large ship traffic, the grid should be relatively small to more accurately monitor and analyze ship dynamics; consider the terrain, waterways, port locations, etc. of the waters to ensure that the grid division conforms to the actual navigation environment; use real-time ship traffic data to dynamically adjust the size and shape of the grid. For example, in areas with dense ships, the grid can be adjusted to a smaller size to improve the accuracy of data collection; combine meteorological data (such as wind speed, tidal current) and marine environmental factors to dynamically adjust the grid. For example, in areas with poor weather conditions, smaller grids may be required to facilitate accurate monitoring and respond to changes.

[0081] Furthermore, when determining the grid area where a ship is located, it is necessary not only to base it on the ship flow, but also to combine the analysis of traffic flow and navigation environment:

[0082] Traffic flow analysis: AIS data can be used to analyze ship traffic, navigation routes, and areas with high ship density. High-flow waterways should be divided into smaller grids to accurately monitor the dynamics of ships; low-flow areas can be appropriately enlarged. But more importantly, traffic flow analysis can also identify congestion points, especially port entrances and waterway intersections with large ship traffic, which may become key planning areas for water-related projects (such as port expansion or channel dredging).

[0083] Navigation environment analysis: The system can dynamically adjust the grid division by combining chart data, tidal changes, channel depth and other information. For example, in shallow water areas or areas severely affected by tides, it is recommended to use a finer grid division to ensure that ships can pass safely. For the site selection recommendations for water-related projects, shallow water areas are not suitable for the construction of large ports or channel expansion, while deep water areas can be used as priority sites.

[0084] It should be noted that by dividing the sea area into multiple grid areas, the ship flow and dynamics in a specific area can be monitored and analyzed more clearly. The detailed division helps to identify densely populated navigation areas and potential congestion risks. By real-time monitoring of the grid areas where the ships are located, managers can adjust navigation strategies in a timely manner, optimize route selection, and ensure navigation safety.

[0085] S2.3. Use the ship's speed and heading data to predict the ship's position at a future time point.

[0086] Furthermore, assuming that the current position of the ship is known ( , ),speed (Unit: km / h) and heading angle (Unit: degrees), then the ship will be (Unit: hours) position ( , )for,

[0087] ;

[0088] ;

[0089] It should be noted that by predicting future positions, the management system can adjust routes in a timely manner, optimize ship flow, and avoid potential congestion and collision risks; by predicting the future positions of ships, port resources and waterway rights can be allocated more effectively, thereby improving overall shipping efficiency.

[0090] S2.4. Map the predicted ship positions to grid areas and count the actual and predicted number of ships in each area.

[0091] Specifically, the mapping process is as follows:

[0092] Determine the boundary coordinates of each grid area and clarify the geographical scope of each grid; for each predicted ship position, compare its longitude and latitude coordinates with the boundary coordinates of each grid area to determine which specific grid area the position belongs to. If the longitude of the ship is between the minimum and maximum longitude of the grid, and the latitude is also between the minimum and maximum latitude of the grid, then the ship position belongs to the grid; after completing the position matching, it is necessary to count the number of ships in each grid area, including the number of ships actually monitored and the number of ships calculated based on the predicted position; using GIS technology, the mapped ship flow data is integrated into a unified platform, including the actual position, predicted position, speed and heading of the ship, to form a comprehensive spatial data set; by counting the number of ships in different grid areas, a flow distribution map is generated to show which areas have higher ship flow and which areas have lower ship flow; through spatial analysis, GIS can identify hot spots of ship flow. These hot spots may be ports, waterway intersections or other important waters where ships frequently enter and exit.

[0093] Furthermore, suppose the boundary coordinates of each grid are defined as:

[0094] Grid boundary minimum longitude , Maximum longitude of grid boundary , minimum latitude of grid boundary , Maximum latitude of grid boundary ; For each predicted ship position ( , ), determine whether it belongs to a specific grid area, and count the number of ships that meet the conditions , the formula is,

[0095] ;

[0096] in, Represents the predicted position list, 1 is an indicator function, returns 1 if the condition is met, otherwise returns 0;

[0097] For each grid area, count the number of ships actually monitored and the number of ships calculated based on the predicted positions , get the total number of ships , the expression is,

[0098] ;

[0099] The specific examples are as follows:

[0100] Assume the following list of predicted locations :

[0101] ;

[0102] And the boundaries of the grid area are, =50.0, =60.0, =140.0, =150.0;

[0103] Check each predicted position one by one to see if it is within the grid area:

[0104] (58.3,148.3) → in the grid, returns 1;

[0105] (55.0,145.0) → in the grid, returns 1;

[0106] (61.0,149.0) → not in the grid, returns 0;

[0107] (57.5,147.0) → in the grid, returns 1;

[0108] Then the number of ships that meet the conditions is: ;

[0109] Assume that the number of ships actually monitored in a grid area =5, then the total number of ships is: .

[0110] S2.5. Calculate the current and predicted ship density based on the statistical results.

[0111] The specific steps for calculating ship density are as follows:

[0112] Determine the area of ​​each grid area. The area of ​​the grid can be calculated using the known latitude and longitude boundaries. Based on the previous mapping results, obtain the actual and predicted number of ships in each grid area. Calculate the actual and predicted ship density of each grid area by the ratio of the number of ships to the area of ​​the area.

[0113] Furthermore, the calculation formula for determining the area of ​​each grid region is,

[0114] ;

[0115] in, represents the area of ​​each grid region, is a constant used to convert longitude and latitude to actual area (usually in km²), which varies depending on the geographic location, usually in mid-latitudes. ≈111 (one degree is approximately equivalent to 111km).

[0116] The actual and predicted ship density for each grid area is calculated as,

[0117] ;

[0118] ;

[0119] in, represents the actual ship density, represents the predicted ship density.

[0120] It should be noted that the calculation of ship density can help identify which areas may be congested, thereby providing important decision-making support for managers. High-density areas usually mean higher risks of collisions and navigation delays. By calculating ship density in real time, the management system can adjust the channel and flow control measures according to the current navigation status to avoid accidents and improve traffic efficiency. Analysis of historical density data can provide a basis for future port expansion, channel planning and resource allocation, ensuring that it can respond to changing shipping needs.

[0121] S2.6. Apply the K-means clustering algorithm to perform cluster analysis on the current density and predicted density to identify the ship flow and congestion in each area; based on the results of the K-means clustering analysis, combined with real-time ship density, flow direction, speed, and meteorological and ocean environmental data, form a real-time updated water boundary.

[0122] Specifically, the K-means clustering algorithm is used for ship density analysis, including the following steps:

[0123] The optimal K value is determined by the elbow rule and other methods, specifically,

[0124] For different value (for example, from 1 to 10), perform K-means clustering, and calculate each The corresponding total square error (SSE) is expressed as:

[0125] ;

[0126] in, Indicates the number of clusters The total square error is an indicator for evaluating the clustering effect. The smaller it is, the better the clustering effect is. Represents the total number of data points in the data set, which determines how many data points need to be calculated for clustering errors. Represents the data point index. When calculating the total squared error (SSE) and data point distribution, Used to traverse each data point to determine its distance from the cluster center or its position in the cluster. Represents the current cluster index. When looking for the minimum distance and updating the cluster center, Used to traverse each cluster center to determine the cluster to which the data point is closest and calculate the new cluster center. Indicates the current number of clusters, that is, the number of cluster centers selected in K-means clustering. Indicates The feature vector of a data point is usually a vector containing density information. Indicates The characteristic vector of cluster centers represents the cluster center of gravity.

[0127] The density data is input into the K-means algorithm for cluster analysis, specifically,

[0128] Random Selection Initial cluster centers , assign each data point to the nearest cluster center, the expression is,

[0129] ;

[0130] in, Indicates The cluster C to which the data points are assigned is a class label indicating Which cluster does it belong to? It means to find the expression that minimizes the following expression. value, Represents data points and cluster centers The Euclidean distance between them is minimized to determine which cluster the data point belongs to.

[0131] Update the cluster center according to the allocation results. The expression is:

[0132] ;

[0133] in, Indicates that it belongs to The set of all data points in a cluster.

[0134] The assignment and update steps are repeated until the cluster centers no longer change or the preset number of iterations is reached.

[0135] Based on the clustering results, the ship flow and congestion in each area can be identified. High-density areas and low-density areas correspond to different clusters. The specific steps are as follows:

[0136] High-density areas: Classes with higher density in the cluster (for example, the density value in the cluster center is higher than a certain threshold) indicate heavy vessel traffic and possible congestion.

[0137] Low-density areas: Less dense classes in the cluster indicate less vessel traffic and are likely to be more unobstructed.

[0138] Furthermore, the specific steps for generating the water area boundary are as follows:

[0139] Using the results of K-means cluster analysis, the ship flow characteristics of different grid areas are identified, and high-flow areas are marked as areas that need special attention, while management measures can be moderately relaxed in low-flow areas; real-time ship density data is obtained to analyze the number and distribution of ships in each area; the flow direction and speed of ships are monitored in real time to understand the movement trend of ships in the waters; meteorological data such as wind speed, rainfall, and visibility are collected, which may affect the navigation safety of ships; the impact of environmental factors such as tides and currents on the behavior and flow distribution of ships is monitored; based on real-time data, decision algorithms are used to analyze the current ship flow status and environmental conditions, and identify the water boundaries that need to be adjusted; in high-density areas, temporary channels can be set up in the cabin to guide ships to pass more safely; in low-density areas, the channels can be expanded to provide more space to form updated water boundaries.

[0140] Furthermore, when using the K-means clustering algorithm for analysis, adding considerations of traffic flow and navigation environment can better explain the ship flow and congestion, and provide a planning basis for water-related projects:

[0141] Traffic flow hotspot identification: Through K-means clustering results, traffic flow hotspots can be identified, such as channel intersections, port entrances with dense ships, etc. In these hotspots, the ship density is large and the traffic flow is complex, and more sophisticated water management and dynamic water boundary adjustments may be required. This also provides a basis for port expansion or channel widening. It is recommended to plan near these hotspots to alleviate congestion.

[0142] Adaptability to the navigation environment: Based on the combination of meteorological and marine environmental data, the system can dynamically adjust the boundaries of water areas. For example, in areas with high wind speeds or strong tidal currents, the system can automatically adjust the water division to prevent ships from entering high-risk areas. For the site selection of water-related projects, it is recommended to avoid these areas with strong winds and waves and choose relatively stable waters for construction.

[0143] S3. Based on the adaptive segmentation results, the ship behavior pattern is analyzed to obtain the anomaly detection results.

[0144] Extract the real-time behavior data of ships from the adaptive partitioning results; define the characteristics of the normal behavior model based on historical data and domain knowledge; use machine learning algorithms to establish a normal behavior model; use historical normal behavior data to train the model and identify normal speed, course changes and ship spacing; compare the real-time collected ship behavior data with the normal behavior model, and calculate the deviation of each ship in the real-time data and the normal behavior model identification results; set a deviation threshold, and when the deviation exceeds the set threshold, automatically mark the behavior as abnormal; automatically generate an abnormal behavior report based on the deviation results.

[0145] Furthermore, the calculation formula for the deviation between real-time data and the normal behavior model is,

[0146] ;

[0147] in, Indicates Deviation of the vessel, Indicates Real-time characteristics of the ships (such as speed, heading, spacing), Indicates Normal behavior characteristics of a ship.

[0148] Specifically, the abnormality report includes the ID and location information of the abnormal ship, the type of abnormality (such as abnormal speed, abnormal course or insufficient safety distance), the time and duration of the abnormality, and the recommended response measures (such as changing course or speed).

[0149] Furthermore, when analyzing ship behavior patterns, combining traffic flow analysis and navigation environment can accurately determine abnormal ship behavior:

[0150] Traffic flow anomalies: In high-traffic areas, sudden anomalies in the speed and course of ships (such as sharp deceleration or deviation) may indicate potential congestion or equipment failure. By analyzing traffic flow data, the system can identify these abnormal behaviors and issue early warnings. For water-related projects, these abnormal behavior data can be used to evaluate future waterway planning and avoid construction in high-risk areas.

[0151] Abnormal navigation environment: In certain navigation environments (such as areas with strong winds or areas with drastic tidal changes), the behavior of ships may be abnormal. For example, strong winds may cause ships to deviate from their routes. The system can analyze these abnormal behaviors based on environmental data to see if they are caused by environmental factors. These analyses provide valuable environmental data for water-related projects, helping decision makers choose relatively stable areas for construction.

[0152] It should be noted that defining the characteristics of normal behavior helps to establish a benchmark for subsequent comparison and analysis; establishing a normal behavior model through machine learning algorithms enables it to automatically identify normal and abnormal behaviors; by comparing real-time with the normal model, deviations from normal behavior can be identified in a timely manner, and calculating deviations helps to quantify the degree of abnormality in ship behavior; by setting deviation thresholds, automated abnormal behavior marking can be achieved, improving response speed, detecting abnormal behavior in a timely manner, and ensuring navigation safety.

[0153] S4. Analyze the interactions between ship groups based on anomaly detection results and predict potential congestion and conflicts.

[0154] S4.1. Collect historical data and historical navigation records related to abnormal ship behavior; select key features based on the ship's behavior patterns and environmental conditions, and perform feature enhancement.

[0155] Furthermore, feature enhancement includes time series features and aggregation features. Time series features refer to recording the speed and heading data of the past N time points to form a time series feature set. Using sliding window technology, window extraction is performed on each time point to generate multiple time series input features; aggregation features refer to grouping ships and aggregating them according to geographical locations or divided grid areas to form new input features.

[0156] S4.2. Apply natural language processing technology to identify ship behavior patterns and related keywords in specific situations in historical voyage record texts, and convert related keywords into numerical features;

[0157] The features extracted from text analysis are combined with key features to obtain the features of the ship and form a feature vector; each ship is regarded as a node in the graph; the relative position and distance between ships are used as edges; a distance threshold is defined, and ships less than the distance threshold are used as interconnected edges; the basic weight of the edge is calculated according to the speed and course of the ship; the frequency of encounters of each pair of ships in the past period of time is counted, historical data is analyzed, and the degree of interference of ships in a congested state is calculated; the encounter frequency and the degree of congestion interference are combined to calculate the comprehensive influence; the comprehensive influence is combined with the basic edge weight to update the final weight of each edge; all ship nodes are connected to their corresponding dynamically updated edges to form a comprehensive spatiotemporal graph convolutional network model; the ship's feature vectors are used for model training; the trained spatiotemporal graph convolutional network model is used to predict real-time data to obtain potential congestion and conflict situations.

[0158] Furthermore, the expression of the basic weight of the edge is,

[0159] ;

[0160] in, Indicates Ships and The weights of the edges between ships, Indicates Ships and The actual distance between the ships, and Respectively represent Ships and The speed of the ship, and Respectively represent Ships and the course of the vessel;

[0161] The expression of the interference degree of ships in congestion state is:

[0162] ;

[0163] in, Indicates Ships and The degree of congestion and interference between ships, Indicates Ships and The number of encounters between ships, represents the interference coefficient, which is used to quantify the effect of the intersection frequency on the interference level;

[0164] The expression of comprehensive influence is:

[0165] ;

[0166] in, Indicates Ships and The combined influence between the ships.

[0167] It should be noted that the ships are regarded as nodes in the graph to form an interactive network between the ships in order to analyze their mutual influence; the interaction intensity between the ships is quantified by defining the relative position and distance; the degree of behavioral interference of ships in a congested state is analyzed by counting the intersection frequency, which helps to identify potential conflicts; the intersection frequency is combined with the degree of interference to provide a basis for congestion and conflict prediction; the interaction intensity between ships is dynamically adjusted by combining the comprehensive influence with the basic edge weight.

[0168] S5. Based on the group behavior prediction results, optimize the ship path planning and generate the optimal route suggestion.

[0169] S5.1. Treat each ship as an independent player in the game theory and define clear goals for each ship.

[0170] Furthermore, the clear goals include minimizing the voyage, minimizing fuel consumption and maximizing safety. Minimizing the voyage means that the ship hopes to choose the shortest route to reduce sailing time and fuel consumption; minimizing fuel consumption means that the ship hopes to choose an energy-saving route within a reasonable sailing time to reduce operating costs; maximizing safety means that the ship needs to avoid potential collisions and congestion as much as possible to ensure sailing safety.

[0171] S5.2. Construct a payoff matrix to quantify the gains and losses of each ship under different strategies and define the optional strategies for the ships.

[0172] Furthermore, the optional strategies available to the vessel include maintaining the current route, changing course to avoid congested areas, and adjusting speed to adapt to the dynamic behavior of other vessels.

[0173] S5.3. Use an iterative algorithm to calculate the Nash equilibrium so that in the equilibrium state, no ship can obtain better benefits by unilaterally changing its strategy.

[0174] Furthermore, using an iterative algorithm to calculate the Nash equilibrium means updating the strategy of each ship through repeated iterations until a stable state is reached, where no ship is willing to change its strategy. In each iteration, the ship evaluates its payoff based on the current payoff matrix and selects the strategy that brings the maximum payoff.

[0175] It should be noted that in addition to iterative algorithms, genetic algorithms can also be used to calculate Nash equilibrium and find the optimal strategy combination by simulating the biological evolution process (selection, crossover, mutation, etc.). This method is suitable for situations where the strategy space is large and complex, and can effectively explore the diversity of strategy combinations.

[0176] S5.4. Based on the Nash equilibrium results, the path optimization algorithm is applied to calculate the optimal route selection.

[0177] Furthermore, the steps to calculate the optimal route selection are as follows,

[0178] The route map constructed based on the spatiotemporal graph convolutional network is input into the selected optimization algorithm (such as Dijkstra or A* algorithm) to calculate the optimal path from the starting point to the end point. The algorithm will evaluate the advantages of each path according to the set weights and output the route with the shortest time, lowest cost or minimum risk. During the execution process, the optimization algorithm will traverse the nodes and edges in the graph and find the best route by minimizing the total weight of the path. This process not only takes into account the basic weight of each edge, but also uses real-time Nash equilibrium data to ensure that the optimization results can dynamically reflect the actual navigation status of the ship.

[0179] S5.5. Use a multi-objective optimization algorithm to combine multiple well-defined objectives, perform multi-objective optimization, and generate optimal route recommendations.

[0180] Furthermore, the steps of generating the optimal route are as follows: using the concept of Pareto frontier, looking for a solution that cannot be improved on all objectives, that is, the Pareto optimal solution; setting the weight of the objectives according to the specific needs of the ship and the navigation environment, and dynamically adjusting the objective weights according to the real-time situation during navigation to adapt to emergencies (such as weather changes, traffic congestion, etc.); generating a set of route suggestions containing multiple Pareto optimal solutions through an optimization algorithm, each route representing a trade-off under different objectives; comparing the route suggestions with the current environmental conditions (such as wind speed, tidal current, channel congestion, etc.) to screen out suitable routes.

[0181] It should be noted that when generating the optimal route, multi-objective optimization combining traffic flow and navigation environment can better provide decision support for ships and water-related projects:

[0182] Traffic flow impact assessment: Through multi-objective optimization, the system can evaluate the impact of different routes on traffic flow. In high-traffic areas, the system can recommend that ships use temporary channels or adjust their speed to reduce interference with other ships. This analysis also provides site selection recommendations for water-related projects to avoid expansion or construction in high-traffic areas.

[0183] Adaptability to navigation environment: In areas with complex environmental conditions, the system can recommend ships to choose safer and less risky routes. For example, in areas with drastic tidal changes or high wind speeds, the system can recommend ships to take detours through optimization algorithms. For water-related projects, it is recommended to select areas with less tidal influence for construction to reduce the impact on project progress and ship traffic.

[0184] This embodiment also provides a computer device, which is suitable for the case of a maritime traffic flow analysis method based on big data, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the maritime traffic flow analysis method based on big data proposed in the above embodiment.

[0185] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. 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 and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0186] The present embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for analyzing maritime traffic flow based on big data as proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, disk or optical disk.

[0187] In summary, the present invention ensures the accuracy and consistency of data and reduces erroneous analysis caused by data noise by: collecting and preprocessing maritime traffic-related data; using the preprocessed data for adaptive dynamic partitioning, combined with the K-means clustering algorithm, it can reflect the changes in ship density and flow in real time; analyzing ship behavior patterns based on adaptive partitioning, and using normal behavior models for anomaly detection, it can quickly identify abnormal behaviors and generate reports; by analyzing the interactions between ship groups, combining historical data and natural language processing technology, constructing a spatiotemporal graph convolutional network model, and predicting potential navigation conflicts; by optimizing path planning, generating optimal route recommendations based on multi-objective optimization, and further improving navigation efficiency and economy.

[0188] Example 2, referring to Table 1, is the second example of the present invention. To further verify the technical solution of the present invention, experimental simulation data of a maritime traffic flow analysis method based on big data is provided.

[0189] First, collect data related to maritime traffic, including AIS (Automatic Identification System) data, meteorological data, and marine environmental data. AIS data provides real-time ship location information, speed, and heading by receiving signals transmitted by ships. Meteorological data is obtained by weather stations and satellites, including wind speed, wind direction, and visibility. Marine environmental data is obtained through buoy and satellite monitoring, including tides and wave heights.

[0190] After data collection is completed, data preprocessing is performed, including data format standardization, deduplication, cleaning of erroneous records, and processing of missing values. The preprocessed data is used to adaptively and dynamically divide the sea area. The specific steps include: extracting the location information, speed, and heading of each ship from the preprocessed data; determining the grid area based on the ship's location information; using the ship's speed and heading to predict the future position, and mapping the predicted position to the grid area, counting the actual and predicted number of ships; calculating the current and predicted ship density; applying the K-means clustering algorithm to analyze the density and identify congestion.

[0191] On the basis of adaptive partitioning, the real-time behavior data of the ship is extracted, the normal behavior model is constructed, and the machine learning algorithm is used for training. By comparing the real-time ship behavior data with the normal behavior model, the deviation is calculated and the threshold is set. When the deviation exceeds the set value, the behavior is marked as abnormal. The generated abnormal behavior report provides the basis for subsequent analysis.

[0192] Next, the interaction between groups of ships is analyzed to predict potential congestion and conflict. By collecting historical data of abnormal ship behavior, key features are selected, and natural language processing technology is combined to extract relevant numerical features. Ships are regarded as nodes in the graph, relative positions and distances are calculated, and distance thresholds are defined to identify interconnected ships. Based on the frequency of intersections and the degree of congestion interference, the comprehensive influence is calculated, and the spatiotemporal graph convolutional network model is used to predict real-time data to identify potential congestion and conflict situations.

[0193] Finally, based on the group behavior prediction results, the ship path planning is optimized. The goal is defined for each ship, the payment matrix is ​​constructed, the Nash equilibrium is calculated using an iterative algorithm, and the path optimization algorithm is applied to generate the optimal route recommendation.

[0194] The number of ships in Experiment 1 was 100, Experiment 2 increased the number of ships to 150, and Experiment 3 increased the number of ships to 200.

[0195] The details are shown in Table 1 below:

[0196] Table 1 Experimental data record table

[0197]

[0198] By comparing the data of the experimental group and the control group in the table, it can be clearly seen that, when the number of ships is the same, the average speed of the present invention is significantly higher than that of the traditional method. For example, in the experiment of 100 ships, the average speed is 16.2 knots, while the traditional method is only 12.8 knots, which is about 26.5% faster. This shows that after adopting the method of the present invention, the operating efficiency of the ship is significantly improved.

[0199] In terms of maximum density, when there are 100 ships, the maximum density of the present invention is 9.5 ships / square kilometer, compared with 13.1 ships / square kilometer of the traditional method, which shows that the present invention can effectively reduce the risk of congestion and provide protection for navigation safety.

[0200] The proportion of abnormal behavior is an important indicator for evaluating navigation safety. In an experiment with 100 ships, the proportion of abnormal behavior of the present invention was 4.2%, while that of the traditional method was 12.5%, showing the obvious advantages of the present invention in identifying and handling abnormal behavior.

[0201] The optimal route suggestion shows the advantages of the present invention in path optimization. In the experiment of 200 ships, the optimal route suggestion of the present invention is 130, while the traditional method is only 80, which shows that the present invention can provide more reasonable route planning for ships and reduce potential conflicts and congestion.

[0202] In summary, through the comparative analysis of experimental data, it is fully proved that the present invention not only improves the operating efficiency of ships in maritime traffic flow analysis, but also effectively reduces the occurrence of congestion and abnormal behavior, providing strong support for future marine traffic management.

[0203] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for analyzing marine traffic flow based on big data, characterized in that: include, Collect and pre-process data related to maritime traffic; Use the processed data to adaptively and dynamically divide the sea area and generate water boundaries that are updated in real time; Based on the adaptive segmentation results, the ship behavior patterns are analyzed to obtain anomaly detection results; Analyze the interactions between groups of ships based on anomaly detection results to predict potential congestion and conflicts; Based on the group behavior prediction results, optimize the ship path planning and generate the best route suggestions; Use pre-processed data to adaptively and dynamically divide the sea area and generate real-time updated water boundaries The following steps are included: Extract the position information, speed and heading of each ship from the preprocessed data; According to the extracted ship position information, determine the grid area where each ship is located; Using the ship’s speed and heading data, predict the ship’s position at a future point in time; Map the predicted ship positions to grid areas and count the actual and predicted number of ships in each area; Calculate current and predicted ship density based on statistical results; Apply K-means clustering algorithm to perform cluster analysis on current density and predicted density to identify the ship flow and congestion in each area; Based on the results of K-means cluster analysis, combined with real-time ship density, flow direction, speed, and meteorological and oceanographic environmental data, a real-time updated water boundary is formed; Based on the anomaly detection results, analyzing the interaction between ship groups and predicting potential congestion and conflict includes the following steps: Collect historical data and historical voyage records related to abnormal vessel behavior; Select key features based on the ship's behavior patterns and environmental conditions, and perform feature enhancement; Apply natural language processing technology to identify ship behavior patterns and related keywords in specific situations in historical voyage record texts, and convert related keywords into numerical features; The features extracted from the text analysis are combined with the key features to obtain the characteristics of the ship and form a feature vector; Consider each ship as a node in the graph; The relative positions and distances between ships are used as edges; Define a distance threshold, and the ships with distances less than the distance threshold are considered as interconnected edges; Calculate the base weight of the edge based on the ship's speed and heading; Count the frequency of each pair of ships passing each other in the past period of time, analyze historical data, and calculate the degree of interference between ships in a congested state; Combine the intersection frequency and congestion interference degree to calculate the comprehensive impact; Combine the comprehensive influence with the basic edge weight to update the final weight of each edge; Connect all ship nodes with their corresponding dynamically updated edges to form a comprehensive spatiotemporal graph convolutional network model; Use the ship's feature vectors for model training; The trained spatiotemporal graph convolutional network model is used to predict real-time data and obtain potential congestion and conflict situations.

2. The method for analyzing marine traffic flow based on big data according to claim 1, characterized in that: The maritime traffic-related data include AIS data, meteorological data and marine environment data; the AIS data refers to the real-time ship position information, speed and heading provided by receiving signals transmitted by ships; the meteorological data refers to the wind speed, wind direction and visibility obtained by weather stations and satellites; the marine environment data refers to the tide and wave height monitored by buoys and satellites; The preprocessing includes data format standardization, deduplication, error record cleaning and missing value processing.

3. The method for analyzing marine traffic flow based on big data according to claim 2, characterized in that: Based on the adaptive segmentation results, analyzing the ship behavior pattern and obtaining the anomaly detection results include the following steps: Extracting the real-time behavior data of the ship from the adaptive segmentation results; Define the characteristics of normal behavior models based on historical data and domain knowledge; Use machine learning algorithms to model normal behavior; Use historical normal behavior data to train models and identify normal speed, course changes, and ship spacing; Compare the real-time collected ship behavior data with the normal behavior model, and calculate the deviation of each ship in the real-time data and the normal behavior model identification results; Set a deviation threshold. When the deviation exceeds the set threshold, the behavior is automatically marked as abnormal. Based on the deviation results, abnormal behavior reports are automatically generated.

4. The method for analyzing marine traffic flow based on big data according to claim 3, characterized in that: Based on the group behavior prediction results, optimizing the ship path planning and generating the optimal route suggestion include the following steps: Treat each ship as an independent player in the game theory and define clear goals for each ship; Construct a payoff matrix to quantify the gains and losses of each ship under different strategies and define the optional strategies of the ships; Use an iterative algorithm to calculate the Nash equilibrium so that in the equilibrium state, no ship can obtain better benefits by unilaterally changing its strategy; Based on the Nash equilibrium results, the path optimization algorithm is applied to calculate the optimal route selection; Use multi-objective optimization algorithms to combine multiple well-defined objectives, perform multi-objective optimization, and generate optimal route recommendations.

5. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the marine traffic flow analysis method based on big data described in any one of claims 1 to 4 are implemented.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the marine traffic flow analysis method based on big data described in any one of claims 1 to 4 are implemented.

Images