Port berth boundary identification method based on multi-source data fusion

Abstract

The invention discloses a port berth boundary identification method based on multi-source data fusion. The method comprises the following steps: screening a continuous low-speed dense area of a ship based on data of an automatic ship identification system; classifying based on a spatial distance relationship between the low-speed dense area and a mainland shoreline boundary; and artificial parking space identification and boundary identification based on satellite remote sensing image data. According to the method, the possible port berth area is identified on the basis of fusing the AIS data, the geographic information data and the satellite remote sensing image, and finally accurate berth identification and boundary identification are performed through a manual intervention method. The method has better applicability, and improves the accuracy of parking space and boundary identification on the basis of guaranteeing the identification efficiency. Precise identification of port berths is realized, and the method has multiple beneficial effects on port operation and management.

Description

Technical Field

[0001] The invention relates to the field of water transportation, and in particular to a port berth boundary marking method based on multi-source data fusion. Background Art

[0002] With the acceleration of globalization, the volume of international trade continues to rise, which puts higher requirements on the throughput capacity of ports. Therefore, in order to realize the automated management of berths and improve the utilization efficiency of berths, it is very necessary to carry out accurate identification and boundary marking of port berths. The existing technology for port berths and their boundary identification is mainly based on the identification of port berths and their boundaries based on AIS data mining, video surveillance, and laser radar. For example, the DBSCAN density clustering algorithm is used to mine ship berthing points; the DBSCAN density clustering algorithm is used not only to mine ship berthing points, but also to further realize the automatic identification of the location, direction, shoreline length and type of port berths on this basis, and effectively monitor the dynamics of ships and the operation of ships in ports. In addition, the port berth identification using video surveillance and laser radar equipment is mainly used for the identification and monitoring of designated berths or berth groups.

[0003] The current berth identification results based on AIS data through clustering methods are affected by data quality and algorithm parameter settings and have poor applicability. Port berth identification using equipment such as video surveillance and lidar is expensive and only applicable to designated berths. Summary of the invention

[0004] The object of the present invention is to provide a port berth boundary identification method based on multi-source data fusion, so as to solve the above-mentioned problems existing in the prior art.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0006] A port berth boundary identification method based on multi-source data fusion, the method comprising:

[0007] Screening of continuous low-speed dense areas of ships based on the data of the ship automatic identification system;

[0008] Classification is based on the spatial distance relationship between the low-speed dense area and the continental coastline boundary;

[0009] Artificial berth identification and boundary marking based on satellite remote sensing image data.

[0010] Preferably, the field content of the ship automatic identification system signal mainly includes: dynamic information and static information. The dynamic information can be used to obtain the real-time navigation status of the ship;

[0011] A specific ship can be identified using the maritime mobile communication service identification code, ship call sign and International Maritime Organization number in the static information.

[0012] Preferably, the method for screening the continuous low-speed dense area of ​​ships based on the ship automatic identification system data is:

[0013] Clean the data of the ship automatic identification system;

[0014] Extraction of continuous low-speed trajectory of ships;

[0015] Ship berthing area extraction based on adaptive DBSCAN density clustering.

[0016] Preferably, the specific method of cleaning is:

[0017] Delete the AIS data point set with longitude outside the range of [-180 180], latitude outside the range of [-90 90], track direction outside the range of [0 360], and speed above 30 knots;

[0018] Based on the ship speed and time data, delete the AIS data point set with unreachable speed;

[0019] Based on the ship track analysis, delete the AIS data point set with frequent and clustered changes in the ship's track during navigation;

[0020] Based on the ship's track direction and ship speed data, the drift data point set in the AIS data is deleted;

[0021] The specific method for extracting the continuous low-speed trajectory of a ship is as follows:

[0022] Extract AIS data from each ship and segment the AIS data according to the set time difference threshold T1;

[0023] Analyze the ship speed characteristics section by section and extract the AIS track point set with speed less than 0.2 knots continuously and more than 10 points;

[0024] The extracted AIS track point set is analyzed ship by ship, and two or more AIS data track segments whose head and tail time difference is less than the set time difference threshold T2 are merged.

[0025] Preferably, the classification is performed based on the spatial distance relationship between the ship berthing area and the land shoreline boundary;

[0026] By calculating the shortest distance between the ship berthing area and the land shoreline boundary, the ship berthing area with a distance less than the set distance threshold D is selected as the candidate berthing area for the ship; the specific calculation process is as follows:

[0027] According to the WGS84 coordinate system, the longitude and latitude coordinates of the land coastline boundary and the longitude and latitude coordinates of the ship's berthing track are converted into plane coordinates;

[0028] Calculate the coordinates of the center point of the ship's berthing track and calculate its distance from each sub-layer in the land coastline boundary layer;

[0029] Select the sub-layer land boundary point set closest to the center point of the ship's berthing track, and calculate the minimum distance d from the center point of the ship's berthing track to each land boundary;

[0030] If the calculated minimum distance d is less than the set distance threshold D, the boundary of the ship's berthing trajectory point set is calculated according to the convex hull algorithm principle, and the boundary is recorded in the ship candidate berth area data set;

[0031] The spatial overlay analysis method is used to process the boundary data in the candidate berth area dataset. The boundaries with overlap exceeding 80% are taken as the union set, and the boundaries are recalculated according to the convex hull algorithm principle and updated to the candidate berth area dataset.

[0032] Preferably, artificial berth identification and boundary marking based on satellite remote sensing image data:

[0033] The candidate berth area for ships is obtained based on the classification of the spatial distance relationship between the ship berthing area and the land coastline boundary. On this basis, the satellite remote sensing image data of the candidate berth area for ships is called, and the method of manual intervention is used to determine whether the candidate berth area for ships is a real berth; if the judgment result is true, the berth boundary is manually marked.

[0034] Preferably, the specific workflow of artificial berth identification and boundary marking based on satellite remote sensing image data is as follows:

[0035] Satellite remote sensing image data acquisition: Based on the longitude and latitude data of the boundaries of the candidate berth area to be screened, high-resolution remote sensing image data of the area is acquired from satellites or aircraft;

[0036] Satellite remote sensing image data processing:

[0037] Berth identification: Based on the retrieved satellite remote sensing image data, manually identify whether there is a dock or berth in the area;

[0038] Boundary marking: Based on the berth identification results, the boundary marking is manually performed for the area identified as the berth;

[0039] Repeat the above workflow to update the identified berths and boundaries into the existing berth database.

[0040] Preferably, the specific method for processing satellite remote sensing image data is:

[0041] Radiometric calibration: converting the digital number (DN) of an image into a physical quantity, such as reflectivity or radiance, for quantitative analysis;

[0042] Atmospheric correction: remove the influence of atmospheric absorption and scattering on the reflected signal of the ground objects, and restore the true reflectivity of the surface;

[0043] Geometric correction: Use ground control points (GCPs) or digital elevation model (DEM) data to correct the geometric distortion of the image to ensure that the image corresponds to the actual position on the ground;

[0044] Orthorectification: Through DEM and sensor parameters, the oblique image is corrected to ortho projection to eliminate the radiation changes caused by terrain shadows and elevation;

[0045] Image fusion: Fusing high-resolution panchromatic images with low-resolution multispectral images to obtain images with high spatial resolution and high spectral resolution;

[0046] Image stitching: seamlessly stitch multiple images together to form a larger continuous coverage area;

[0047] Color uniformity processing: Adjust the colors of images at different times or in different scenes to make them visually consistent and facilitate comparative analysis.

[0048] Preferably, the boundary marking method is specifically as follows:

[0049] The marked berth boundaries should be rectangular as much as possible;

[0050] The marked berth boundary should include all ship anchorage points, and the boundary should maintain a certain distance from the ship anchorage points;

[0051] The marked berth boundaries do not overlap with existing berth boundaries;

[0052] If the berth to be marked is close to an existing berth, consideration should be given to merging the two berths and re-marking the boundary.

[0053] The beneficial effects of the present invention are as follows: the present invention discloses a port berth boundary identification method based on multi-source data fusion, the method comprising: screening continuous low-speed dense areas of ships based on ship automatic identification system data; classification based on the spatial distance relationship between the low-speed dense areas and the continental shoreline boundary; manual berth identification and boundary identification based on satellite remote sensing image data. This application identifies possible port berth areas based on the fusion of AIS data, geographic information data and satellite remote sensing images, and finally accurately identifies berths and marks boundaries through manual intervention methods. It has better applicability, and improves the accuracy of berth and boundary identification on the basis of ensuring identification efficiency. Accurate identification of port berths has many beneficial effects on port operations and management, mainly including the following aspects:

[0054] (1) Improve port operation efficiency:

[0055] By accurately identifying the location and type of berths, ship berthing can be arranged more effectively, reducing waiting time and unnecessary movements. Accurate information on berths helps optimize the allocation of manpower, equipment and materials, ensuring that various resources are used rationally, thereby improving terminal utilization and overall operational efficiency.

[0056] (2) Ensure the safety of ships:

[0057] Accurate berth information helps avoid collisions between ships and reduce the probability of safety accidents. In an emergency, accurate berth information can help quickly locate the position of the ship, take timely measures and reduce losses.

[0058] (3) Support data analysis and decision making:

[0059] Analyzing the historical operation of berths can help identify bottlenecks in operations and provide a basis for improvement. In addition, data analysis based on berth information can be used to predict future demand and support long-term planning and development strategies.

[0060] (4) Enhanced data-driven decision making:

[0061] Accurate berth information helps to achieve data sharing between ports, shipping companies, shippers and other parties, and promote data-based collaborative optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 This is a flow chart of a port berth boundary identification method based on multi-source data fusion in the present application;

[0063] Figure 2 It is a flow chart of the specific method of AIS data cleaning of this application;

[0064] Figure 3 It is a schematic diagram of the association between the AIS signal and the archive data of the present application;

[0065] Figure 4 This is a schematic diagram of ship berthing point extraction based on adaptive DBSCAN density clustering in this application;

[0066] Figure 5 This is a schematic diagram of the remote sensing image call and boundary identification results of this application. DETAILED DESCRIPTION

[0067] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation methods described herein are only used to explain the present invention and are not used to limit the present invention.

[0068] A port berth boundary identification method based on multi-source data fusion, the method includes the following three stages:

[0069] Screening of continuous low-speed dense areas of ships based on the data of the ship automatic identification system;

[0070] Classification is based on the spatial distance relationship between the low-speed dense area and the continental coastline boundary;

[0071] Artificial berth identification and boundary marking based on satellite remote sensing image data.

[0072] Preferably, the field content of the ship automatic identification system signal mainly includes: dynamic information and static information. The dynamic information can be used to obtain the real-time navigation status of the ship;

[0073] A specific ship can be identified using the maritime mobile communication service identification code, ship call sign and International Maritime Organization number in the static information.

[0074] It should be noted in this embodiment that the Automatic Identification System (AIS) is a technology used for radio communication between ships and between ships and land. The purpose of AIS is to provide real-time information such as ship location, identification, navigation status, etc. to enhance the safety and navigation efficiency of maritime traffic. The transmission frequency of AIS signals is in the second to minute level. Generally, the AIS transmission frequency when the ship is sailing is higher than the frequency when the ship is moored.

[0075] AIS signal is a string of raw data that records the ship's own coordinates, speed, ship name and other data. Due to its high-frequency characteristics, each ship is continuously and frequently broadcasting to the outside world. Therefore, due to its large data volume but low information content, the first step is to perform basic processing of AIS signals to solve the determination of the ship's unique identification, so as to effectively match other attributes of the ship and correctly correspond to subsequent data mining work. In the second step, due to the base stations at the physical layer and the equipment layer spread all over the world, and there are artificial satellite base stations, the standards and quality of the data are different, and the dynamic and static manual filling information of the AIS signal equipment is sent asynchronously, so it is necessary to clean and complete the AIS signal, remove outliers and non-standard data, merge dynamic and static data, and finally repair certain known abnormal data types.

[0076] Preferably, the method for screening the continuous low-speed dense area of ​​ships based on the ship automatic identification system data is:

[0077] Clean the data of the ship automatic identification system;

[0078] Extraction of continuous low-speed trajectory of ships;

[0079] Ship berthing area extraction based on adaptive DBSCAN density clustering.

[0080] (1) AIS data cleaning

[0081] The MMSI code is the identifier of the ship in the AIS data. The first three digits of the MMSI code are the flag state code, and the last six digits are the ship's specific identification code. Since the MMSI code is manually entered, there are objective input errors or human modifications. There are mainly three types of MMSI code errors, namely, the MMSI length is not 9 digits, the first three digits of the MMST code cannot correspond to the correct country and region identification code, and the MMSI code cannot correspond to a valid ship type. Therefore, the data with the above errors is deleted during the data cleaning process.

[0082] At present, the commercial berths in ports mainly serve passenger ships, cargo ships, ro-ro ships and other types of ships. In order to improve the efficiency and accuracy of port berth identification, the AIS data of the above-mentioned types of ships are mainly identified. Therefore, it is necessary to match the dynamic and static information of AIS to filter the data of the target ship type. In actual situations, it is difficult for us to perform a true full sample analysis based on AIS signal data at one time. Instead, we need to combine the different ship parameter properties recorded in various different ship information to match data from different sources to achieve a lower error rate. Figure 3 The dynamic information part of the AIS signal only contains the information of the MMSI ID, which needs to be matched with other data to obtain the complete field information. Because the static information part of the AIS signal is generated asynchronously, the dynamic and static information must first be spliced ​​according to the timestamp based on the same MMSI code, and then the ship name, IMO number, etc. can be used for further data association.

[0083] Because although the MMSI code of each ship is unique based on the broadcasting principle of the AIS system, that is, the MMSI code in the dynamic information and static information of a ship is consistent, there are a large number of repeated or missing MMSI codes in the actual data, so the static information and dynamic information data cannot be matched. For such incomplete ship data information, only the ship AIS signal data containing complete dynamic and static information is retained.

[0084] AIS data may be affected by hardware failure, signal interference, environmental factors and other factors, which may generate noise and lead to inaccurate or distorted data. Therefore, in order to ensure the accuracy of port berth identification, it is necessary to clean the longitude, latitude and speed data of AIS data on the basis of removing duplicate data. The idea of ​​AIS data cleaning method is to remove noise points based on deleting numerical anomalies and combining the characteristics of ship motion.

[0085] Preferably, the specific method of AIS data cleaning is:

[0086] Delete the AIS data point set with longitude outside the range of [-180 180], latitude outside the range of [-90 90], track direction outside the range of [0 360], and speed above 30 knots;

[0087] Based on the ship speed and time data, delete the AIS data point set with unreachable speed;

[0088] Based on the ship track analysis, delete the AIS data point set with frequent and clustered changes in the ship's track during navigation;

[0089] Based on the ship's track direction and ship speed data, the drift data point set in the AIS data is deleted;

[0090] (2) Extraction of continuous low-speed ship trajectories:

[0091] When a ship is moored on the water, it will swing around the mooring point or anchor point due to the influence of wind and current. The swing amplitude is affected by the mooring method of the ship. Generally speaking, the swing amplitude of a moored ship is smaller than that of an anchored ship. Therefore, a series of continuous low-speed motion points will be generated during the mooring process. And considering that the speed of a ship may change suddenly due to wind and current, ship maneuvering or data anomalies during the mooring process, after extracting the low-speed points from the AIS data, the obtained AIS point set needs to be merged to improve the recognition efficiency of the berth.

[0092] The specific method for extracting the continuous low-speed trajectory of a ship is as follows:

[0093] Extract AIS data from each ship and segment the AIS data according to the set time difference threshold T1;

[0094] Analyze the ship speed characteristics section by section and extract the AIS track point set with speed less than 0.2 knots continuously and more than 10 points;

[0095] The extracted AIS track point set is analyzed ship by ship, and two or more AIS data track segments whose head and tail time difference is less than the set time difference threshold T2 are merged.

[0096] (3) Ship berthing area extraction based on adaptive DBSCAN density clustering

[0097] In view of the unique dynamic behavior of ships on the water, their continuous trajectories at low speed may be mooring trajectories, wandering trajectories or slow-speed sailing trajectories. The mooring trajectories are different from other types of trajectories. Their points are generally concentrated in a certain range of waters. Therefore, in order to extract the ship mooring area from the continuous low-speed trajectory of the ship, this patent adopts an adaptive DBSCAN density clustering method for identification.

[0098] The traditional DBSCAN (Density-Based Spatial Clustering of Applications with Noise) clustering algorithm relies on two key parameters: ε (epsilon, neighborhood radius) and MinPts (minimum number of neighborhood points). However, these parameters may require different optimal values ​​on different data sets or data subsets, which makes manual parameter adjustment difficult, especially when the density of berthing trajectory points for different ships is different. The adaptive DBSCAN algorithm used in this patent aims to solve this problem by dynamically adjusting these parameters to adapt to the local data density to avoid over-clustering or insufficient clustering. The basic process of the algorithm is as follows:

[0099] (a) Calculate local density: For each point p in the trajectory segment dataset, calculate the number of points in its surrounding neighborhood. Obtain a local density estimate for each point;

[0100] (b) Determine core points: Based on the calculated local density, determine which points can be regarded as core points (core points are those points that have enough neighbors in their neighborhood);

[0101] (c) Dynamic adjustment of neighborhood: For each core point, the neighborhood size is adjusted according to its local density, so that the neighborhood is smaller in high-density areas and larger in low-density areas;

[0102] (d) Expanding clusters: Starting from any unprocessed core point, find all directly density-reachable points (i.e., points within the adjusted neighborhood) and mark them as belonging to the same cluster. Mark these points as visited and check if they are also core points. If so, continue expanding the cluster until no new points can be added;

[0103] (e) Handling boundary points and noise points: assign boundary points to the cluster where the nearest core point is located, and mark noise points as not belonging to any cluster (boundary points are those points that do not belong to the neighborhood of any core point, but they are in the neighborhood of a core point. Noise points are those points that are neither core points nor boundary points.);

[0104] (f) repeating steps d and e;

[0105] (g) Output results: The final output is the clustering division of the data set, including the members of each cluster and the points marked as noise.

[0106] Preferably, the classification is performed based on the spatial distance relationship between the ship berthing area and the land shoreline boundary;

[0107] Land coastline boundary data, i.e., geographic information describing the boundary between land and sea, is one of the basic data in many fields such as earth science, oceanography, geographic information system (GIS) and environmental planning. It usually contains information such as the geographic coordinates of boundary points and boundary point numbers. There are various ways to obtain this type of data, and the main sources include government agencies, international organizations, academic research institutions and commercial data providers.

[0108] The commonly used storage format for land coastline boundary data is SHP format. Shapefile is a vector data storage format developed by ESRI (Environmental Systems Research Institute) for geographic information systems (GIS). A complete Shapefile dataset is actually composed of a set of related files, not just a single shp file. SHP is the main file used to store geometric data. It contains the geometric information of each feature, such as the coordinates of the point, the vertex list of the line, the boundary of the polygon, etc.

[0109] Considering that port berths are generally located in coastal waters along the mainland or islands, the shortest distance from the ship berthing area to the land coastline boundary is calculated, and the ship berthing area with a distance less than the set distance threshold D is selected as the candidate berthing area for the ship; the specific calculation process is as follows:

[0110] (1) According to the WGS84 coordinate system, the longitude and latitude coordinates of the land coastline boundary and the longitude and latitude coordinates of the ship's berthing track are converted into plane coordinates;

[0111] (2) Calculate the coordinates of the center point of the ship's berthing track and calculate its distance from each sub-layer in the land coastline boundary layer;

[0112] (3) Select the land boundary point set of the sub-layer closest to the center point of the ship's berthing track, and calculate the minimum distance d from the center point of the ship's berthing track to each land boundary;

[0113] (4) If the calculated minimum distance d is less than the set distance threshold D, the boundary of the ship's berthing trajectory point set is calculated according to the convex hull algorithm principle, and the boundary is recorded in the ship candidate berth area data set;

[0114] (5) The spatial overlay analysis method is used to process the boundary data in the candidate berth area dataset. The boundaries with overlap exceeding 80% are taken as the union set, and the boundaries are recalculated according to the convex hull algorithm principle and updated to the candidate berth area dataset.

[0115] Preferably, artificial berth identification and boundary marking based on satellite remote sensing image data:

[0116] The candidate berthing area for ships obtained based on the classification of the spatial distance relationship between the ship berthing area and the land shoreline boundary may be a real berthing area, or it may be an anchorage or waiting area near the port area. Therefore, it is necessary to call the satellite remote sensing image data of the candidate berthing area for ships on this basis, and judge whether the candidate berthing area for ships is a real berth through manual intervention; if the judgment result is true, the berth boundary is manually marked.

[0117] Preferably, the specific workflow of artificial berth identification and boundary marking based on satellite remote sensing image data is as follows:

[0118] Satellite remote sensing image data acquisition: Based on the longitude and latitude data of the boundaries of the candidate berth area to be screened, high-resolution remote sensing image data of the area is acquired from satellites or aircraft;

[0119] Satellite remote sensing image data processing;

[0120] Berth identification: Based on the retrieved satellite remote sensing image data, manually identify whether there is a dock or berth in the area;

[0121] Boundary marking: Based on the berth identification results, the boundary marking is manually performed for the area identified as the berth;

[0122] Repeat the above workflow to update the identified berths and boundaries into the existing berth database.

[0123] Preferably, in order to improve the visual effect of the image, enhance the contrast and clarity of the image, and facilitate the need for berth identification, the following processing is required:

[0124] The specific method of satellite remote sensing image data processing is:

[0125] (a) Radiometric calibration: converting the digital number (DN) of an image into physical quantities, such as reflectivity or radiance, for quantitative analysis;

[0126] (b) Atmospheric correction: remove the influence of atmospheric absorption and scattering on the reflected signal of the ground object and restore the true reflectivity of the ground surface;

[0127] (c) Geometric correction: using ground control points (GCPs) or digital elevation model (DEM) data to correct the geometric distortion of the image and ensure that the image corresponds to the actual position on the ground;

[0128] (d) Orthorectification: Using DEM and sensor parameters, the oblique image is corrected to ortho projection to eliminate the radiation changes caused by terrain shadows and elevation;

[0129] (e) Image fusion: Fusion of high-resolution panchromatic images with low-resolution multispectral images to obtain images with high spatial resolution and high spectral resolution;

[0130] (f) Image stitching: stitching multiple images together seamlessly to form a larger continuous coverage area;

[0131] (g) Color uniformity processing: Adjust the colors of images at different times or in different scenes to make them visually consistent and facilitate comparative analysis.

[0132] Preferably, the boundary marking method is specifically as follows:

[0133] (a) The marked berth boundary should be rectangular as far as possible;

[0134] (b) The marked berth boundary should include all ship anchorage points and the boundary should maintain a certain distance from the ship anchorage points;

[0135] (c) the marked berth boundary does not overlap with the existing berth boundary;

[0136] (d) If the berth to be marked is close to an existing berth, consideration should be given to merging the two berths and re-marking the boundary.

[0137] The beneficial effects of the present invention are as follows: the present invention discloses a port berth boundary identification method based on multi-source data fusion, the method comprising: screening continuous low-speed dense areas of ships based on ship automatic identification system data; classification based on the spatial distance relationship between the low-speed dense areas and the continental shoreline boundary; manual berth identification and boundary identification based on satellite remote sensing image data. This application identifies possible port berth areas based on the fusion of AIS data, geographic information data and satellite remote sensing images, and finally accurately identifies berths and marks boundaries through manual intervention methods. It has better applicability, and improves the accuracy of berth and boundary identification on the basis of ensuring identification efficiency. Accurate identification of port berths has many beneficial effects on port operations and management, mainly including the following aspects:

[0138] (1) Improve port operation efficiency:

[0139] By accurately identifying the location and type of berths, ship berthing can be arranged more effectively, reducing waiting time and unnecessary movements. Accurate information on berths helps optimize the allocation of manpower, equipment and materials, ensuring that various resources are used rationally, thereby improving terminal utilization and overall operational efficiency.

[0140] (2) Ensure the safety of ships:

[0141] Accurate berth information helps avoid collisions between ships and reduce the probability of safety accidents. In an emergency, accurate berth information can help quickly locate the position of the ship, take timely measures and reduce losses.

[0142] (3) Support data analysis and decision making:

[0143] Analyzing the historical operation of berths can help identify bottlenecks in operations and provide a basis for improvement. In addition, data analysis based on berth information can be used to predict future demand and support long-term planning and development strategies.

[0144] (4) Enhanced data-driven decision making:

[0145] Accurate berth information helps to achieve data sharing between ports, shipping companies, shippers and other parties, and promote data-based collaborative optimization.

[0146] In summary, accurate identification of port berths can not only significantly improve port operation efficiency, but also enhance safety, improve customer experience, support data analysis and decision-making, promote intelligent management, reduce environmental impact, and enhance data-driven decision-making capabilities.

[0147] In summary, the present invention analyzes the continuous low-speed sections of ships through AIS data, and screens the continuous low-speed dense areas of ships through the DBSCAN density clustering method. On this basis, the satellite remote sensing image data of the low-speed dense area whose distance to the coastal boundary is less than the distance threshold is selected. Finally, the berths are identified and the berth boundaries are marked by convolutional graph neural network, and a global port berth information database is established to prepare for future tasks such as ship arrival and departure identification, cargo flow and flow direction prediction, etc.

[0148] Compared with the prior art, the multi-source data fusion port berth boundary identification method adopted in the present invention can identify and screen the continuous low-speed dense area of ​​ships based on AIS data, and judge whether it is a high-probability area of ​​port berths according to its geographic information characteristics. On this basis, according to the satellite remote sensing image data of the high-probability area, it can manually judge whether it is a berth in the port area and mark the boundary of the berth. On the basis of ensuring the efficiency of berth boundary identification, more accurate berth identification is achieved.

[0149] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be considered as the scope of protection of the present invention.

Claims

1. A port berth boundary identification method based on multi-source data fusion, characterized in that: The method comprises: Screening of continuous low-speed dense areas of ships based on the data of the ship automatic identification system; Classification is performed based on the spatial distance relationship between the low-speed dense area and the continental coastline boundary; Artificial berth identification and boundary marking based on satellite remote sensing image data.

2. The port berth boundary identification method based on multi-source data fusion according to claim 1 is characterized in that: The field content of the ship automatic identification system signal mainly includes two parts: dynamic information and static information. The dynamic information can be used to obtain the real-time navigation status of the ship; A specific ship can be identified by using the maritime mobile communication service identification code, ship call sign, and International Maritime Organization number in the static information.

3. The port berth boundary identification method based on multi-source data fusion according to claim 2 is characterized in that: The method for screening the continuous low-speed dense area of ​​ships based on the ship automatic identification system data is: Cleaning the ship automatic identification system data; Extraction of continuous low-speed trajectory of ships; Ship berthing area extraction based on adaptive DBSCAN density clustering.

4. The port berth boundary identification method based on multi-source data fusion according to claim 3 is characterized in that: The specific method of the cleaning is: Delete the AIS data point set with longitude outside the range of [-180 180], latitude outside the range of [-90 90], track direction outside the range of [0 360], and speed above 30 knots; Based on the ship speed and time data, delete the AIS data point set with unreachable speed; Based on the ship track analysis, delete the AIS data point set with frequent and clustered changes in the ship's track during navigation; Based on the ship's track direction and ship speed data, the drift data point set in the AIS data is deleted; The method for extracting the continuous low-speed trajectory of the ship is specifically as follows: Extract AIS data from each ship and segment the AIS data according to the set time difference threshold T1; Analyze the ship speed characteristics section by section and extract the AIS track point set with speed less than 0.2 knots continuously and more than 10 points; The extracted AIS track point set is analyzed ship by ship, and two or more AIS data track segments whose head and tail time difference is less than the set time difference threshold T2 are merged.

5. The port berth boundary identification method based on multi-source data fusion according to claim 4 is characterized in that: Classification is based on the spatial distance relationship between the ship berthing area and the land shoreline boundary; By calculating the shortest distance between the ship berthing area and the land shoreline boundary, the ship berthing area with a distance less than the set distance threshold D is selected as the candidate berthing area for the ship; the specific calculation process is as follows: According to the WGS84 coordinate system, the longitude and latitude coordinates of the land coastline boundary and the longitude and latitude coordinates of the ship's berthing track are converted into plane coordinates; Calculate the coordinates of the center point of the ship's berthing track and calculate its distance from each sub-layer in the land coastline boundary layer; Select the sub-layer land boundary point set closest to the center point of the ship's berthing track, and calculate the minimum distance d from the center point of the ship's berthing track to each land boundary; If the calculated minimum distance d is less than the set distance threshold D, the boundary of the ship's berthing trajectory point set is calculated according to the convex hull algorithm principle, and the boundary is recorded in the ship candidate berth area data set; The spatial overlay analysis method is used to process the boundary data in the candidate berth area dataset. The boundaries with overlap exceeding 80% are taken as the union set, and the boundaries are recalculated according to the convex hull algorithm principle and updated to the candidate berth area dataset.

6. The port berth boundary identification method based on multi-source data fusion according to claim 5 is characterized in that: Artificial berth identification and boundary marking based on satellite remote sensing image data: Based on the classification of the spatial distance relationship between the ship berthing area and the land shoreline boundary, the candidate ship berthing area is obtained, and on this basis, the satellite remote sensing image data of the candidate ship berthing area is called, and the method of manual intervention is used to determine whether the candidate ship berthing area is a real berth; If the judgment result is true, the berth boundary is manually drawn.

7. The port berth boundary identification method based on multi-source data fusion according to claim 6 is characterized in that: The specific workflow of artificial berth identification and boundary marking based on satellite remote sensing image data is as follows: Satellite remote sensing image data acquisition: Based on the longitude and latitude data of the boundaries of the candidate berth area to be screened, high-resolution remote sensing image data of the area is acquired from satellites or aircraft; Satellite remote sensing image data processing: Berth identification: Based on the retrieved satellite remote sensing image data, manually identify whether there is a dock or berth in the area; Boundary marking: Based on the berth identification results, the boundary marking is manually performed for the area identified as the berth; Repeat the above workflow to update the identified berths and boundaries into the existing berth database.

8. The port berth boundary identification method based on multi-source data fusion according to claim 7 is characterized in that: The specific method for processing the satellite remote sensing image data is as follows: Radiometric calibration: converting the digital number (DN) of an image into a physical quantity, such as reflectivity or radiance, for quantitative analysis; Atmospheric correction: remove the influence of atmospheric absorption and scattering on the reflected signal of the ground objects, and restore the true reflectivity of the surface; Geometric correction: Use ground control points (GCPs) or digital elevation model (DEM) data to correct the geometric distortion of the image to ensure that the image corresponds to the actual position on the ground; Orthorectification: Through DEM and sensor parameters, the oblique image is corrected to ortho projection to eliminate the radiation changes caused by terrain shadows and elevation; Image fusion: Fusing high-resolution panchromatic images with low-resolution multispectral images to obtain images with high spatial resolution and high spectral resolution; Image stitching: seamlessly stitch multiple images together to form a larger continuous coverage area; Color uniformity processing: Adjust the colors of images at different times or in different scenes to make them visually consistent and facilitate comparative analysis.

9. The port berth boundary identification method based on multi-source data fusion according to claim 8 is characterized in that: The boundary identification method is specifically as follows: The marked berth boundaries should be rectangular as much as possible; The marked berth boundary should include all ship anchorage points, and the boundary should maintain a certain distance from the ship anchorage points; The marked berth boundaries do not overlap with existing berth boundaries; If the berth to be marked is close to an existing berth, consideration should be given to merging the two berths and re-marking the boundary.

Images