Satellite data-based digital buoy positioning method and system

Abstract

The application relates to the technical field of buoy positioning, and discloses a digital buoy positioning method and system based on satellite data. The method comprises the following steps: acquiring satellite data and preprocessing the satellite data, and extracting feature information of a buoy; dividing a buoy position partition based on the feature information, and constructing a mapping relationship between the feature information and the position partition; subsequently, extracting a key partition, calculating a combination matching degree of position data to obtain a position bias combination; then, updating positioning information in the position partition based on the position bias combination; finally, generating a buoy positioning management library according to the updated mapping relationship. Through the steps of accurately extracting feature information, scientifically dividing a position partition, and establishing a reasonable mapping relationship, the method improves the accuracy and timeliness of buoy positioning, and the generated positioning management library can provide reliable position support for buoy-related applications.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of buoy positioning, in particular to a digital buoy positioning method and system based on satellite data. BACKGROUND

[0002] In many fields such as marine monitoring, water traffic management, fishery operation and environmental monitoring, the accurate position information of buoy as an important water monitoring and marking device is crucial for the orderly development of various work. The traditional buoy positioning method relies on ground station signal or short-distance wireless communication technology, which has obvious limitations in practical application. When the coverage of ground station is limited, or in open waters far from land and complex marine environment, the ground station signal will be greatly weakened or even disappear, resulting in interruption of buoy positioning signal and inability to continuously obtain accurate position data. At the same time, the short-distance wireless communication technology is limited by communication distance and can only realize the transmission and acquisition of buoy positioning information in a small range, which is difficult to meet the demand of positioning multiple buoys in a large range of water.

[0003] With the development of satellite technology, satellite positioning has been gradually applied to the field of buoy positioning. However, the current buoy positioning method based on satellite data often fails to fully extract the feature information of the buoy in the process of satellite data processing, resulting in lack of accurate basis for judging the position of the buoy. When dividing the position partition of the buoy, a simple geographical area division method is often used without considering the actual characteristics of the buoy and the characteristics of the satellite data, which makes the relevance between the position partition and the actual position of the buoy low. In addition, when establishing the mapping relationship between the feature information and the position partition, the mapping logic is relatively single, which cannot accurately reflect the buoy position range corresponding to different feature information, and further affects the accuracy of updating the subsequent positioning information. At the same time, in the process of extracting key partitions and calculating the combination matching degree of position data, there is a lack of scientific and reasonable calculation method, which makes it difficult to accurately select the key partitions that are important for buoy positioning and also cannot accurately calculate the combination matching degree of position data, resulting in deviation between the obtained position combination and the actual position of the buoy, and finally the generated buoy positioning management library lacks accuracy, which cannot provide reliable position support for the actual buoy management and application, and seriously affects the performance of buoy in various fields of application. SUMMARY

[0004] The present application aims to provide a digital buoy positioning method based on satellite data to solve the problems raised in the background.

[0005] To achieve the above-mentioned purpose, the present application provides a digital buoy positioning method based on satellite data, which comprises:

[0006] acquiring satellite data and preprocessing the satellite data to extract feature information of the buoy.

[0007] Divide a buoy position partition based on the feature information;

[0008] Construct a mapping relationship between the feature information and the position partition;

[0009] Extract a key partition and calculate a combined matching degree of position data to obtain a position bias combination;

[0010] Update positioning information in the position partition based on the position bias combination;

[0011] Generate a buoy positioning management library according to the updated mapping relationship.

[0012] Preferably, the feature information of the buoy includes:

[0013] Obtain a signal classification model for a buoy signal in satellite data;

[0014] Classify the satellite data using the signal classification model to obtain at least one signal category;

[0015] Identify feature points, including position points, velocity points, and related points, existing in the satellite data under the corresponding signal category to form a feature sample set;

[0016] Analyze the position points, velocity points, and related points in the feature sample set respectively to obtain a position point area, a velocity point analysis area, and a related point association area corresponding to the feature sample set in sequence as the feature information.

[0017] Preferably, obtaining the position point area, the velocity point analysis area, and the related point association area corresponding to the feature sample set includes:

[0018] Merge the position points, velocity points, and related points in the feature sample set according to the signal category to obtain a plurality of merging results;

[0019] Extract a position point pair in the merging result, compare the position point pair with a position database, and obtain a position point area;

[0020] Extract a change rate of the velocity points and an association strength of the related points in the merging result, divide the merging result according to the change rate of the velocity points and the association strength of the related points, and obtain a velocity point analysis area and a related point association area.

[0021] Preferably, dividing the buoy position partition includes:

[0022] The position partition is judged, the positioning times and positioning times of the buoy in the position partition are analyzed, the position partition is fitted according to the positioning times and the positioning times, and a mapping relationship of the position partition to the actual position is constructed.

[0023] Preferably, constructing the mapping relationship of the feature information and the position partition comprises:

[0024] The position data corresponding to the satellite data and the historical trajectory are called to generate a plurality of unlabeled positioning recognition results;

[0025] It is judged whether the plurality of unlabeled positioning recognition results is a target positioning recognition result, if it is the target positioning recognition result, the target positioning recognition result is regarded as the position partition.

[0026] Preferably, constructing the mapping relationship of the feature information and the position partition comprises: the information of the position point area, the speed point analysis area and the related point association area existing in the feature information, the description information of the position partition and the type of the position partition, and the mapping relationship of the feature information and the position partition is constructed.

[0027] Preferably, calculating the combination matching degree of the position data comprises:

[0028] The feature information and the position partition are clustered and analyzed according to the data type, the position type and the function type, and the maximum cluster center after clustering analysis is set as a key partition;

[0029] The keywords of the key partition are extracted, the similarity between each keyword is calculated, and the public sequence related to the similarity between each keyword is set;

[0030] The position data existing in the public sequence is extracted by using the public sequence related to the similarity between each keyword, and the longest common subsequence between each position data is set, and the length value of the longest common subsequence is set as the combination matching degree between each position data;

[0031] The combination matching degree between each position data is set as a position bias combination according to the time distribution probability of each position data.

[0032] Preferably, updating the positioning information in the position partition comprises:

[0033] The time distribution probability of each position data is extracted from the position bias combination;

[0034] The target path of the position bias combination is set according to the time period corresponding to the time distribution probability of each position data.

[0035] fitting a target path of each position data in the position bias combination to obtain a fitted target path, and setting a probability value of the fitted target path in each time period as a combination occurrence probability of the position bias combination;

[0036] comparing the combination occurrence probability of the position bias combination with a reference condition of historical trajectories in the position partition, identifying an occurrence difference value, and classifying positioning information in the position partition according to the occurrence difference value to complete updating.

[0037] Preferably, generating the buoy positioning management library comprises:

[0038] According to the updated mapping relationship, extracting a positioning target in satellite data, and sorting the positioning target according to an occurrence probability of the positioning target to obtain a positioning sequence.

[0039] Combining the positioning target and the positioning sequence in a structured form to obtain the buoy positioning management library.

[0040] Preferably, the application further comprises a digital buoy positioning system based on satellite data, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the steps of the digital buoy positioning method based on satellite data when executing the computer program.

[0041] Compared with the prior art, the application has the following beneficial effects:

[0042] The digital buoy positioning method based on satellite data can mine key information closely related to the position of the buoy from the satellite data by preprocessing the satellite data and extracting feature information of the buoy, thereby providing an accurate information basis for subsequent buoy positioning work. Compared with the traditional method in which satellite data processing is relatively rough and feature information extraction is insufficient, the present method effectively improves the integrity and accuracy of the feature information of the buoy, so that the subsequent positioning operation based on the feature information has a more reliable basis.

[0043] In the process of dividing the buoy position partition based on the feature information, the limitations of simply dividing according to the geographical area in the traditional method are overcome, and the feature information of the buoy and the characteristics of the satellite data are fully combined, so that the position partition is more in line with the actual position distribution of the buoy, and the relevance of the position partition and the actual position of the buoy is enhanced. This scientific partition method can more accurately define the possible position range of the buoy, lay a good foundation for subsequent mapping relationship construction and positioning information updating, and avoid positioning deviation problems caused by unreasonable partition.

[0044] The mapping relationship between the feature information and the location partition is constructed, which breaks the single disadvantage of the traditional mapping logic, and a multi-dimensional and multi-logic mapping relationship is established according to the characteristics of different feature information of the buoy and the corresponding location partition attributes. This mapping relationship can more comprehensively and accurately reflect the corresponding relationship between the feature information and the location partition. When new feature information of the buoy is obtained, the corresponding location partition can be quickly located through the mapping relationship, which greatly improves the efficiency and accuracy of the conversion of feature information to location partition, and reduces the location judgment errors caused by insufficient mapping logic.

[0045] The key partition is extracted and the combination matching degree of the location data is calculated to obtain the location bias combination. A scientific and reasonable key partition extraction method can accurately select the key partition that plays a decisive role in the positioning of the buoy from a large number of location partitions, and exclude irrelevant or less influential partition interference, focusing on the core positioning area. At the same time, the combination matching degree of the location data is calculated through a reasonable calculation method, which can accurately judge the degree of fit between different location data combinations and the actual location of the buoy, and then obtain a location bias combination that is more consistent with the actual location of the buoy, effectively reducing the deviation between the location bias combination and the actual location, and improving the positioning accuracy.

[0046] The location information in the location partition is updated based on the location bias combination, which can timely integrate the latest and accurate location bias combination into the location information in the location partition, realizing dynamic optimization of the location information. Compared with the situation that the location information updating is lagging and the updating basis is insufficient in the traditional method, the location information of the location partition in this method always maintains a high consistency with the actual location of the buoy, ensuring the timeliness and accuracy of the location information, and providing the latest reference for the subsequent judgment of the buoy location.

[0047] The buoy positioning management library is generated according to the updated mapping relationship, which integrates the optimized location information and mapping relationship through multiple links, so that the data in the buoy positioning management library has high precision and reliability. The management library can clearly and accurately present the location information of the buoy and the related feature association, providing comprehensive and accurate location data support for the buoy management and application in the fields of marine monitoring, water traffic management, fishery operation, etc., which helps to improve the efficiency of buoy application in various fields, promotes the application of buoy in actual scenes to be more efficient and orderly, and solves the problem of insufficient precision of traditional buoy positioning management library and inability to meet the actual application requirements. BRIEF DESCRIPTION OF DRAWINGS

[0048] Figure 1 The working principle diagram of the digital buoy positioning method based on satellite data is described in the present application;

[0049] Figure 2 The flowchart for extracting the feature information of the buoy is described in the present application;

[0050] Figure 3 Flow chart for dividing the buoy position partition;

[0051] Figure 4 Flow chart for position data combination matching degree calculation and position bias combination acquisition. DETAILED DESCRIPTION

[0052] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work are within the protection scope of the present application.

[0053] Please refer to Figure 1 The present application provides a digital buoy positioning method and system based on satellite data, which comprises:

[0054] By acquiring satellite data and preprocessing, the feature information of the buoy is extracted, then the buoy position partition is divided based on the feature information, the mapping relationship between the feature information and the position partition is constructed, the key partition is extracted and the combination matching degree of the position data is calculated to obtain the position bias combination, and then the positioning information in the position partition is updated, and finally the buoy positioning management library is generated. In specific implementation, the satellite data includes but is not limited to remote sensing image, signal strength data and buoy motion parameters, the preprocessing operation covers data cleaning, format standardization and noise filtering. Feature information extraction involves analysis and classification of buoy-related signals in satellite data, position partition division is based on spatial distribution characteristics and motion state of the buoy for region segmentation, and mapping relationship construction is realized through data association and pattern recognition. Key partition extraction is based on clustering analysis method, combination matching degree calculation uses sequence comparison and probability statistics method, and position bias combination generation depends on time distribution and path fitting. Positioning information update is completed through difference comparison and classification adjustment, and buoy positioning management library is integrated and stored in the form of structured data.

[0055] Embodiment 1: refer to Figure 2Satellite data is first received and pre-processed, which includes decoding, calibration, and de-noising of the raw satellite-transmitted data stream. Data sources can include multi-spectral imagers, synthetic aperture radar, and auxiliary data from the Automatic Identification System (AIS). After format unification, these data are input into a pre-trained signal classification model. This model is a deep neural network architecture trained on a large amount of historical satellite data containing various known types of buoy signals and environmental interference samples. The model is able to recognize spectral features, spatial distribution patterns, and temporal sequence behaviors of signals, thus classifying input satellite data segments into different signal categories, such as fixed navigation buoys, floating monitoring platforms, mobile operational buoys, or non-buoy interference signals (such as ships, oceanic drifts, or background noise).

[0056] For each identified signal category, the system further resolves its contained feature points. Position points are extracted by analyzing the signal's georeferencing information. Each signal is accompanied by latitude and longitude coordinates, which can be subject to certain deviations due to satellite positioning errors. The system identifies coordinates of the same signal at consecutive timestamps, forming a sequence of position points. Velocity points are calculated relying on these consecutive position information. By calculating the displacement difference between adjacent time points and dividing by the time interval, instantaneous velocity vectors are obtained. These velocity values and their directions collectively constitute velocity point data. Correlation points are determined more complexly, involving multi-dimensional attribute analysis of signals. For example, the system analyzes signal intensity stability, interaction patterns with neighboring signals (such as signal collision or synchronization phenomena), and the matching degree of their spectral features with known buoy types. These attributes are calculated through correlation algorithms, ultimately generating indicators representing the strength of signal interconnections.

[0057] All these feature points: position points, velocity points, and correlation points, are grouped and merged according to their belonging signal categories. All data points under each signal category are aggregated into a merged result dataset. For example, all signals classified as "fixed navigation buoys" and their feature points are merged into an independent data structure.

[0058] The system processes these merged results to generate final feature information. For the position point area, the system extracts position point pairs (i.e., coordinate pairs of the same signal at different times) from the merged results and compares these pairs with a high-precision reference location database. This database stores precise coordinates of known buoys and their acceptable error ranges. Through coordinate matching algorithms, the system determines whether each position point pair falls within the tolerance range of known buoys, thus identifying reliable position point areas. These areas represent geographical ranges where signal positions highly coincide with known buoy positions.

[0059] For the velocity point analysis zone, the system calculates the rate of change for each velocity point. The rate of change is determined by analyzing the trend of velocity values over time, such as calculating the frequency of acceleration or sudden changes in velocity direction. The system sets a rate of change threshold, and velocity points with a rate of change above the threshold are classified as high dynamic regions, while those below the threshold are classified as low dynamic regions. High dynamic regions may correspond to moving buoys or buoys affected by strong currents, while low dynamic regions correspond to fixed or steadily floating buoys. These regions are labeled as velocity point analysis zones, providing a quantitative description of the state of buoy movement. The system assesses the strength of association for each relevant point, i.e., the degree of interaction between the signal and other signals or environmental factors. The strength of association is calculated using statistical methods such as cross-correlation functions or cluster analysis. Points with high strength of association are classified as strong association regions, indicating that the signal may belong to a buoy network or interact with other marine facilities; points with low strength of association are classified as weak association regions, representing independent or isolated buoys. These regions together form the relevant point association zone.

[0060] The position point zone, velocity point analysis zone, and relevant point association zone are integrated into a complete set of feature information. This set of information not only describes the static position of the buoy, but also reveals its dynamic behavior and usage context, providing a multi-level, multi-dimensional data foundation for subsequent partitioning and mapping relationship construction. The entire implementation process relies on continuous data stream iteration, and the model will be updated regularly with new satellite data to adapt to changes in buoy position or state, ensuring the continuous accuracy of feature information extraction.

[0061] Example 2: Referring to Figure 3 The division of position zones is based on feature information extracted from satellite data, which includes the position point zone, velocity point analysis zone, and relevant point association zone of the buoy, collectively describing the characteristics of the buoy in terms of spatial distribution and dynamic behavior. The system first divides the geographical area into several initial grid cells based on the data from the position point zone. Each grid cell represents a potential position zone, and its size can be adjusted according to the density of buoy distribution, using smaller grids in dense buoy areas and larger grids in sparse areas.

[0062] The system judges each initial zone and analyzes the number of positioning times and positioning times of the buoys in the zone. The number of positioning times refers to the number of times the satellite successfully identifies and records the buoy signal in a specific time period within the zone. The positioning time includes the specific time stamp of each record and the duration of the signal. These data are extracted from the satellite's original observation records and historical database. For example, for a grid cell, the system will count the number of times the cell is scanned by the satellite in the past week, as well as the number and time points of the captured buoy signals during each scan.

[0063] After obtaining the number of fixes and fix time data for each partition, the system performs a fitting operation. The fitting process employs a regression analysis method, taking the number of fixes and fix time as independent variables, and the actual geographic coordinates of the partition as dependent variables. Specifically, the system establishes a model that determines the geographic location of a partition in reality by analyzing the relationship between the high-frequency areas of historical fix counts in a partition and the known actual positions of buoys. At the same time, fix time data is used to analyze the timing patterns of the buoys, such as some buoys may only be frequently recorded during specific time periods (such as daytime or specific tidal stages), and these time patterns are incorporated into the fitting model to improve the accuracy of the mapping.

[0064] The fitting process produces a mathematical relationship that associates each location partition (represented by a grid cell) with an accurate actual geographic location. This mapping relationship can be represented as a function, with the input being the identifier of the partition and its statistical characteristics (such as the average number of fixes and the main fix time period), and the output being the latitude and longitude coordinates or a geographic range. For example, the system may determine that a certain partition number corresponds to XX degrees north XX minutes, XX degrees west XX minutes, with an error radius. The fitting process also considers the spatial relationship between location partitions. When mapping to actual geographic locations, adjacent partitions also maintain spatial continuity in their coordinate outputs, avoiding abrupt jumps. The system uses spatial interpolation algorithms to smooth the mapping results at the boundaries of the partitions, ensuring that the entire mapping relationship is geographically coherent.

[0065] After fitting is complete, each location partition has a mapping relationship to the actual location. This set of relationships is stored in a mapping table for subsequent steps. The mapping relationship is dynamic and will be periodically updated as new satellite data continues to flow in. The system monitors the fix data stream of each partition, and when it finds that the number of fixes or time patterns of a certain partition have changed significantly, it will trigger a re-fitting process to adjust its mapping relationship to adapt to changes in the buoy position or state. For example, if a buoy is moved to a new location, the fix data pattern of the partition it is in will change, and the system updates the actual location mapping of the partition through re-fitting, thereby maintaining the accuracy of the fix information.

[0066] The entire implementation process relies on the comprehensive analysis of historical data and real-time data, and the system not only processes current satellite scanning data, but also calls a large number of historical trajectory records to identify long-term stable patterns and short-term abnormal changes. In this way, the division of location partitions and the construction of mapping relationships not only reflect the immediate state of the buoy, but also incorporate the historical context of its behavior, forming a robust and adaptive spatial framework that lays the foundation for subsequent management and updating of fix information.

[0067] Example 3: Constructing the mapping relationship between feature information and location partitions is a crucial stage. This process begins by retrieving the location data and historical trajectories corresponding to the satellite data. The location data includes real-time acquired buoy coordinate information, which includes accuracy indicators such as latitude and longitude values ​​and their potential error ranges. The historical trajectories are records of the buoy's movement paths over a past period, retrieved from the database. These records are organized in a time series and include the location, speed, and possible additional attributes at each point in time.

[0068] Based on this data, the system generates multiple unlabeled location identification results. The generation process employs a pattern recognition algorithm, which analyzes the spatial and temporal patterns inherent in the location data and historical trajectories. The algorithm first performs spatial clustering on the location data, identifying areas where buoy signals frequently appear. Simultaneously, it analyzes the shape, direction, and speed changes of historical trajectories, extracting common movement patterns such as straight-line navigation, circling operations, or stationary mooring. Each identification result is a data object containing its corresponding geographical region, movement pattern characteristics, and time information, but these results are not yet associated with any known buoy identifiers or zone labels. The system determines whether these unlabeled location identification results are target location identification results. Target location identification results are those that highly match the existence and behavior of actual buoys. This determination is accomplished through a similarity calculation. The system maintains a benchmark dataset, which contains the precise locations of known buoys, their respective zone identifiers, and typical behavioral patterns. Each unlabeled identification result is compared with records in the benchmark dataset.

[0069] The calculation involves multiple dimensions. Spatial dimension comparison checks whether the geographic area of ​​the identified result overlaps with the geographic extent of a known partition, and whether its center coordinates are within an acceptable error range. Temporal dimension comparison analyzes whether the occurrence time pattern of the identified result matches the historical activity time of known buoys. Behavioral dimension comparison evaluates whether the extracted movement pattern (such as trajectory shape and velocity distribution) matches the typical behavior of known buoys. A function to quantify this degree of matching can be expressed as:

[0070] ;

[0071] in: Indicates the result of label-free recognition. With known partitions The overall matching degree between them. Represents spatial similarity, calculated by The region and The geographical extent is determined by the overlapping area or the distance between the center points. Represents time similarity, through comparison the activity time series of and the historical time pattern of represent the behavior similarity, which is calculated by analyzing the similarity between the moving trajectory features (such as direction, curvature, speed change) of and the typical trajectories of , , and are weight factors, used to adjust the relative importance of the three dimensions of space, time and behavior in the final matching degree, and their values are pre-set according to the reliability of different dimensions in the actual application scenario.

[0072] The system calculates the matching degree of each unlabeled recognition result with all known partitions. If there is a partition such that the value exceeds a pre-set threshold, then the recognition result is determined as the target positioning recognition result. Once it is determined as the target result, it is considered as an instance or representation of its matching partition , that is, it is "considered" as the location partition.

[0073] After successfully identifying the target positioning recognition result and associating it with the location partition, the system proceeds to build the final mapping relationship. This mapping relationship is a structured association set, whose content includes the specific representation of feature information, the description information of the location partition, and the type attribute of the partition. The representation of feature information comes from the output of Embodiment 1, i.e., the location point area (describing the reliable coordinate range), the speed point analysis area (describing the motion state feature), and the related point association area (describing the signal interaction intensity). The description information of the location partition includes the unique identifier of the partition, its geographical boundary, the estimated number of internal buoys, the average motion state, etc. The type of the location partition is a classification label, such as a fixed area, a moving area, or a mixed area, which is determined by the dominant behavior pattern of the target recognition results in the partition.

[0074] The process of building the mapping relationship is essentially to associate and bind each data item of the feature information with each attribute of the location partition. For example, the system will record that the pattern with a specific location point area range and speed point analysis area feature is consistently mapped to a location partition with a certain identifier and type. This mapping relationship can be stored and represented in the form of a database table, a matrix, or a configuration file, so that the system can quickly query and determine the most likely corresponding location partition according to the newly acquired feature information in subsequent processing. The entire process is iterative and self-learning, and new matching success cases are incorporated into the baseline dataset for optimization of future judgment processes, so that the mapping relationship becomes more accurate and reliable over time.

[0075] Example 4: Refer to Figure 4 The calculation of the combined matching degree of location data is the core step of constructing the location bias combination. The system receives the feature information and the divided location partition data from the previous examples. The feature information contains detailed parameters of the location point area, the speed point resolution area, and the related point association area, while the location partition contains the geographical range, type identification, and internal buoy statistical information of each partition.

[0076] The system first performs clustering analysis on these two sets of data, grouping them according to three types: data type, location type, and function type. Data type clustering classifies data according to their intrinsic attributes, such as classifying numerical data like latitude and longitude coordinates, speed values (knots), signal strength (dB), etc. Location type clustering is based on geographical characteristics, such as dividing partitions located in major shipping lanes, near ports, and open sea areas into different groups. Function type clustering is based on the purpose of the buoy, such as navigation buoys, environmental monitoring buoys, and mooring buoys. The clustering analysis uses an improved K-means algorithm that can handle both numerical and categorical data. After multiple iterations, the algorithm outputs several cluster centers, each representing a collection of data with similar characteristics. The system selects the cluster center with the most data points and identifies the data group it represents as the key partition. This key partition represents the most common and typical buoy characteristics and location patterns in the current data set.

[0077] The system extracts keywords from the key partition. These keywords are not text but quantitative indicators or feature vectors that can highly summarize the core features of the partition. For example, for a key partition identified as "nearshore navigation buoy", its keywords may include: average water depth, predominant signal frequency, average flow speed range, and the most common buoy material code. Each keyword is accompanied by its statistical distribution in the partition (such as mean, mode). The system uses a cosine similarity-based measurement method, but it is adapted to handle multi-dimensional mixed data types. It calculates the directional consistency between each pair of keyword vectors, resulting in a similarity score between 0 and 1. The higher the score, the higher the probability that the features represented by the two keywords coexist in the key partition.

[0078] Based on these pairwise similarity scores, the system constructs a common sequence. This common sequence is not a strict time sequence, but a sequence of feature appearance patterns. It describes which keywords tend to appear simultaneously, and the relative order in which they appear. For example, the analysis might show that the keyword "high signal strength" always appears with the keywords "shallow water" and "steel material", and that "shallow water" is usually the first feature observed. Using this common sequence, the system extracts from the raw data all position data that match this sequence pattern. Position data here refers to specific, individual buoy observation records, each containing its coordinates, timestamp, and observed feature values. The system filters out those observation records whose feature value change patterns highly match the common sequence.

[0079] The system sets the longest common subsequence between these position data. The concept of subsequence is here extended beyond strings and applied to the feature sequence of buoy observation records. The system compares any two observation records (e.g. record A and record B), and analyzes their feature change paths at a series of consecutive time points. The longest common subsequence is the longest, perfectly identical feature change fragment between the two records. The length of this fragment (i.e. the number of identical feature change steps it contains) is quantified as a numerical value, and defined as the combined match degree between the two position data. The higher the match degree, the more similar the behavior patterns of the two buoys in the past time period. Referring to Table 1, the system computes the combined match degree for all filtered position data pairwise, forming a match degree matrix.

[0080] Table 1: Position data combined match degree

[0081]

[0082] The system combines these combined match degrees with the time distribution probabilities of the position data to set the position bias combination. The time distribution probability describes the frequency at which each position data (i.e. each buoy) is observed in the historical records at different time periods (e.g. different hours of the day, different dates of the month). The system looks for groups of position data that have high combined match degrees (indicating similar behavior patterns) and highly overlapping time distribution probabilities (indicating frequent active observation at the same time periods). These groups are filtered out and defined as a position bias combination. This combination represents a set of buoys that have strong correlations in both behavior and time of appearance, providing a high-confidence data basis for subsequent positioning information updates. The whole computation process is dynamic and periodically re-executed as new data comes in, to ensure that the position bias combination reflects the current latest marine environment and buoy status.

[0083] Example 5: Updating positioning information in location partition begins with a deep analysis of the location-biased combination, which is the set of buoy data filtered out in the previous step that exhibit high correlation in behavior patterns and occurrence times. The system first resolves the temporal distribution probability of each piece of location data within the combination. The temporal distribution probability is statistically derived from historical satellite observation records, describing the frequency distribution of each buoy being successfully located at different time periods (e.g., each hour within a 24-hour cycle, each day within a week, or specific tidal phases). For example, a buoy might have a significantly higher probability of being observed between 8:00 AM and 10:00 AM than other time periods, and this pattern is quantified as a probability value and stored.

[0084] Based on these temporal distribution probabilities, the system sets a target path for the location-biased combination, which is not a physical trajectory but a serialized representation of the expected occurrence locations of the buoys within specific time periods. The system associates each time period (e.g., a 1-hour time window) with a spatial region where the location-biased combination is most likely to occur within that time period. For example, for the time period between 8:00 AM and 9:00 AM, the system calculates a geographic center coordinate and a distribution radius based on the historical location data of the buoys within the combination for that time period, forming a "target location point" for that time period. Connecting these target location points in chronological order results in a target path for the location-biased combination within a day or a cycle. This path reflects the expected spatial distribution patterns of the buoy group in the temporal dimension.

[0085] The system fits a target path to each piece of location data in the location-biased combination. Since the target paths of different buoys within the same combination might have slight differences (e.g., individual buoy deviations), the system uses curve fitting techniques to find a smooth path that best represents the overall movement trend of the group. The fitting process considers the geometric shape, direction change points, and velocity continuity of the paths of individual buoys. The fitted target path is a continuous spatial curve, with each point corresponding to an expected average location at a specific time point. By discretizing the fitted target path into individual time periods, the system calculates a probability value for the path in each time period, called the combination occurrence probability. This probability value depends on two factors: one is the degree of overlap between the geographic region covered by the fitted path in that time period and the distribution of historical observation points; the other is the proportion of the number of valid observations in that time period in the total number of observations in history. For example, if the region covered by the fitted path between 9:00 AM and 10:00 AM contains 90% of the historical observation points for that time period, and there are sufficient observation samples for that time period, the combination occurrence probability will be assigned a higher value.

[0086] The system then compares the combined occurrence probability of the location bias combination with the historical trajectory reference cases stored in the location partition. The historical trajectory reference cases record the frequency of the internal buoy trajectory data in the partition being actually recorded and used when the partition was scanned by the satellite in the past. For example, a certain partition has its internal buoy trajectory data called by the navigation system 100 times and referenced by the environmental monitoring report 50 times in the past 30 days. The comparison process calculates the difference value between the combined occurrence probability value and the historical reference frequency. The difference value identifies the deviation between the expected probability and the actual usage, which can be caused by buoy state changes, environmental interference, or data acquisition abnormalities.

[0087] According to the difference value, the system updates the positioning information in the location partition. The update is based on a rule-based decision tree model. The model sets a difference threshold: positioning information with a difference value below the first threshold is classified into a high confidence category, indicating that the expected probability is highly consistent with the actual usage, and the information is reliable; positioning information with a difference value between the first and second thresholds is classified into a medium confidence category, indicating that there is an acceptable deviation, and the information is basically reliable but needs attention; positioning information with a difference value above the second threshold is classified into a low confidence category, indicating that there is a significant inconsistency, and the information may have a large error or be outdated. The system assigns a corresponding confidence label to each positioning information and adjusts its weight and priority in the partition database accordingly. For example, low-confidence information will be marked and prioritized for re-verification or replacement in subsequent data collection.

[0088] After updating the positioning information, the system generates a buoy positioning management library according to the updated mapping relationship. First, the system extracts the positioning target from the latest satellite data stream, i.e., the buoy entity identified and positioned at the current time. Each positioning target is accompanied by its occurrence probability, which is composed of two parts: one is the combined occurrence probability of the location bias combination it belongs to; the other is the independent occurrence frequency of the buoy itself in history. The system combines these two probability values to calculate the final occurrence probability score through weighted calculation. The system sorts all positioning targets in descending order of occurrence probability score to obtain the positioning order. The sorting result reflects the location reliability and importance priority of the buoy at the current time. For example, a buoy with a high combined occurrence probability and a high historical occurrence frequency will be ranked at the top of the list.

[0089] The system combines the positioning targets and the positioning sequence in a structured database format. The positioning targets are stored as database records, each record contains the buoy unique identifier, real-time coordinates, confidence label, belonging partition, appearance probability score, etc. fields. The positioning sequence is embodied as a sorting index field in the database, or an independent priority list table. Together they form the buoy positioning management library. This library supports efficient query, for example, buoy information can be quickly retrieved by partition, confidence or priority, providing real-time, ordered buoy position data source for marine supervision, navigation safety or resource management. The whole process runs periodically in an automated manner, ensuring that the management library remains timely with satellite data updates.

[0090] It should be noted that the relational terms herein such as first and second and the like are used solely to distinguish one from another entity or action without necessarily requiring or implying that there is any such relationship or order between such entities or actions. Also, the terms "comprises", "comprising", or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0091] While embodiments of the application have been shown and described, it is to be understood that the embodiments described are only by way of example and that changes can be made to the embodiments without departing from the spirit and the scope of the application, which is defined in the appended claims and their equivalents.

Claims

1. A method for positioning a digital buoy based on satellite data, characterized by, The method comprises the following steps: acquiring satellite data and preprocessing the satellite data to extract feature information of the buoy; dividing the buoy position partition based on the feature information; constructing a mapping relationship between the feature information and the position partition; extracting key partitions and calculating the combined matching degree of the position data to obtain a position bias combination; updating the positioning information in the position partition based on the position bias combination; generating a buoy positioning management library according to the updated mapping relationship; calculating the combined matching degree of the position data comprises: performing cluster analysis on the feature information and the position partition according to data types, position types and function types, and setting the maximum cluster center after cluster analysis as the key partition; extracting the keywords of the key partition and calculating the similarity between each keyword, and setting the public sequence related to the similarity between each keyword; using the public sequence related to the similarity between each keyword to extract the position data existing in the public sequence, and setting the longest common subsequence between each position data as the combined matching degree between each position data; setting the position bias combination according to the time distribution probability of each position data.

2. The method of claim 1, wherein, extracting the feature information of the buoy comprises: acquiring a signal classification model for the buoy signal in the satellite data; classifying the satellite data using the signal classification model to obtain at least one signal category; identifying feature points, including position points, velocity points and related points, existing in the satellite data under the corresponding signal category to form a feature sample set; analyzing the position points, velocity points and related points in the feature sample set respectively, and sequentially obtaining the position point area, velocity point analysis area and related point association area corresponding to the feature sample set as the feature information.

3. The method of claim 2, wherein, acquiring the position point area, velocity point analysis area and related point association area corresponding to the feature sample set comprises: merging the position points, velocity points and related points in the feature sample set according to the signal category to obtain a plurality of merging results; extracting position point pairs in the merging results, comparing the position point pairs with a position database to obtain a position point area; extracting the change rate of the velocity points and the association strength of the related points, and dividing the merging results according to the change rate of the velocity points and the association strength of the related points to obtain a velocity point analysis area and a related point association area.

4. The satellite data based digital buoy positioning method of claim 1, wherein, dividing the buoy position partition comprises: judging the position partition, analyzing the positioning times and positioning times of the buoy in the position partition, fitting the position partition according to the positioning times and positioning times, and constructing a mapping relationship from the position partition to the actual position.

5. The satellite data based digital buoy positioning method of claim 1, wherein, constructing the mapping relationship between the feature information and the position partition comprises: calling the position data and historical trajectory corresponding to the satellite data to generate a plurality of unlabeled positioning recognition results; judging whether the plurality of unlabeled positioning recognition results are target positioning recognition results, if they are the target positioning recognition results, regarding the target positioning recognition results as the position partition.

6. The satellite data based digital buoy positioning method of claim 3, wherein, The mapping relationship between the feature information and the location partition is constructed by using information representing a location point area, a speed point analysis area and a related point association area in the feature information, description information of the location partition and a type of the location partition.

7. The satellite data based digital buoy positioning method of claim 1, wherein, Updating the positioning information in the location partition comprises: extracting a time distribution probability of each location data from the location bias combination; setting a target path of the location bias combination according to a time period corresponding to the time distribution probability of each location data; fitting the target path of each location data in the location bias combination to obtain a fitted target path, and setting a probability value of the fitted target path in each time period as a combination occurrence probability of the location bias combination; comparing the combination occurrence probability of the location bias combination with a reference condition of a historical trajectory in the location partition, identifying a difference value, and classifying the positioning information in the location partition according to the difference value to complete the updating.

8. The satellite data based digital buoy positioning method of claim 1, wherein, The processor, when executing the computer program, implements the steps of the digital buoy positioning method based on satellite data according to any one of claims 1 to 8. The processor, when executing the computer program, implements the steps of the digital buoy positioning method based on satellite data according to any one of claims 1 to 8. ​ 9. A digital buoy positioning system based on satellite data, comprising a memory, a processor and a computer program stored in the memory and running on the processor, characterized in that, ​

Images