A method and system for tracing a vessel involved in a collision of a light buoy

Abstract

The application provides a method and system for tracing a collision-causing ship through a light buoy, the method comprising: collecting a collision signal through an acoustic and vibration sensor arranged on the light buoy, determining a collision event and generating an accurate time stamp through time-frequency domain fusion processing and deep learning model identification; expanding a preset time window with the time stamp, obtaining GNSS positioning data of the light buoy and AIS data of surrounding ships in the period; calculating an actual distance between the collided light buoy and the passing ship according to the GNSS positioning data and the AIS data of the passing ship; determining whether the passing ship is suspicious according to a comparison result of the actual distance and a theoretical distance, and screening out a suspicious ship; and determining the collision-causing ship from the suspicious ship by detecting a change in the speed and the heading of the suspicious ship. The application can improve the efficiency of collision accident investigation and quickly and accurately determine the collision-causing ship.

Description

Technical Field

[0001] This application relates to the field of buoy collision tracing technology, specifically to a method and system for tracing the source of a collision involving a buoy and tracking the offending vessel. Background Technology

[0002] In the field of marine and port management, light buoys serve as important navigational markers, and their safety and stability are crucial for ship navigation. Currently, investigations into light buoy collisions primarily rely on on-site accident investigations and AIS (Automatic Identification System) trajectory playback. However, existing technologies have significant shortcomings and deficiencies in investigating light buoy collision accidents.

[0003] On the one hand, on-site accident investigations suffer from significant delays, difficulties in evidence collection, and cumbersome procedures such as paint testing or scratch comparison by the Maritime Safety Administration. On the other hand, although AIS trajectory playback is a fundamental method for tracing the offending vessel, relying solely on manual judgment based on changes in the vessel's movement before and after passing through the area of ​​the light buoy presents numerous problems. Manual judgment lacks objective standards, and different individuals may make differing assessments; there is no complete and systematic analytical method for processing and analyzing data, and relying solely on human experience makes it difficult to guarantee the accuracy and reliability of judgments; moreover, manually analyzing large amounts of AIS and light buoy data requires a significant investment of time and manpower, resulting in low efficiency and making it difficult to quickly and accurately identify the offending vessel.

[0004] While existing technologies utilize AIS data and buoy data, they fail to fully leverage the advantages of digitalization and lack in-depth mining and intelligent analysis of this data.

[0005] In view of this, the present invention proposes a method and system for tracing the source of a collision involving a light buoy and tracking the offending vessel, which can improve the efficiency of accident investigation, enhance the accuracy of accident judgment, and reduce labor costs. Summary of the Invention

[0006] To address the current problems of low investigation efficiency and difficulty in quickly and accurately identifying the offending vessel in light buoy collision accidents, this invention provides a method and system for tracing the source of a collided light buoy and tracking the offending vessel, thereby resolving the aforementioned technical deficiencies.

[0007] In a first aspect, the present invention proposes a method for tracing the source of a collision involving a light buoy and tracking the offending vessel, the method comprising the following steps:

[0008] S1. Collect and preprocess the ship's AIS data, which includes MMSI code, position, speed, heading, and ship dimensions. Collect underwater acoustic emission signals at the time of collision using a low-frequency acoustic sensor array deployed on the buoy, and simultaneously collect vibration and impact signals of the buoy structure using a broadband vibration sensor. Perform time-frequency domain fusion processing on the acoustic emission signals and vibration and impact signals, and input them into a pre-trained deep learning neural network model for identification. If the matching degree between the identified signal features and the ship collision features exceeds a preset threshold, it is determined that the buoy has collided, and a determination signal containing a collision timestamp is generated.

[0009] S2. Based on the collision timestamp of the judgment signal, extend a preset time window forward and backward, obtain the GNSS positioning data of the collided buoy within the preset time window, and filter the AIS data of passing ships with the collided buoy as the coordinate center and a set radius range r.

[0010] S3. Based on GNSS positioning data and AIS data of passing vessels, calculate the actual distance between the collided light buoy and the passing vessel; based on the comparison between the actual distance and the theoretical distance, determine whether the passing vessel is suspicious and screen out the suspicious vessels;

[0011] S4. By detecting changes in the speed and course of the suspected vessels, the vessel responsible for the accident is identified from among the suspected vessels. This includes the following sub-steps:

[0012] S41. Obtain the speed sequence and course sequence of the suspected vessel;

[0013] S42. Mark speed drop points in the speed sequence where the speed drop point decreases more than the previous data point by a factor exceeding a preset speed threshold; mark heading change points in the heading sequence where the heading angle changes more than the previous data point by a factor exceeding a preset angle threshold.

[0014] S43. A suspected vessel that meets all of the following conditions shall be identified as the vessel responsible for the accident:

[0015] There is a point where the speed drops suddenly and the timestamp of the drop is within the preset time window;

[0016] There is a course change point and the difference between the change timestamp and the descent timestamp is less than a preset time threshold.

[0017] Preferably, after step S4, the method further includes: S5, tracking and visually replaying the suspected vessel, specifically including the following sub-steps:

[0018] S51. Extract continuous time-series radar echo data of a preset area centered on the struck buoy within a preset time window;

[0019] S52. Connect the continuous radar echo points of the suspected vessel within a preset time window in sequence according to the outer edge of its echo end to generate a strip-shaped movement path of the vessel; connect the continuous radar echo points of the collided buoy within the same time period in sequence according to the outer edge of its echo end to generate a strip-shaped movement path of the buoy.

[0020] S53. In a high-resolution electronic nautical chart scenario, the generated ship's strip-shaped movement path and the light buoy's strip-shaped movement path are dynamically overlaid and displayed, with the ship's strip-shaped movement path and the light buoy's strip-shaped movement path being rendered in different colors.

[0021] S54. In dynamic overlay visualization, if the strip-shaped movement path area of ​​the light buoy is completely or partially inside the strip-shaped movement path area of ​​the vessel, it will help determine that the suspected vessel is the vessel that caused the accident.

[0022] S55. Match the ship target in the radar echo data with the AIS signal in real time. If the match is successful, obtain the ship's AIS static information. If no AIS signal is matched, extend the playback time window of the radar echo data and continue to execute steps S51 to S54 to track until the AIS signal is matched and the information is successfully obtained, or the relevant data of the ship's docking berth is recorded.

[0023] Preferably, the method for tracing and tracking the offending vessel after a collision with a light buoy provided by the present invention further includes: collecting light buoy monitoring data and preprocessing it, and further verifying whether a collision has occurred based on the preprocessed light buoy monitoring data, including the following sub-steps:

[0024] S21. Analyze the voltage value sequence in the preprocessed light buoy monitoring data;

[0025] If there is no valid voltage data in two consecutive feedback cycles in the voltage value sequence, and the voltage value recorded in the last time is greater than the preset voltage value, then the voltage fault flag is triggered.

[0026] S22. Extract GNSS positioning data from the preprocessed light buoy monitoring data and calculate the position offset between adjacent periods according to the time series.

[0027] If the cumulative offset distance within the preset time period exceeds the preset distance, a displacement fault flag will be triggered.

[0028] S23. If either the voltage fault flag or the displacement fault flag is triggered, the verification result is that the lamp buoy has collided.

[0029] Preferably, in step S3, the comparison between the actual distance and the theoretical distance is used to determine whether the passing vessel is suspicious, and suspicious vessels are screened out. This specifically includes the following sub-steps:

[0030] S31. Based on the ship's dimensions and the diameter of the collided buoy from the ship's AIS data, construct theoretical distances according to the collision location. These theoretical distances include the theoretical distance for mid-ship collisions. Theoretical distance for stern collision Theoretical distance to the bow collision The calculation expression is as follows:

[0031]

[0032]

[0033]

[0034] In the formula, Indicates the length of the ship that passed. Indicates the width of the passing vessel. Indicates the diameter of the buoy that was struck;

[0035] S32. Compare the actual distance with the three theoretical distances constructed in step S31 to determine whether the passing vessel is suspicious, and screen out suspicious vessels, including first-level suspicious vessels, second-level suspicious vessels, and third-level suspicious vessels:

[0036] If the actual distance is less than the theoretical distance for a mid-ship collision If so, the passing vessel is determined to be a Class I suspected vessel;

[0037] If the actual distance is within the theoretical distance for a stern collision Theoretical distance to collision with the ship If the vessel passes through the area between these points, it is determined to be a Class II suspected vessel.

[0038] If the actual distance is within the theoretical distance for a bow collision Theoretical distance to collision with the stern If the vessel passes through the area between these points, it is determined to be a Class III suspected vessel.

[0039] If the actual distance is greater than or equal to the theoretical distance of a bow collision If so, it is determined that the passing vessel is not suspected.

[0040] Preferably, in step S3, the actual distance between the struck buoy and the passing vessel is calculated based on GNSS positioning data and AIS data of the passing vessel, specifically including the following sub-steps:

[0041] S301. Obtain the GNSS positioning coordinates of the struck buoy and the AIS position coordinates of the passing vessel;

[0042] S302. Calculate the real-time physical distance between the GNSS positioning coordinates of the struck buoy and the AIS position coordinates of the passing ship using the Euclidean distance formula.

[0043] S303. Within a preset time period, select the minimum physical distance value from the real-time physical distances as the actual distance.

[0044] Preferably, in step S1, underwater acoustic emission signals at the time of collision are collected by a low-frequency acoustic sensor array deployed on the buoy, and vibration and impact signals of the buoy structure are collected by a broadband vibration sensor; the acoustic emission signals and vibration and impact signals are fused in the time and frequency domains and input into a pre-trained deep learning neural network model for identification; if the matching degree between the identified signal features and the ship collision features exceeds a preset threshold, it is determined that the buoy has collided, and a determination signal containing a collision timestamp is generated, specifically including the following sub-steps:

[0045] S11. Acquire underwater acoustic emission signals at the time of collision using a low-frequency acoustic sensor array at a first sampling frequency to obtain an acoustic emission signal time series; acquire vibration and impact signals of the buoy structure using a broadband vibration sensor at a second sampling frequency to obtain a vibration and impact signal time series; and use a GPS time synchronization module to timestamp-align the acoustic emission signal time series and the vibration and impact signal time series.

[0046] S12. Perform bandpass filtering on the acoustic emission signal time series to remove preset low-frequency noise and high-frequency interference, and obtain the filtered acoustic emission signal; perform high-pass filtering on the vibration and shock signal time series to remove environmental vibration baseline drift, and obtain the filtered vibration and shock signal; perform amplitude normalization processing on the filtered acoustic emission signal and vibration and shock signal respectively.

[0047] S13. Perform short-time Fourier transform on the preprocessed acoustic emission signal and vibration shock signal respectively to calculate the time spectrum of the acoustic emission signal and the time spectrum of the vibration shock signal; extract the Mel frequency cepstral coefficient feature vector from the time spectrum of the acoustic emission signal and the wavelet packet energy feature vector from the time spectrum of the vibration shock signal; perform feature-level fusion of the Mel frequency cepstral coefficient feature vector and the wavelet packet energy feature vector to generate a fused feature matrix;

[0048] S14. Input the fused feature matrix into the pre-trained deep learning neural network model. The deep learning neural network model is a hybrid model combining convolutional neural network and long short-term memory network. It is used to output the matching score between signal features and ship collision features. The matching score is calculated through a normalized exponential function layer.

[0049] S15. If the matching score exceeds the preset threshold, it is determined that the light buoy has collided, and a judgment signal containing the collision timestamp is generated.

[0050] Preferably, in step S1, the ship's AIS data is preprocessed, including:

[0051] The AIS data of ships are grouped according to the MMSI code to generate independent ship datasets;

[0052] Based on the UTC timestamp, sort the data and calculate the time interval between adjacent data points in the independent ship dataset. If the time interval exceeds a preset threshold, the data is divided into independent trajectory segments.

[0053] Preprocessed AIS data is obtained by correcting GPS drift errors in independent trajectory segments using linear interpolation.

[0054] Secondly, the present invention proposes a system for tracing and tracking the offending vessel after a buoy collision, used to perform the method for tracing and tracking the offending vessel after a buoy collision as described above, the system comprising:

[0055] The data acquisition and preprocessing module is configured to acquire and preprocess the ship's AIS data, which includes MMSI code, position, speed, heading, and ship dimensions. It acquires underwater acoustic emission signals at the time of a collision using a low-frequency acoustic sensor array deployed on the buoy, and simultaneously acquires vibration and impact signals of the buoy structure using a broadband vibration sensor. The acoustic emission and vibration / impact signals are fused in the time and frequency domains and input into a pre-trained deep learning neural network model for identification. If the matching degree between the identified signal features and the ship's collision features exceeds a preset threshold, a collision is determined, and a determination signal containing a collision timestamp is generated.

[0056] The collision determination module is configured to use the collision timestamp of the determination signal as a reference, extend forward and backward by a preset time window, obtain the GNSS positioning data of the collided buoy within the preset time window, and filter the AIS data of passing ships with the collided buoy as the coordinate center and a set radius range r.

[0057] The suspect analysis module is configured to calculate the actual distance between the collided buoy and the passing vessel based on the GNSS positioning data of the collided buoy and the AIS data of the passing vessel; and to determine whether the passing vessel is suspicious based on the comparison between the actual distance and the theoretical distance, and to filter out the suspicious vessels.

[0058] The accident determination module is configured to identify the vessel responsible for the accident from among the suspected vessels by detecting changes in the speed and course of the suspected vessels. Specifically, it includes the following sub-steps:

[0059] S41. Obtain the speed sequence and course sequence of the suspected vessel;

[0060] S42. Mark speed drop points in the speed sequence where the speed drop point decreases more than the previous data point by a factor exceeding a preset speed threshold; mark heading change points in the heading sequence where the heading angle changes more than the previous data point by a factor exceeding a preset angle threshold.

[0061] S43. A suspected vessel that meets all of the following conditions shall be identified as the vessel responsible for the accident:

[0062] There is a point where the speed drops suddenly and the timestamp of the drop is within the preset time window;

[0063] There is a course change point and the difference between the change timestamp and the descent timestamp is less than a preset time threshold.

[0064] Thirdly, the present invention proposes a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of any of the above-described methods for tracing the source of a collision between a light buoy and the offending vessel.

[0065] Fourthly, the present invention proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of any of the above-described methods for tracing the source of a collision involving a light buoy and tracking the offending vessel.

[0066] Compared with the prior art, the beneficial results of the present invention are as follows:

[0067] (1) Improved efficiency of accident investigation: This invention changes the traditional model that relies on manual judgment through a systematic data processing and analysis method, effectively reducing the tedious steps and subjectivity of manual operation. It can quickly filter out key information from a large amount of data, greatly improving the efficiency of collision accident investigation and realizing the rapid and accurate identification of the offending vessel.

[0068] (2) Improve the accuracy of accident judgment: This invention performs in-depth mining and intelligent analysis on the collected AIS data and light buoy data. By constructing a theoretical distance model and comparing it with the actual distance, the suspicion level of the offending vessel is classified, and a judgment is made by combining multiple factors, thereby improving the accuracy and reliability of judging ship collision accidents and effectively reducing the possibility of misjudgment.

[0069] (3) Reduce labor costs: Traditional methods require a lot of manpower to screen and analyze data, while the present invention reduces the workload of manual analysis through automated and intelligent processing, reduces the input of labor costs, and improves work efficiency, enabling maritime management departments to carry out accident investigation and handling work more efficiently.

[0070] (4) Enhance accident prevention capabilities: By analyzing the drift motion characteristics of the light buoy and identifying abnormal data, the present invention can promptly detect potential risks of the light buoy and take preventive and handling measures in advance, thereby enhancing accident prevention capabilities, better ensuring the safety and stability of the light buoy, and reducing the probability of collision accidents.

[0071] (5) Supports multi-dimensional data fusion and intelligent analysis: This invention innovatively integrates light buoy monitoring data and ship AIS data, and uses various algorithms such as Kalman filtering and linear interpolation for data preprocessing, effectively improving data quality and reliability. At the same time, it introduces a theoretical distance model of the collision site and detection methods for changes in speed and heading, analyzing ship behavior from multiple dimensions to more accurately identify the offending ship.

[0072] (6) Excellent scalability and adaptability: The system design of this invention is reasonable, not only supporting software upgrades and optimizations to improve functionality and performance, but also having wide applicability, capable of meeting the needs of different aquatic environments and light buoy types. By dynamically adjusting the threshold parameters, it can adapt to various complex actual situations, further improving the accuracy and effectiveness of the system. Attached Figure Description

[0073] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments, taken with reference to the accompanying drawings:

[0074] Figure 1 This is a flowchart of the method for tracing and tracking the offending vessel after a buoy collision according to the present invention;

[0075] Figure 2 This is a flowchart for judging collision accident data according to the present invention;

[0076] Figure 3a The theoretical distance of mid-ship collision according to the present invention A schematic diagram of the calculation;

[0077] Figure 3b The theoretical distance of stern collision according to the present invention A schematic diagram of the calculation;

[0078] Figure 3c This is the theoretical distance of bow collision according to the present invention. A schematic diagram of the calculation;

[0079] Figure 4 This is a flowchart for tracking the offending vessel according to the present invention;

[0080] Figure 5 A dynamic overlay visualization of the present invention is shown;

[0081] Figure 6 A structural diagram of the collision tracing system for the light buoy of the present invention is shown;

[0082] Figure 7 This is a schematic diagram of the structure of a computer system suitable for implementing the embodiments of the present invention. Detailed Implementation

[0083] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0084] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0085] Firstly, this invention proposes a method for tracing the source of a collision involving a light buoy and tracking the offending vessel. Figure 1 A flowchart of the method for tracing and tracking the offending vessel after a collision with a buoy, as shown in this invention, is provided. Please refer to it. Figure 1 The method includes the following steps:

[0086] S1. Collect and preprocess the vessel's AIS data. AIS data (Automatic Identification System) includes: MMSI code (Maritime Mobile Service Identity), location, speed, heading, and vessel dimensions. Underwater acoustic emission signals at the time of collision are collected using a low-frequency acoustic sensor array deployed on the buoy, while vibration and impact signals of the buoy structure are collected using a broadband vibration sensor. The acoustic emission signals and vibration and impact signals are fused in the time and frequency domains and input into a pre-trained deep learning neural network model for identification. If the matching degree between the identified signal features and the vessel collision features exceeds a preset threshold, a collision is determined, and a determination signal containing a collision timestamp is generated. The specific implementation process is as follows:

[0087] S11. Signal Acquisition and Synchronization:

[0088] When a vessel comes into contact with or comes into very close proximity to the light buoy, specific physical signals are generated. A low-frequency acoustic sensor array synchronously acquires underwater acoustic emission signals at a first sampling frequency (preferably 96 kHz in this embodiment), generating an acoustic emission signal time series S_audio(t). A broadband vibration sensor acquires vibration and impact signals of the light buoy structure at a second sampling frequency (preferably 48 kHz in this embodiment), generating a vibration and impact signal time series S_vibration(t). The GPS module in the central signal processing unit provides UTC timestamps with microsecond accuracy and assigns a unified time tag to the data from both signal channels, ensuring that the signals processed subsequently are completely synchronized in time and eliminating analysis errors caused by sampling asynchrony.

[0089] S12, Signal Preprocessing:

[0090] The acquired raw signals contain a significant amount of environmental noise, requiring filtering and standardization. For S_audio(t), a bandpass filter with a passband frequency of 100Hz to 20kHz is used to remove low-frequency water flow noise (such as wave impact) and high-frequency electronic interference, while retaining the mid-to-low-frequency acoustic characteristics that may be caused by collisions, resulting in the filtered acoustic emission signal S_audio_f(t). For S_vibration(t), a high-pass filter with a cutoff frequency of 5Hz is used to remove low-frequency baseline drift caused by ocean currents, natural buoy swaying, etc., highlighting the transient impact components caused by collisions, resulting in the filtered vibration impact signal S_vibration_f(t).

[0091] To prevent amplitude differences from affecting model training and recognition, amplitude normalization is performed on S_audio_f(t) and S_vibration_f(t) respectively, and their amplitude range is linearly scaled to the interval [-1, 1] to obtain the normalized signals S_audio_n(t) and S_vibration_n(t).

[0092] S13. Time-frequency domain feature extraction and fusion:

[0093] To fully extract the features of non-stationary collision signals, the signals are transformed to the time-frequency domain and feature fusion is performed. Short-time Fourier transforms (STFTs) are performed on S_audio_n(t) and S_vibration_n(t) respectively, with a Hanning window function, a window length of 1024 sampling points, and an overlap rate of 50%. The time-frequency spectrograms of the acoustic emission signal (Spectrogram_audio) and the vibration and impact signal (Spectrogram_vibration) are calculated respectively.

[0094] Feature extraction: A 39-dimensional Mel-frequency cepstral coefficient (MFCC) feature vector, Feature_MFCC, is extracted from Spectrogram_audio. This feature can well simulate the characteristics of human hearing and has a strong ability to characterize the low-frequency resonance characteristics of impact acoustics. An energy feature vector, Feature_WaveletEnergy, based on 3-level wavelet packet decomposition, is extracted from Spectrogram_vibration. This feature can finely characterize the energy distribution of vibration and impact signals in each frequency band.

[0095] Feature fusion: The two feature vectors, Feature_MFCC and Feature_WaveletEnergy, are concatenated along their feature dimensions to generate a fusion feature matrix, Fusion_Matrix, that integrates acoustic and vibration information. This feature-level fusion strategy comprehensively utilizes the complementary information of the two signals, significantly improving the robustness and accuracy of subsequent pattern recognition.

[0096] S14, Deep Learning Model Recognition:

[0097] The fused feature matrix is ​​input into a pre-trained deep learning neural network model for recognition. The training and application of this model are as follows:

[0098] Model Structure: This embodiment uses a hybrid model combining a Convolutional Neural Network (CNN) and a Long Short-Term Memory (LSTM) network. First, the CNN layer (containing two convolutional layers and a pooling layer) automatically extracts local spatiotemporal features from the Fusion_Matrix. Then, the feature sequence output by the CNN is fed into the LSTM layer (containing a single 32-unit LSTM layer) to capture the temporal dependencies between features. Finally, a matching score P_collision is output through fully connected layers and a Softmax layer.

[0099] Model Training: The model is trained offline using a large amount of historical data (including confirmed collision event data and a large amount of non-collision environmental noise data). During training, the Adam optimizer is used to minimize the cross-entropy loss function until the model converges.

[0100] Online identification: The real-time generated Fusion_Matrix is ​​input into the deployed model, and the model outputs a matching score P_collision between [0, 1], which represents the probability that the current signal belongs to a ship collision event.

[0101] S15, Collision Detection:

[0102] The matching score P_collision output by the model is compared with a preset threshold (in this embodiment, the threshold is set to 0.85 after extensive testing and verification).

[0103] If P_collision ≥ 0.85, a collision is determined to have occurred on the buoy. The central signal processing unit then generates a determination signal containing a precise collision timestamp. This timestamp is calculated by adding the precisely calibrated signal processing delay T_delay (including filtering, transformation, feature extraction, and model forward propagation time) to the signal acquisition start time T_start, i.e., T_collision = T_start + T_delay.

[0104] Ultimately, the generated judgment signal (containing information such as T_collision and P_collision) is sent to the remote monitoring center through the wireless communication module, providing the most critical triggering basis and accurate time reference for subsequent ship tracing and tracking.

[0105] Through the above embodiments, the present invention achieves high-precision and automated identification of buoy collision events, effectively reducing false alarms and false negatives, and providing a reliable data foundation for subsequent steps.

[0106] Step S1 further includes: collecting and preprocessing light buoy monitoring data, which includes GNSS positioning data (Global Navigation Satellite System), voltage value sequences, and drift status. The drift status is a Boolean flag indicating whether a displacement threshold has been exceeded.

[0107] In practice, data is collected from multiple information sources at regular intervals according to a pre-set collection cycle. These sources include BeiDou links, GPRS (General Packet Radio Service), AIS links, and VTS (Vessel Traffic Service) links, covering various data types such as light buoy monitoring data and AIS data.

[0108] In step S1, the preprocessing of the light buoy monitoring data is as follows:

[0109] The GNSS positioning data of the light buoys is processed using a Kalman filter algorithm to eliminate drift errors. Simultaneously, null values ​​and redundant fields (such as repeatedly recorded non-critical parameters like ambient temperature and wind speed) are filtered out, while core parameters such as position, voltage, and attitude are retained, ultimately yielding pre-processed light buoy monitoring data.

[0110] In step S1, the ship's AIS data is preprocessed, including:

[0111] First, the raw AIS data is grouped according to the MMSI code to separate different ship datasets. Then, for the AIS data of the same ship, the time interval between adjacent data points is calculated based on the UTC timestamp (Coordinated Universal Time). If the interval exceeds a preset threshold, it is determined as a signal interruption, and the data is segmented into independent trajectory segments. Finally, using the light buoy coordinates as the center, a radius range is set to filter out ship trajectories passing through this area, reducing the amount of invalid data calculated. At the same time, interpolation is used to correct GPS (Global Positioning System) drift errors in the AIS data to ensure trajectory continuity, thus obtaining preprocessed AIS data.

[0112] In the data preprocessing stage, this invention employs multi-source data fusion calibration technology to improve trajectory accuracy. Specifically, this includes: establishing a ship kinematic model for AIS trajectory data; fusing GNSS positioning and gyroscope angular velocity data through extended Kalman filtering (EKF); defining the state vector as X=[x, y, v, θ, ω]^T (where x and y represent position, v represents velocity, θ represents heading angle, ω represents angular velocity, and T represents acceleration constraint threshold); constructing the observation equation based on the WGS84 coordinate system and the UTM (Universal Transverse Mercator) local coordinate transformation matrix; for trajectory segments with signal interruption, using an adaptive time window cubic spline interpolation method; setting acceleration constraints based on ship tonnage (radial acceleration threshold ≤0.3m / s² for ships of 10,000 tons); and ensuring the interpolation point density is positively correlated with the original data acquisition frequency; simultaneously introducing a differential correction mechanism to eliminate ionospheric errors using GNSS data from the base station, reducing positioning errors from the conventional 10-meter level to within 2 meters, providing a high-precision data foundation for subsequent distance calculations. This preprocessing process significantly improves the spatial alignment accuracy of the ship and light buoy trajectories through a triple guarantee of kinematic modeling, physical constraint interpolation, and differential correction.

[0113] Continue to refer to Figure 1 The present invention proposes a method for tracing the source of a collision involving a light buoy and tracking the offending vessel, which further includes the following steps:

[0114] S2. Based on the collision timestamp of the judgment signal, extend forward and backward by a preset time window, obtain the GNSS positioning data of the collided buoy within the preset time window, and filter out the AIS data of passing ships with the collided buoy as the coordinate center and a set radius range r.

[0115] For example, using the collision timestamp T_collision as a baseline, a time interval Δt is extended forward (in the historical direction) and backward (in the future direction) to construct a preset time window [T_collision - Δt, T_collision + Δt]. The value of Δt is determined by comprehensively considering the ship's speed, the frequency of AIS data updates, and the ship's possible reaction time after the collision. Based on the analysis of a large amount of inland waterway and port channel ship navigation data, this embodiment preferably sets Δt to 10 minutes. This means that all relevant data within a 20-minute time window, from 10 minutes before the collision to 10 minutes after the collision, will be retrieved. This time window is sufficient to cover the entire process of a ship traveling at 12 knots entering the monitoring range before the collision, the collision occurring, and its initial departure after the collision.

[0116] Step S2 also includes further determining whether a collision has occurred based on the preprocessed buoy monitoring data, including the following sub-steps:

[0117] S21. Analyze the voltage value sequence in the preprocessed light buoy monitoring data;

[0118] If there is no valid voltage data in two consecutive feedback cycles in the voltage value sequence, and the voltage value recorded in the last time is greater than the preset voltage value, then the voltage fault flag is triggered.

[0119] S22. Extract GNSS positioning data from the preprocessed light buoy monitoring data and calculate the position offset between adjacent periods according to the time series.

[0120] If the cumulative offset distance within the preset time period exceeds the preset distance, a displacement fault flag will be triggered.

[0121] S23. If either the voltage fault sign or the displacement fault sign is triggered, it is determined that the light buoy has been hit.

[0122] Steps S21-S23 can be used to further verify and confirm whether a collision actually occurred after step S1 determines that a collision has occurred based on the matching degree between signal features and ship collision features. Of course, in practical applications, steps S21-S23 can also be used as an alternative to step S1, which determines whether a collision has occurred based on the matching degree between signal features and ship collision features, that is, the collision accident is judged directly based on steps S21-S23.

[0123] In the collision determination step S2, the present invention further adopts a dynamic threshold adaptive mechanism to improve detection accuracy. Specifically, it includes: establishing a normal working mode baseline based on historical monitoring data of the light buoy, including the average data transmission cycle μ_t, the standard deviation of drift distance σ_d, and the voltage fluctuation range; in real-time monitoring, the voltage threshold is dynamically adjusted according to V_threshold = 11V + 0.5σ_v (σ_v is the historical voltage standard deviation), and the displacement threshold is dynamically corrected according to the sea area current velocity coefficient k to D_threshold = 15m × (1 + k·v_flow) (v_flow is the real-time current velocity data); at the same time, the speed change threshold Δv is set according to the ship tonnage classification (e.g., Δv=3 knots for tonnage <1000 tons, Δv=4.5 knots for 1000-5000 tons, and Δv=5.5 knots for >5000 tons), and the moving average of the heading angle change rate is calculated through a sliding time window. If the change rate exceeds twice the baseline value within three consecutive cycles, the heading anomaly flag is triggered. This multidimensional dynamic threshold system effectively suppresses false alarms and ensures the spatiotemporal adaptability of collision criteria by integrating historical data statistical features, real-time environmental parameters, and ship characteristics.

[0124] Figure 2 The flowchart of the collision accident data judgment of the present invention is shown, as follows: Figure 2 As shown, in practical applications, light buoys may suffer varying degrees of damage after being struck by a ship. According to the classification standards of the navigation aid management department, collision accidents can be categorized into five types: The first type involves the light buoy completely sinking and being destroyed, with all its onboard electronic equipment damaged, resulting in the navigation aid telemetry and control platform being unable to receive any telemetry signals. The second type involves damage to the visible structural components of the light buoy (such as the lookout board and light frame), but the main body of the light buoy remains in its original position and the electronic equipment's communication function is normal. In this case, the combined effect of the impact force generated by the ship's collision, the anchor chain, and the pull of the sinker will cause the navigation aid telemetry and control platform to record a brief displacement of the light buoy. The third type involves damage to the light buoy's above-water structure, causing it to move after the collision, while the electronic equipment's communication fails, resulting in the navigation aid telemetry and control platform not receiving the light buoy's telemetry signals. The fourth type involves the light buoy's electronic equipment communication function remaining intact, but due to the excessive impact force generated by the ship's collision, exceeding the anchor chain's bearing limit and causing it to break, the light buoy loses its restraint and drifts away from its original position. At this time, the navigation beacon telemetry and remote control platform will show that the distance of the light buoy deflection increases instantly after the collision and continues to increase over time; the fifth type is that after the collision, the light buoy not only experiences electronic communication failure, but also drifts away from its original position due to the impact force, causing the navigation beacon telemetry and remote control platform to be unable to obtain its telemetry signal.

[0125] To more accurately determine whether an accident has occurred and its specific outcome, two key types of information are primarily relied upon: First, the data transmission status of the buoy is assessed. If the buoy's normally functioning data transmission is suddenly interrupted, this likely indicates that the buoy has been struck by a vessel. Second, the buoy's drift is considered. If the buoy shows a momentary displacement, it usually means that it has been collided with by a passing vessel. However, when the buoy stops transmitting data, its positioning information cannot be obtained, making it difficult to determine whether the buoy has drifted away.

[0126] Therefore, the first, third, and fifth scenarios in the original accident classification were grouped into the same category. Ultimately, the buoy data was reclassified into three categories: the first category is no data transmission, encompassing the first (buoy completely destroyed and equipment total loss), the third (damaged and displaced surface structure, equipment communication failure), and the fifth (equipment communication failure and buoy drifting away) scenarios in the original classification; the second category is where the buoy can transmit data but has actually drifted away from its original position, corresponding to the fourth scenario in the original classification; the third category is where the buoy not only transmits data normally but also remains in its original position without drifting away, consistent with the second scenario in the original classification.

[0127] The primary channel for collecting light buoy movement data is the navigation buoy telemetry and control platform. By constructing a navigation buoy telemetry and control database, real-time monitoring of light buoy data based on public networks and the BeiDou communication network can be achieved, thereby improving the level of light buoy maintenance and management efficiency. Light buoy data encompasses both static and dynamic information. Static information includes light buoy parameters and thresholds preset by management personnel, while dynamic information involves the light buoy's real-time coordinates, displacement distance, and operating voltage. The navigation buoy telemetry and control platform has a light buoy fault alarm function; however, due to fluctuations in equipment stability, signal transmission obstruction, and inappropriate threshold settings, site navigation buoy management personnel may receive a large number of fault alarm messages daily. If management personnel lack experience and are unable to quickly identify light buoy collision accidents, the golden opportunity to track down the offending vessel may be missed.

[0128] Based on the classification of collision results between ships and light buoys, such collision events exhibit three typical characteristics on the navigational aid telemetry and control platform: no data feedback from the signal terminal, continuous increase in light buoy offset distance, and a sudden surge in light buoy offset distance. Based on these characteristics, a light buoy collision data anomaly identification process was developed in practical work to accurately filter out data related to light buoy collisions from massive amounts of light buoy fault alarm information.

[0129] When fault reports of no signal, displacement, or continuous displacement of the buoy are received, these fault data will be analyzed and classified. The specific judgment rules are as follows:

[0130] (1) If the real-time position of the light buoy exceeds the preset safe distance range and the offset state continues, it can be judged that the light buoy is suspected of being collided with by a ship.

[0131] (2) When the real-time position of the light buoy changes and its displacement distance exceeds 15 meters within a 20-minute time span, it can be determined that the light buoy is likely to be hit by a collision.

[0132] (3) If the light buoy fails to send a signal to the platform for two consecutive data transmission cycles, and the voltage value of the last telemetry data is greater than 11V, it can also be listed as a suspected collision object. If the voltage of the last telemetry data is less than 11V, the signal interruption should be tentatively attributed to insufficient energy supply. Further confirmation is needed based on the on-site investigation to determine whether the light buoy signal interruption is due to energy failure or a ship collision.

[0133] It should be understood that the above three standards for detecting abnormal collisions of light buoys are not fixed. Due to the differences in actual operational data among different aquatic environments and light buoy types, it is necessary to dynamically optimize and adjust the threshold parameters in each judgment condition based on accurate data obtained from real-time monitoring of the light buoys, in order to ensure the accuracy and effectiveness of the detection mechanism.

[0134] Continue to refer to Figure 1 The present invention proposes a method for tracing the source of a collision involving a light buoy and tracking the offending vessel, which further includes the following steps:

[0135] S3. Calculate the actual distance between the collided buoy and the passing vessel based on GNSS positioning data and AIS data of the passing vessel; determine whether the passing vessel is suspicious based on the comparison between the actual distance and the theoretical distance, and screen out the suspicious vessels.

[0136] In step S3, based on GNSS positioning data and AIS data from the passing vessel, the actual distance between the struck buoy and the passing vessel is calculated, specifically including the following sub-steps:

[0137] S301. Obtain the GNSS positioning coordinates of the struck buoy and the AIS position coordinates of the passing vessel;

[0138] S302. Calculate the real-time physical distance between the GNSS positioning coordinates of the struck buoy and the AIS position coordinates of the passing ship using the Euclidean distance formula.

[0139] S303. Within a preset time period, select the minimum physical distance value from the real-time physical distances as the actual distance for subsequent comparison and analysis with the theoretical distance.

[0140] In step S3, the comparison between the actual distance and the theoretical distance determines whether the passing vessel is suspicious, and suspicious vessels are screened out. The specific process is as follows:

[0141] S31. Based on the ship's dimensions and the diameter of the collided buoy from the ship's AIS data, and based on the geometric relationships between different parts of the ship and the buoy, three theoretical distances are constructed according to the collision location: the theoretical distance of mid-ship collision. Theoretical distance for stern collision Theoretical distance to the bow collision .

[0142] Figure 3a The theoretical distance for mid-ship collision is shown. Calculation diagram, Figure 3b The theoretical distance for stern collision is shown. Calculation diagram, Figure 3c The theoretical distance of bow collision is shown. The diagram illustrates the calculation process, where the red circle represents the GNSS antenna and the green circle represents the buoy. (Example:) Figures 3a-3c As shown, calculating these three theoretical distances requires considering the length (L), width (B), and diameter (R) of the buoy being struck. Depending on the different collision methods between the ship and the buoy, the corresponding geometric model and Pythagorean theorem are used to determine the ship / buoy GPS distance (S). Typically, the theoretical distance for a bow collision is... Maximum, theoretical distance of stern collision Secondly, the theoretical distance of a collision within the ship. Minimum. The specific calculation expression is as follows:

[0143]

[0144]

[0145]

[0146] In the formula, Indicates the length of the ship that passed. Indicates the width of the passing vessel. Indicates the diameter of the buoy that was struck;

[0147] S32. Compare the actual distance with the three theoretical distances constructed in step S31 to determine whether the passing vessel is suspicious, and screen out suspicious vessels, including first-level suspicious vessels, second-level suspicious vessels, and third-level suspicious vessels:

[0148] like If so, the passing vessel is determined to be a Class I suspected vessel;

[0149] like If so, the passing vessel is determined to be a Class II suspected vessel;

[0150] like If the vessel passes through the area between these points, it is determined to be a Class III suspected vessel.

[0151] like If so, it is determined that the passing vessel is not suspected, among which S 实 Indicates the actual distance.

[0152] Figure 4 A flowchart of the accident vessel tracking process of the present invention is shown, in conjunction with reference. Figure 1 and Figure 4 The present invention proposes a method for tracing the source of a collision involving a light buoy and tracking the offending vessel, which further includes the following steps:

[0153] S4. By detecting changes in the speed and course of the suspected vessels, the vessel responsible for the accident can be identified from among the suspected vessels.

[0154] In step S4, based on the suspected vessel's speed and heading changes, combined with the typical motion characteristics after a vessel collides with a light buoy, the system can accurately identify the offending vessel. After a vessel collides with a light buoy, it typically exhibits the following characteristics: the offending vessel's AIS navigation trajectory has a significant spatial correlation with the trajectory recorded by the buoy's telemetry and remote control data; at the moment of collision, the vessel's speed may experience irregular changes such as sudden increases, sudden decreases, or continuous fluctuations; the vessel's heading angle may suddenly deviate, deviating from its original navigation path.

[0155] To accurately capture these characteristics, the speed and heading sequences of the suspected vessel are first obtained. The system detects points in the speed sequence where speed drops sharply (i.e., points where the speed decreases by more than a preset speed threshold compared to the preceding data point at an adjacent time stamp). Simultaneously, it detects points in the heading sequence where the heading angle changes by more than a preset angle threshold compared to the preceding data point at an adjacent time stamp. The system identifies a suspected vessel as the vessel responsible for the incident if it simultaneously meets the following conditions: a speed drop point exists and the drop point time stamp is within a preset time window; a heading change point exists and the difference between the change point time stamp and the speed drop time stamp is less than a preset time threshold.

[0156] Following step S4, the process also includes: S5, tracking and visually replaying the suspected vessel, which specifically includes the following sub-steps:

[0157] S51. Extract continuous time-series radar echo data of a preset area centered on the struck buoy within a preset time window;

[0158] S52. Connect the continuous radar echo points of the suspected vessel within a preset time window in sequence according to the outer edge of its echo end to generate a strip-shaped movement path of the vessel; connect the continuous radar echo points of the collided buoy within the same time period in sequence according to the outer edge of its echo end to generate a strip-shaped movement path of the buoy.

[0159] S53. In a high-resolution electronic nautical chart scenario, the generated ship's strip-shaped movement path and the light buoy's strip-shaped movement path are dynamically overlaid and displayed, with the ship's strip-shaped movement path and the light buoy's strip-shaped movement path being rendered in different colors.

[0160] S54. In dynamic overlay visualization, if the strip-shaped movement path area of ​​the light buoy is completely or partially inside the strip-shaped movement path area of ​​the vessel, it will help determine that the suspected vessel is the vessel that caused the accident.

[0161] S55. Match the ship target in the radar echo data with the AIS signal in real time. If the match is successful, obtain the ship's AIS static information. If no AIS signal is matched, extend the playback time window of the radar echo data and continue to execute steps S51 to S54 to track until the AIS signal is matched and the information is successfully obtained, or the relevant data of the ship's docking berth is recorded.

[0162] Figure 5 A dynamic overlay visualization of the present invention is shown, such as... Figure 5 As shown, the green strip path represents the continuous radar echo trajectory of the suspected vessel between 03:53:40 and 03:54:00. The outer edges of the echo ends are connected sequentially to form a vessel-shaped movement path whose width varies with the vessel's attitude. The red strip path represents the radar echo trajectory of the collided buoy during the same time period, also connected to form a buoy-shaped movement path based on the outer edges of the echo ends. In the high-resolution electronic chart scene, the two strip paths are dynamically superimposed with different colors, and the visualization interface renders them in real time: when the red buoy strip area falls completely or partially into the green vessel strip area, an automatic "suspected collision" prompt is displayed, and the timestamp and relative positional relationship at that moment are recorded, thus completing the collision assistance determination intuitively and quickly.

[0163] Meanwhile, utilizing the AIS-radar echo matching mechanism in step S5: firstly, the current radar target is compared with the AIS signal in real time. If a match is successful, the vessel's MMSI, length, beam, and other static information are immediately retrieved to form a complete chain of evidence. If a match fails, the radar echo playback window is automatically extended, and the strip path extraction and overlay verification in steps S51-S54 continue until an AIS signal is successfully matched or the vessel's final docking history is traced, recording its latitude, longitude, and time for subsequent tracing. Through the dual strategy of "dynamic strip path overlay + continuous AIS matching," the shortcomings of traditional ARPA playback—requiring manual dragging of the timeline, lack of attitude information, and unintuitive echo overlap judgment—are effectively overcome, achieving efficient, accurate, and fully traceable visual tracing of light buoy collision events.

[0164] By monitoring abnormal vessel movements in the waters surrounding the buoy in real time, the system comprehensively assesses the likelihood of a collision. Furthermore, by replaying the AIS tracks of surrounding vessels and the buoy's movement data, it determines whether the track spacing meets the criteria for severity assessment. If the criteria are met, the system tracks the offending vessel based on the spacing, speed changes, and changes in bow direction, matching the suspect vessel's static AIS data and tracing its complete navigation trajectory. Simultaneously, the system supports replaying the offending vessel's AIS track along a timeline, dynamically displaying key parameters such as position, speed, and heading, and automatically marking abnormal behavior points to assist investigators in visually reconstructing the accident process and identifying suspects.

[0165] Further reference Figure 6 As an implementation of the above method, in a second aspect, the present invention provides an embodiment of a structural diagram of a buoy collision tracing and tracking system 600 for tracing the source of an accident involving a vessel. This system can be specifically applied to various electronic devices. The buoy collision tracing and tracking system 600 includes the following modules:

[0166] The data acquisition and preprocessing module 610 is configured to acquire and preprocess the AIS data of the vessel. The AIS data includes: MMSI code, position, speed, heading, and vessel dimensions. It acquires underwater acoustic emission signals at the time of a collision using a low-frequency acoustic sensor array deployed on the buoy, and simultaneously acquires vibration and impact signals of the buoy structure using a broadband vibration sensor. The acoustic emission signals and vibration and impact signals are fused in the time and frequency domains and input into a pre-trained deep learning neural network model for identification. If the matching degree between the identified signal features and the vessel collision features exceeds a preset threshold, a collision is determined, and a determination signal containing a collision timestamp is generated.

[0167] The collision determination module 620 is configured to use the collision timestamp of the determination signal as a reference, extend forward and backward by a preset time window, obtain the GNSS positioning data of the collided buoy within the preset time window, and filter out the AIS data of passing ships with the collided buoy as the coordinate center and a set radius range r.

[0168] The suspect analysis module 630 is configured to calculate the actual distance between the collided buoy and the passing vessel based on the GNSS positioning data of the collided buoy and the AIS data of the passing vessel; and to determine whether the passing vessel is suspicious based on the comparison between the actual distance and the theoretical distance, and to filter out the suspicious vessels.

[0169] The accident determination module 640 is configured to identify the vessel responsible for the accident from among the suspected vessels by detecting changes in the speed and course of the suspected vessels. Specifically, it includes the following sub-steps:

[0170] S41. Obtain the speed sequence and course sequence of the suspected vessel;

[0171] S42. Mark speed drop points in the speed sequence where the speed drop point decreases more than the previous data point by a factor exceeding a preset speed threshold; mark heading change points in the heading sequence where the heading angle changes more than the previous data point by a factor exceeding a preset angle threshold.

[0172] S43. A suspected vessel that meets all of the following conditions shall be identified as the vessel responsible for the accident:

[0173] There is a point where the speed drops suddenly and the timestamp of the drop is within the preset time window;

[0174] There is a course change point and the difference between the change timestamp and the descent timestamp is less than a preset time threshold.

[0175] After a collision between a ship and a buoy, the ship's trajectory will inevitably pass through the area where the buoy is located, and its motion may change. Typically, a ship colliding with a buoy may exhibit the following characteristics:

[0176] (1) The AIS navigation track of the vessel involved in the accident and the track recorded by the telemetry and remote control data of the navigation mark are adjacent to each other. Although they may not directly intersect, their spatial positions are significantly correlated.

[0177] (2) When a collision occurs, the ship’s speed may experience irregular changes such as sudden increase, sudden decrease or continuous fluctuation due to the driver’s control or the damping force of the light buoy;

[0178] (3) Under the influence of the driver’s active adjustment or the resistance of the light buoy, the ship’s heading angle will suddenly deviate from the original sailing path;

[0179] (4) Construct a dynamic overlay visualization of collision light buoys and suspected ship radar echoes in a high-resolution scene, where radar echoes are superimposed.

[0180] This invention monitors in real time whether vessels in the waters surrounding a light buoy exhibit the aforementioned abnormal motion attitudes, comprehensively assessing the likelihood of a collision between the vessel and the buoy. By replaying the AIS trajectories of surrounding vessels and the buoy's motion data, it determines whether the trajectory spacing meets the criteria for determining the severity level. If so, it tracks the offending vessel based on the spacing, speed changes, and changes in bow direction. If the judgment conditions are met, the system further matches the suspect vessel's static AIS data (including MMSI number and size information) and traces its complete navigation trajectory. Simultaneously, the system supports replaying the offending vessel's AIS trajectory along a timeline (e.g., ±1 hour), dynamically displaying key parameters such as position, speed, and heading, and automatically marking abnormal behavior points (e.g., emergency deceleration or sudden changes in heading), assisting investigators in visually reconstructing the accident process and identifying the suspect.

[0181] Thirdly, the present invention proposes a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of any of the above-described methods for tracing the source of a collision between a light buoy and the offending vessel.

[0182] Fourthly, the present invention proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of any of the above-described methods for tracing the source of a collision involving a light buoy and tracking the offending vessel.

[0183] The following is for reference. Figure 7 It shows a schematic diagram of the structure of a computer system 700 suitable for implementing terminal devices or servers in the embodiments of this application. Figure 7 The terminal device or server shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0184] like Figure 7 As shown, the computer system 700 includes a central processing unit (CPU) 701, which can perform various appropriate actions and processes based on programs stored in read-only memory (ROM) 702 or programs loaded from storage section 708 into random access memory (RAM) 703. The RAM 703 also stores various programs and data required for the operation of the system 700. The CPU 701, ROM 702, and RAM 703 are interconnected via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0185] The following components are connected to the I / O interface 705: an input section 706 including a keyboard, mouse, etc.; an output section 707 including a liquid crystal display (LCD) and speakers, etc.; a storage section 708 including a hard disk, etc.; and a communication section 709 including a network interface card such as a LAN card and a modem, etc. The communication section 709 performs communication processing via a network such as the Internet. A drive 710 is also connected to the I / O interface 705 as needed. A removable medium 711, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 710 as needed so that computer programs read from it can be installed into the storage section 708 as needed.

[0186] Specifically, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 709, and / or installed from removable medium 711. When the computer program is executed by central processing unit (CPU) 701, it performs the functions defined in the methods of this application. It should be noted that the computer-readable medium described in this application can be a computer-readable signal medium or a computer-readable medium or any combination thereof. The computer-readable medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, a computer-readable medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable medium that can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. Program code contained on a computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0187] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0188] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0189] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this application.

Claims

1. A method for tracing and tracking the offending vessel after a light buoy is struck, characterized in that, Includes the following steps: S1. Collect and preprocess the ship's AIS data, which includes MMSI code, position, speed, heading, and ship dimensions; collect underwater acoustic emission signals at the time of collision using a low-frequency acoustic sensor array deployed on the buoy, and simultaneously collect vibration and impact signals of the buoy structure using a broadband vibration sensor; perform time-frequency domain fusion processing on the acoustic emission signals and vibration and impact signals, and input them into a pre-trained deep learning neural network model for identification; if the matching degree between the identified signal features and the ship collision features exceeds a preset threshold, it is determined that the buoy has collided, and a determination signal containing a collision timestamp is generated; S2. Based on the collision timestamp of the judgment signal, extend forward and backward by a preset time window, obtain the GNSS positioning data of the collided buoy within the preset time window, and filter out the AIS data of passing ships with the collided buoy as the coordinate center and a radius range r. S3. Based on the GNSS positioning data and the AIS data of the passing vessel, calculate the actual distance between the collided light buoy and the passing vessel; based on the comparison between the actual distance and the theoretical distance, determine whether the passing vessel is suspicious, and filter out the suspicious vessels; S4. By detecting changes in the speed and course of the suspected vessels, the vessel responsible for the accident is identified from the suspected vessels. This includes the following sub-steps: S41. Obtain the speed sequence and heading sequence of the suspected vessel; S42. Mark speed drop points in the speed sequence, where the speed drop point decreases more than a preset speed threshold compared to the previous data point; mark heading change points in the heading sequence, where the heading angle changes more than a preset angle threshold compared to the previous data point. S43. A suspected vessel that meets all of the following conditions shall be identified as the vessel responsible for the accident: There exists a point where the speed suddenly drops and the timestamp of the drop occurs within the preset time window; There is a course change point and the difference between the change timestamp and the descent timestamp is less than a preset time threshold.

2. The method for tracing and tracking the offending vessel after a collision with a light buoy according to claim 1, characterized in that, Following step S4, the method further includes: S5, tracking and visually replaying the suspected vessel, specifically including the following sub-steps: S51. Extract continuous time-series radar echo data of a preset area centered on the struck buoy within a preset time window; S52. Connect the continuous radar echo points of the suspected vessel within the preset time window in sequence according to the outer edge of its echo end to generate a strip-shaped movement path of the vessel; connect the continuous radar echo points of the collided light buoy within the same time period in sequence according to the outer edge of its echo end to generate a strip-shaped movement path of the light buoy. S53. In a high-resolution electronic nautical chart scenario, the generated ribbon-shaped movement path of the ship and the ribbon-shaped movement path of the light buoy are dynamically overlaid and displayed, wherein the ribbon-shaped movement path of the ship and the ribbon-shaped movement path of the light buoy are rendered with different colors to distinguish them. S54. In the dynamic overlay visualization display, if the strip-shaped movement path area of ​​the light buoy is completely or partially inside the strip-shaped movement path area of ​​the vessel, it will help determine that the suspected vessel is the vessel that caused the accident. S55. Match the ship target in the radar echo data with the AIS signal in real time. If the match is successful, obtain the ship's AIS static information. If no AIS signal is matched, extend the playback time window of the radar echo data and continue to execute steps S51 to S54 for tracking until the AIS signal is matched and the information is successfully obtained, or the relevant data of the ship's docking at the pier is recorded.

3. The method for tracing and tracking the offending vessel after a collision with a light buoy according to claim 1, characterized in that, Also includes: Collect and preprocess buoy monitoring data, and further verify whether the buoy has collided based on the preprocessed buoy monitoring data, including the following sub-steps: S21. Analyze the voltage value sequence in the preprocessed light buoy monitoring data; If there is no valid voltage data in two consecutive feedback cycles in the voltage value sequence, and the voltage value recorded at the last time is greater than the preset voltage value, then a voltage fault flag is triggered. S22. Extract GNSS positioning data from the preprocessed light buoy monitoring data and calculate the position offset between adjacent periods according to the time series. If the cumulative offset distance within the preset time period exceeds the preset distance, a displacement fault flag will be triggered. S23. If either the voltage fault flag or the displacement fault flag is triggered, the verification result is that the light buoy has collided.

4. The method for tracing and tracking the offending vessel after a collision with a light buoy according to claim 1, characterized in that, In step S3, based on the comparison between the actual distance and the theoretical distance, it is determined whether the passing vessel is suspicious, and suspicious vessels are screened out. This specifically includes the following sub-steps: S31. Based on the ship dimensions and the diameter of the collided light buoy in the AIS data of the passing ship, construct theoretical distances according to the collision location. These theoretical distances include the theoretical distance for mid-ship collisions. Theoretical distance for stern collision Theoretical distance to the bow collision The calculation expression is as follows: ； ； ； In the formula, Indicates the length of the passing ship. This indicates the width of the passing vessel. Indicates the diameter of the buoy that was struck; S32. Compare the actual distance with the three theoretical distances constructed in step S31 to determine whether the passing vessel is suspicious, and filter out suspicious vessels, including first-level suspicious vessels, second-level suspicious vessels, and third-level suspicious vessels: If the actual distance is less than the theoretical distance for a mid-ship collision If so, the vessel that passed by is determined to be a Class I suspected vessel; If the actual distance is within the theoretical distance of the stern collision... Theoretical distance to collision with the ship If the vessel in question passes through the area, it is determined to be a Class II suspected vessel. If the actual distance is within the theoretical distance of the bow collision Theoretical distance to collision with the stern If the vessel in question passes through the area, it is determined to be a Class III suspected vessel. If the actual distance is greater than or equal to the theoretical bow collision distance If so, it is determined that the passing vessel is not suspected.

5. The method for tracing the source of a collision involving a light buoy according to claim 1, characterized in that, In step S3, based on the GNSS positioning data and the AIS data of the passing vessel, the actual distance between the struck light buoy and the passing vessel is calculated, specifically including the following sub-steps: S301. Obtain the GNSS positioning coordinates of the struck buoy and the AIS position coordinates of the passing vessel; S302. Calculate the real-time physical distance between the GNSS positioning coordinates of the struck buoy and the AIS position coordinates of the passing ship using the Euclidean distance formula; S303. Within a preset time period, select the minimum physical distance value from the real-time physical distances as the actual distance.

6. The method for tracing and tracking the offending vessel after a collision with a light buoy according to claim 1, characterized in that, In step S1, underwater acoustic emission signals at the time of collision are collected by a low-frequency acoustic sensor array deployed on the buoy, and vibration and impact signals of the buoy structure are collected by a broadband vibration sensor. The acoustic emission signals and vibration and impact signals are fused in the time and frequency domains and input into a pre-trained deep learning neural network model for identification. If the matching degree between the identified signal features and the ship collision features exceeds a preset threshold, it is determined that the buoy has collided, and a determination signal containing a collision timestamp is generated. Specifically, this includes the following sub-steps: S11. Acquire underwater acoustic emission signals at the time of collision using the low-frequency acoustic sensor array at a first sampling frequency to obtain an acoustic emission signal time series; acquire vibration and impact signals of the light buoy structure using the broadband vibration sensor at a second sampling frequency to obtain a vibration and impact signal time series; and use a GPS time synchronization module to timestamp-align the acoustic emission signal time series and the vibration and impact signal time series. S12. Bandpass filtering is applied to the acoustic emission signal time series to remove preset low-frequency noise and high-frequency interference, thereby obtaining the filtered acoustic emission signal; high-pass filtering is applied to the vibration and shock signal time series to remove environmental vibration baseline drift, thereby obtaining the filtered vibration and shock signal; amplitude normalization is applied to the filtered acoustic emission signal and the vibration and shock signal respectively. S13. Perform short-time Fourier transform on the preprocessed acoustic emission signal and vibration impact signal respectively to calculate the time spectrum of the acoustic emission signal and the time spectrum of the vibration impact signal; extract the Mel frequency cepstral coefficient feature vector from the time spectrum of the acoustic emission signal and the wavelet packet energy feature vector from the time spectrum of the vibration impact signal; perform feature-level fusion of the Mel frequency cepstral coefficient feature vector and the wavelet packet energy feature vector to generate a fused feature matrix; S14. Input the fused feature matrix into a pre-trained deep learning neural network model. The deep learning neural network model is a hybrid model combining a convolutional neural network and a long short-term memory network. It is used to output the matching score between the signal features and the ship collision features. The matching score is calculated through a normalized exponential function layer. S15. If the matching score exceeds a preset threshold, it is determined that the light buoy has collided, and a determination signal containing the collision timestamp is generated.

7. The method for tracing the source of a collision involving a light buoy according to claim 1, characterized in that, In step S1, the ship's AIS data is preprocessed, including: The ship's AIS data is grouped according to the MMSI code to generate an independent ship dataset; Based on the UTC timestamp, the time interval between adjacent data points in the independent ship dataset is calculated. If the time interval exceeds a preset threshold, the data is divided into independent trajectory segments. The GPS drift error in the independent trajectory segment is corrected by linear interpolation to obtain preprocessed AIS data.

8. A system for tracing and tracking the offending vessel after a buoy collision, used to execute the method for tracing and tracking the offending vessel after a buoy collision as described in any one of claims 1 to 7, characterized in that, The system includes: The data acquisition and preprocessing module is configured to acquire and preprocess the AIS data of the vessel, which includes: MMSI code, position, speed, heading, and vessel dimensions. It acquires underwater acoustic emission signals at the time of a collision using a low-frequency acoustic sensor array deployed on the buoy, and simultaneously acquires vibration and impact signals of the buoy structure using a broadband vibration sensor. The acoustic emission signals and vibration and impact signals are fused in the time and frequency domains and input into a pre-trained deep learning neural network model for identification. If the matching degree between the identified signal features and the vessel collision features exceeds a preset threshold, a collision is determined, and a determination signal containing a collision timestamp is generated. The collision determination module is configured to use the collision timestamp of the determination signal as a reference, extend forward and backward by a preset time window, obtain the GNSS positioning data of the collided buoy within the preset time window, and filter out the AIS data of passing ships with the collided buoy as the coordinate center and a set radius range r. The suspect analysis module is configured to calculate the actual distance between the collided buoy and the passing vessel based on the GNSS positioning data of the collided buoy and the AIS data of the passing vessel; and to determine whether the passing vessel is suspicious based on the comparison between the actual distance and the theoretical distance, and to filter out the suspicious vessels. The accident determination module is configured to identify the vessel responsible for the accident from among the suspected vessels by detecting changes in the speed and course of the suspected vessels. Specifically, it includes the following sub-steps: Obtain the speed and heading sequence of the suspected vessel; Mark speed drop points in the speed sequence, where the speed drop point decreases more than a preset speed threshold compared to the previous data point; mark heading change points in the heading sequence, where the heading angle changes more than a preset angle threshold compared to the previous data point. The suspected vessel that meets all of the following conditions will be identified as the vessel responsible for the incident: There exists a point where the speed suddenly drops and the timestamp of the drop occurs within the preset time window; There is a course change point and the difference between the change timestamp and the descent timestamp is less than a preset time threshold.

9. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method for tracing the source of a collision involving a buoy as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the method for tracing the source of a collision involving a buoy as described in any one of claims 1 to 7.

Images