Ship navigation trajectory prediction method and system
By combining inertial navigation with historical trajectory models and employing credibility assessment and Kalman filters for data fusion, the problem of trajectory prediction when historical data is sparse is solved, achieving high-precision and robust ship navigation trajectory prediction.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-19
- Publication Date
- 2026-03-31
AI Technical Summary
Existing technologies suffer from sparse or missing historical data when ships navigate to new routes, remote sea areas, or areas where AIS signals are lost. This causes historical trajectory prediction models to fail, and the errors of inertial navigation systems accumulate over time, making it difficult to achieve high-precision and robust trajectory prediction.
By combining inertial navigation with historical trajectory models, and through confidence assessment and dynamic data fusion, Kalman filters are used for error correction, data fusion weights are dynamically adjusted, and high-confidence historical trajectories are used to correct inertial navigation errors, thereby achieving continuity and high accuracy in trajectory prediction.
Highly reliable trajectory prediction was achieved under sparse data conditions, suppressing inertial navigation error drift and improving prediction accuracy and robustness in complex environments.
Smart Images
Figure CN121764087A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of ship navigation and trajectory prediction technology, specifically to a method and system for predicting ship navigation trajectories by integrating inertial navigation information and historical trajectory data. Background Technology
[0002] With the rapid development of the global shipping industry, the demand for accurate analysis and prediction of ship trajectories is constantly increasing. Existing Automatic Identification Systems (AIS) can provide a large amount of historical navigation data. Based on this data, academia and industry have proposed a variety of trajectory prediction models, such as navigation pattern learning methods based on Long Short-Term Memory Networks (LSTM), Gated Recurrent Units (GRU), or clustering algorithms.
[0003] However, the existing technology has the following significant shortcomings: When ships navigate to new routes, remote waters, or areas where AIS signals are lost, historical data is sparse or missing, resulting in the historical trajectory database not containing the corresponding motion patterns. In this situation, traditional prediction models based on historical pattern matching fail, trajectory prediction accuracy decreases significantly, and continuous prediction may even become impossible. While inertial navigation systems (INS) possess full autonomy and high short-term accuracy, their errors accumulate over time due to the zero bias and noise inherent in gyroscopes and accelerometers, leading to position drift and making it difficult to maintain long-term accuracy. Therefore, achieving highly reliable trajectory prediction even with sparse historical data, while simultaneously suppressing error drift in INS, is a pressing technical challenge in this field. Summary of the Invention
[0004] The main objective of this invention is to propose a method and system for predicting ship navigation trajectories that combines inertial navigation and historical trajectory models, so as to achieve continuous prediction capability and high-precision robustness in data-sparse scenarios, while also achieving adaptive correction of inertial navigation system errors.
[0005] A method for predicting ship navigation trajectories, specifically employing the following technical means: The ship's inertial navigation data is acquired, and inertial navigation calculation is performed based on the inertial navigation data to obtain the first trajectory prediction result; Acquire historical navigation data of the vessel, and obtain a second trajectory prediction result based on the historical navigation data using a trajectory prediction model; The credibility of the historical navigation data is evaluated in terms of spatial density and pattern similarity to obtain a credibility index for the historical data. Based on the credibility index evaluation results, the data fusion weights of the first trajectory prediction result and the second trajectory prediction result are dynamically determined; The first trajectory prediction result and the second trajectory prediction result are fused according to the data fusion weight to generate the final ship navigation trajectory prediction result.
[0006] Furthermore, the credibility assessment includes: assessing the spatial density of the historical navigation data, and assessing the similarity between the trajectory shape of the first trajectory prediction result and the pattern of the historical navigation data.
[0007] Furthermore, the similarity between the trajectory shape of the first trajectory prediction result and the historical navigation data pattern is evaluated using a dynamic time warping algorithm or a Fraser distance algorithm.
[0008] Furthermore, dynamically determining the data fusion weights of the first trajectory prediction result and the second trajectory prediction result includes: when the credibility of the historical navigation data is lower than a preset threshold, increasing the fusion weight of the first trajectory prediction result and correspondingly decreasing the fusion weight of the second trajectory prediction result; when the credibility of the historical navigation data is higher than the threshold, increasing the fusion weight of the second trajectory prediction result.
[0009] Furthermore, the relationship between the data fusion weights and the credibility index is a non-linear mapping relationship, which is represented by the Sigmoid function.
[0010] Furthermore, a Kalman filter is used for fusion processing, wherein: the first trajectory prediction result is used as the state prediction value, the second trajectory prediction result is used as the measurement observation value, and the measurement noise covariance matrix R of the Kalman filter is dynamically adjusted according to the credibility evaluation result to achieve adaptive optimization of fusion accuracy.
[0011] Furthermore, the state vector of the Kalman filter includes sensor error parameters of the inertial navigation system, and the Kalman filter estimates and corrects these sensor error parameters simultaneously when updating using measured observations.
[0012] A ship navigation trajectory prediction system, characterized in that it includes: The data acquisition module is used to acquire real-time angular velocity and acceleration data from the ship's IMU sensors; and to acquire historical position, speed, and heading status data of the ship from the AIS historical database or real-time receiving equipment. The inertial navigation calculation module performs inertial navigation calculations based on inertial navigation data to obtain the first trajectory prediction result; The ship trajectory prediction module obtains a second trajectory prediction result based on historical navigation data and a trajectory prediction model. The credibility assessment module evaluates the credibility of the historical navigation data in terms of spatial density and pattern similarity, and obtains the credibility index of the historical data. The data fusion weighting module dynamically determines the data fusion weights of the first trajectory prediction result and the second trajectory prediction result based on the credibility index evaluation results. The ship trajectory generation module fuses the first trajectory prediction result and the second trajectory prediction result according to the data fusion weight to generate the final ship navigation trajectory prediction result.
[0013] This invention provides a method and system for predicting ship navigation trajectories. The method establishes a measurement noise covariance matrix. With credibility indicators The inverse mapping function relationship is essentially an information entropy-driven covariance inflation strategy. When the amount of information in the external environment decreases, the system automatically increases the measurement uncertainty parameter. Therefore, this method mathematically guarantees that the filter can automatically degenerate into a pure inertial inference mode in data-sparse regions, while asymptotically converging to the optimal estimate in data-rich regions. The state space of the filter is expanded by incorporating the internal parameters of the inertial navigation system as augmented state variables to be estimated into the system equations. High-confidence historical trajectories are used as observed ground truth to identify and correct these hardware errors in real time.
[0014] In addition, this method uses a dynamic time warping algorithm as a temporal feature extractor to calculate the morphological isomorphism between the current inertial navigation trajectory and the historical trajectory. This can effectively identify false high-confidence scenarios where historical data is dense but does not match the current navigation intention, thus avoiding misleading predictions from erroneous historical data and significantly improving the accuracy of assessment in complex navigation environments.
[0015] Therefore, this ship navigation trajectory prediction method and system intelligently combines the advantages of INS and historical data models, providing high robustness and high accuracy trajectory prediction under all operating conditions. Attached Figure Description
[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0017] Figure 1 This is a flowchart of a ship navigation trajectory prediction method provided in an embodiment of the present invention.
[0018] Figure 2This is a functional module block diagram of a ship navigation trajectory prediction system provided in an embodiment of the present invention. Detailed Implementation
[0019] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.
[0020] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0021] like Figure 1 As shown, the present invention provides a method for predicting ship navigation trajectories, which specifically includes the following steps: S1: Data Acquisition and Preprocessing: Acquire ship's inertial measurement unit (IMU) data (including angular velocity measured by gyroscopes and acceleration measured by accelerometers); acquire ship's historical navigation data (such as AIS data).
[0022] S2: Inertial navigation calculation: Based on the inertial navigation data, the first trajectory prediction result (i.e., INS calculated trajectory) is obtained through the inertial navigation calculation model (Straight-through inertial navigation SINS algorithm).
[0023] S3: Ship navigation trajectory prediction: Based on the historical navigation data, a second trajectory prediction result (i.e., the trajectory predicted by the historical data model) is obtained through the ship trajectory prediction model (LSTM network).
[0024] S4: Credibility Assessment: Establish a credibility assessment unit to evaluate the credibility of the historical navigation data across one or more dimensions.
[0025] In a preferred embodiment, the evaluation includes not only evaluating the spatial density of historical data in the current area, but also evaluating the similarity of the trajectory patterns of the first trajectory prediction result (INS trajectory) and the historical navigation data (DTW or Fréchet distance algorithm).
[0026] S4: Dynamic Data Fusion: Establish a data fusion model, the core of which is to dynamically determine the data fusion weights of the first trajectory prediction result and the second trajectory prediction result in the fusion based on the evaluation results of the credibility.
[0027] When the confidence level is lower than a preset threshold (i.e., when historical data is insufficient or the pattern does not match), the weight of the first trajectory prediction result (INS trajectory) is automatically increased.
[0028] S5: Trajectory Output: Based on the data fusion weights, the first trajectory prediction result and the second trajectory prediction result are fused to generate the fused ship navigation trajectory.
[0029] In a preferred embodiment, the dynamic data fusion step is implemented using an adaptive Kalman filter (EKF): The first trajectory prediction result (INS trajectory) is used as the state prediction value of the filter. The second trajectory prediction result (historical model trajectory) is used as the measurement observation value of the filter. Its key innovation lies in the measurement noise covariance matrix of the filter. It is dynamically adjusted based on the results of the credibility assessment. When the credibility is low, As the value increases, the filter automatically reduces its trust in historical model trajectories.
[0030] The state vector of the Kalman filter is expanded to include key sensor error parameters of the inertial navigation system.
[0031] like Figure 2 The present invention also provides a ship navigation trajectory prediction system, which includes: The data acquisition module is used to acquire real-time raw angular velocity and acceleration data from the ship's IMU sensors (Inertial Measurement Units); and to acquire historical navigation status data such as the ship's position, speed, and heading from the AIS historical database or real-time receiving equipment. This module is also responsible for preprocessing heterogeneous data, such as time synchronization and noise removal. The inertial navigation calculation module is an important component of this invention. It receives IMU data from the data acquisition module and performs standard strapdown inertial navigation (SINS) calculations. This includes attitude updates, velocity updates, and position updates. By integrating and transforming the data from the gyroscope and accelerometer, this module outputs a first trajectory prediction result (…). This trajectory demonstrates extremely high accuracy over a short period and does not rely on any external data. The ship trajectory prediction module receives historical AIS data from the data acquisition module. In one embodiment, this module embeds a deep learning model (such as an LSTM network) pre-trained with massive amounts of AIS data. Based on the input ship's current motion state (such as latitude and longitude, speed over ground, heading angle, etc.), it predicts a second trajectory for a future period or a historical period that needs to be supplemented. This result reflects the typical navigation patterns of ships in a specific sea area; The credibility assessment module evaluates the credibility of the historical navigation data in terms of spatial density and pattern similarity, and obtains the credibility index of the historical data. The data fusion weighting module dynamically determines the data fusion weights of the first trajectory prediction result and the second trajectory prediction result based on the credibility index evaluation results. The ship trajectory generation module fuses the first trajectory prediction result and the second trajectory prediction result according to the data fusion weight to generate the final ship navigation trajectory prediction result.
[0032] Example 1 This embodiment aims to address the prediction failure problem when historical data is sparse or navigation patterns change abruptly. It constructs a two-dimensional reliability evaluation system based on spatial density and pattern similarity, and combines this with a nonlinear sigmoid function to achieve smooth adaptive switching of weights. This scheme ensures that the system maintains the optimal fusion ratio between inertial navigation estimation and historical trajectory prediction under different data quality environments. The specific steps are as follows: 1. Acquiring and Solving Inertial Navigation Data (Obtaining the First Trajectory Prediction Result) The system first acquires raw data (angular velocity and acceleration) from the ship's IMU sensors. Using the inertial navigation solution model (in this embodiment, the strapdown inertial navigation SINS algorithm is used), the raw data is integrated to calculate the ship's first trajectory prediction result in real time (denoted as...). The results are characterized by high accuracy in a short period of time and independence from external information.
[0033] 2. Acquire Historical Navigation Data and Predict (Obtain the Second Trajectory Prediction Result) The system simultaneously acquires the ship's historical navigation data (such as AIS data). Using a pre-trained ship trajectory prediction model (in this embodiment, an LSTM long short-term memory network is used), based on the ship's current motion state, the system predicts the second trajectory prediction result (denoted as...). This result reflects the usual navigation patterns in this sea area.
[0034] 3. Evaluate the credibility of historical data (obtain credibility index): This step involves a multi-dimensional evaluation of the historical data used as a basis to obtain a credibility index. This embodiment innovatively employs an evaluation method that combines spatial density and pattern similarity: Spatial density assessment ): The location of the current prediction point (e.g.) The sea area is divided into grids, and the distribution density of historical data points in the grids is statistically analyzed.
[0035] Pattern similarity assessment ( ):extract Short-term trajectory segments are analyzed using Dynamic Time Warping (DTW) or Fréchet distance algorithms to calculate their morphological similarity to historical trajectory segments. This effectively identifies situations where "there is abundant historical data, but it does not match the current navigation intention."
[0036] Final credibility It is a weighted function of the two mentioned above. (The preset weighting coefficient is 0.5).
[0037] 4. Dynamically determine data fusion weights based on the obtained credibility index. This embodiment uses a non-linear sigmoid function to dynamically determine the weights in order to achieve smooth soft switching: Historical model weights The calculation is as follows ( This is the credibility "switching threshold", set to 0.4; (This is the steepness coefficient of the switching curve).
[0038] INS weight The calculation is as follows:
[0039] 5. Fusion to generate the final trajectory. When historical data highly matches the current pattern ( ), The value quickly approaches 1, indicating that the system trusts historical data.
[0040] When historical data is sparse or the pattern does not match ( ), It rapidly approaches 0. The value rapidly approaches 1. The system will then smoothly and decisively switch to relying primarily on INS calculations, thus achieving the goal of effectively compensating for insufficient historical data using inertial navigation.
[0041] Final fusion trajectory The calculation is as follows:
[0042] This achieves the objective of the present invention: the system automatically switches to relying mainly on inertial navigation for calculation when historical data is insufficient, thereby successfully making up for the lack of historical data.
[0043] Example 2 This embodiment focuses on the closed-loop correction and fusion process based on the extended Kalman filter (EKF). This fusion process is used not only for outputting the trajectory but also for reverse calibration of the error parameters of the INS sensor, thus achieving closed-loop correction.
[0044] Extended Kalman Filter (EKF) State Vector Definition: Unlike traditional filters that only provide position and velocity, the state vector of this invention... The key sensor error parameters for the INS must be included:
[0045] in, The positions are north and east. For northbound and eastbound speeds; For heading or attitude angle; This is the gyroscope bias. This refers to the accelerometer bias.
[0046] Acquire and solve inertial navigation data (corresponding to EKF time update / state prediction). The system acquires IMU data and solves it using the SINS algorithm. In the Kalman filter, this step constitutes state prediction. The system then calculates the first trajectory prediction result obtained from the inertial navigation solution. ) is directly used as the prior state estimate of the filter. ).
[0047] The system acquires historical navigation data and makes predictions (corresponding to EKF measurement acquisition). The system utilizes the historical trajectory prediction model to obtain the second trajectory prediction result. ), and use it as the measured observation value of the filter ( ).
[0048]
[0049] The assessment of historical data reliability is the same as in Example 1; the system calculates a reliability index for the historical data. This indicator reflects the reliability of the second trajectory prediction results as a "measurement".
[0050] Dynamically determining data fusion weights (corresponding to EKF measurement noise matrix adjustment) in Kalman filtering is implicit in the Kalman gain. In the calculation of ), and Depends on the measurement noise covariance matrix This step innovatively utilizes step S3. Values to dynamically adjust the matrix :
[0051] When credibility When reduced, the matrix The value increases (the noise variance increases). According to the Kalman filtering principle, this will lead to an increase in the calculated Kalman gain. The weight of the second trajectory prediction result (measurement value) is reduced in the update step, thereby automatically decreasing the weight of the first trajectory prediction result (state prediction value).
[0052] The fusion generates the final trajectory (corresponding to EKF measurement updates / state corrections). The filter utilizes the adjusted... Matrix and measured observations ( ), for prior state ( The system updates the data to generate the final ship trajectory prediction result (posterior state estimate). ).
[0053] In this step, if the historical data is reliable ( The high-resolution filter not only outputs the fused, precise trajectory, but also utilizes the measurement residuals to adjust the sensor error parameters in the state vector. , The system performed estimations and corrections, achieving closed-loop calibration of the inertial navigation system. This ensures that when the ship next sails into a data-sparse region, the calculation accuracy of the inertial navigation solution module itself has been improved, greatly enhancing the overall robustness of the system.
[0054] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.
[0055] In the above embodiments of the present invention, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.
[0056] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.
[0057] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.
[0058] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.
[0059] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for predicting ship navigation trajectories, characterized in that, include: The ship's inertial navigation data is acquired, and inertial navigation calculation is performed based on the inertial navigation data to obtain the first trajectory prediction result; Acquire historical navigation data of the vessel, and obtain a second trajectory prediction result based on the historical navigation data using a trajectory prediction model; The credibility of the historical navigation data is evaluated in terms of spatial density and pattern similarity to obtain a credibility index for the historical data. Based on the credibility index evaluation results, the data fusion weights of the first trajectory prediction result and the second trajectory prediction result are dynamically determined; The first trajectory prediction result and the second trajectory prediction result are fused according to the data fusion weight to generate the final ship navigation trajectory prediction result.
2. The method for predicting a ship's navigation trajectory according to claim 1, characterized in that: The credibility assessment includes: assessing the spatial density of the historical navigation data, and assessing the similarity between the trajectory shape of the first trajectory prediction result and the pattern of the historical navigation data.
3. The method for predicting ship navigation trajectory according to claim 1, characterized in that, The similarity between the trajectory shape of the first trajectory prediction result and the historical navigation data pattern is evaluated using a dynamic time warping algorithm or a Fraser distance algorithm.
4. The method for predicting ship navigation trajectory according to claim 1, characterized in that, Dynamically determining the data fusion weights of the first trajectory prediction result and the second trajectory prediction result includes: when the credibility of the historical navigation data is lower than a preset threshold, increasing the fusion weight of the first trajectory prediction result and correspondingly decreasing the fusion weight of the second trajectory prediction result; when the credibility of the historical navigation data is higher than the threshold, increasing the fusion weight of the second trajectory prediction result.
5. The method for predicting a ship's navigation trajectory according to claim 1, characterized in that: The relationship between the data fusion weights and the credibility index is a non-linear mapping relationship, which is represented by the Sigmoid function.
6. The method for predicting a ship's navigation trajectory according to claim 1, characterized in that: A Kalman filter is used for fusion processing, wherein the first trajectory prediction result is used as the state prediction value, the second trajectory prediction result is used as the measurement observation value, and the measurement noise covariance matrix R of the Kalman filter is dynamically adjusted according to the credibility evaluation result to achieve adaptive optimization of fusion accuracy.
7. The method for predicting a ship's navigation trajectory according to claim 6, characterized in that: The state vector of the Kalman filter includes sensor error parameters of the inertial navigation system. When the Kalman filter is updated using measured observations, it simultaneously estimates and corrects the sensor error parameters.
8. A ship navigation trajectory prediction system, characterized in that: include: The data acquisition module is used to acquire real-time angular velocity and acceleration data from the ship's IMU sensors; And obtain the ship's historical position, speed, and heading status data from the AIS historical database or real-time receiving equipment; The inertial navigation calculation module performs inertial navigation calculations based on inertial navigation data to obtain the first trajectory prediction result; The ship trajectory prediction module obtains a second trajectory prediction result based on historical navigation data and a trajectory prediction model. The credibility assessment module evaluates the credibility of the historical navigation data in terms of spatial density and pattern similarity, and obtains the credibility index of the historical data. The data fusion weighting module dynamically determines the data fusion weights of the first trajectory prediction result and the second trajectory prediction result based on the credibility index evaluation results. The ship trajectory generation module fuses the first trajectory prediction result and the second trajectory prediction result according to the data fusion weight to generate the final ship navigation trajectory prediction result.