Large model-based ship track prediction system and method
Through multi-channel data collection and feature fusion, anomaly detection and meta-learning models are constructed, which solves the robustness and safety of ship track prediction in new environments and extreme conditions, and achieves high-precision and real-time track prediction. and security decision-making assistance.
Patent Information
- Application Number
- CN202411004353.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-25
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-07-25
AI Technical Summary
The existing ship track prediction technology has poor prediction results in new environments and extreme conditions, lacks robustness and real-time data adaptability, and is difficult to evaluate track safety in real time, which poses safety risks.
Collect multimodal data through multiple channels, perform feature extraction and fusion, build anomaly detection model and meta-learning model, quickly adapt to specific ship behavior, combine real-time data to make predictions, and establish a security decision-making auxiliary system to trigger alerts and provide security suggestions. .
It improves the accuracy and safety of ship track prediction, enhances the robustness and real-time adaptability of the model, and ensures safety decision-making support in complex sea conditions.
Smart Images

Figure CN119004358A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of marine intelligent navigation technology, and in particular to a ship track prediction system and method based on a large model. Background Art
[0002] In marine transportation and military operations, ship track prediction is crucial to ensure navigation safety, avoid collision accidents, plan efficient routes and tactical deployment. In recent years, with the rapid development of artificial intelligence technology, especially the application of deep learning and big data analysis, ship track prediction technology has been significantly improved. Traditional track prediction methods mainly rely on physical models and statistical analysis. Although these methods can reflect the basic laws of ship movement, their prediction accuracy is often limited in complex and changeable marine environments. Physical models are limited by simplified assumptions about environmental factors, while statistical methods are difficult to capture the impact of nonlinear and sudden events.
[0003] In recent years, ship track prediction methods based on machine learning have gradually become a research hotspot. Such methods use multi-source information such as historical track data, environmental data, and ship characteristics to train deep neural network models, which can more accurately predict the future course and speed of the ship. However, the current prediction model still has some key problems. On the one hand, the generalization ability of the model is limited, and the prediction effect on new environments or specific ship behaviors is poor, especially when dealing with abnormal data and extreme conditions. The robustness of the model needs to be enhanced. On the other hand, most methods lack the ability to quickly adapt to real-time data, resulting in a decrease in prediction timeliness and accuracy. In addition, existing technologies also have limitations in safety decision-making assistance. It is difficult to evaluate the safety of the predicted track in real time, and there is a lack of effective alarm and suggestion mechanisms, which may cause safety hazards in complex sea conditions. Summary of the invention
[0004] In view of the above existing problems, the present invention is proposed.
[0005] Therefore, the present invention provides a ship track prediction system and method based on a large model to solve the problem that the generalization ability of the model is limited, and the prediction effect on new environments or specific ship behaviors is poor, especially in processing abnormal data and extreme conditions, the robustness of the model needs to be enhanced. On the other hand, most methods lack the ability to quickly adapt to real-time data, resulting in reduced timeliness and accuracy of prediction. In addition, the existing technology also has limitations in assisting safety decision-making. It is difficult to evaluate the safety of the predicted track in real time, and there is a lack of effective alarm and suggestion mechanisms, which may cause safety hazards in complex sea conditions.
[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:
[0007] In a first aspect, an embodiment of the present invention provides a ship track prediction method based on a large model, which comprises:
[0008] Collect multimodal data through multiple channels;
[0009] Extract features from multimodal data separately and output fused environmental perception features;
[0010] An anomaly detection model is built based on the fused environmental perception features. The fused environmental perception feature vector is input into the anomaly detection model. The abnormal data is identified and filtered through the causal reasoning algorithm to output pure features.
[0011] A meta-learning model is constructed using the meta-learning mechanism and pure features. Pure features are input into the meta-learning model, and a personalized prediction model is output through a sub-model that quickly adapts to the specific ship behavior pattern.
[0012] Combine the personalized prediction model with real-time data, use the large model to predict the ship's track, and output the predicted track;
[0013] A safety decision support system is established to compare the predicted track with the navigation rules. When the predicted result exceeds the safety threshold, an alarm is triggered and safety recommendations are provided to the ship.
[0014] As a preferred solution of the ship track prediction method based on a large model of the present invention, wherein: the multi-modal data is collected through multiple channels, and the specific steps are:
[0015] Collect unstructured text description data, historical track data, and environmental condition data from multiple sources;
[0016] Collect radar images and sonar signals through sensors;
[0017] The collected data are preprocessed by denoising and filling missing values.
[0018] As a preferred solution of the ship track prediction method based on a large model of the present invention, wherein: the feature extraction of multimodal data is performed separately, and the fused environment perception features are output, the specific steps are:
[0019] The BERT model is used to convert unstructured text descriptions into numerical vectors to obtain text feature vectors, which are expressed as follows:
[0020] T = BERT(wo);
[0021] Where T represents the text features obtained after being processed by the BERT model, and wo represents the input text sequence;
[0022] The sliding window technology is used to extract the change rate of the speed, direction and position of the track to obtain the track feature vector. The sliding window size is set to W, the step size of the sliding window is s, and the expression is:
[0023]
[0024] Among them, H represents the track characteristics, i represents the starting time point of the window, represents the summation operation starting from time point i and ending at time point i+W-1, v t represents the velocity vector at time t, θ t represents the direction angle at time t, x t represents the position coordinate at time t, x t+s -x t represents the displacement vector between time t and time t+s;
[0025] Discrete Fourier transform is used to perform spectrum analysis on environmental condition data, and the meteorological data at time t is set as m t , the ocean current data at time t is o t , the characteristic expression of environmental conditions is extracted as follows:
[0026] E = (DFT(m), DFT(o));
[0027] Where E represents the environmental condition characteristics, DFT represents discrete Fourier transform;
[0028] Convolutional neural network CNN is used to extract features from sensor data. The two-dimensional matrix of sensor data is set to so, and the expression is:
[0029] S = CNN(so) = ReLU(K*so);
[0030] Among them, S represents sensor features, K represents CNN filter kernel, ReLU represents activation function, and * represents convolution operation;
[0031] The extracted text features, track features, environmental condition features and sensor features are fused by weighted averaging, and the expression is:
[0032] Z=α T T+α H H+α E E+α S S;
[0033] α T +α H +α E +α S =1;
[0034] Among them, Z represents the fused environmental perception features, α T , α H , α E and α S They represent the weights of text features, track features, environmental condition features, and sensor features respectively.
[0035] As a preferred solution of the ship track prediction method based on a large model described in the present invention, wherein: the anomaly detection model is constructed based on the fused environmental perception features, the fused environmental perception feature vector is input into the anomaly detection model, the abnormal data is identified and filtered through the causal reasoning algorithm, and the pure features are output. The specific steps are:
[0036] The anomaly detection model is constructed by using multivariate Gaussian distribution, information entropy, and the latent causal variable model LCM in causal reasoning. The expression of the anomaly detection model is:
[0037]
[0038] Among them, A(Z) represents the output of the anomaly detection model, represents the integral symbol, μ represents the mean vector, Σ represents the covariance matrix, N(Z; μ, Σ) represents the multivariate Gaussian distribution function, ι(Z) represents the information filtering function, exp(-ι(Z)) represents the exponential function, and du is the differential element, which represents a small change in the integral variable μ;
[0039] The expression of information filtering function ι(Z) is:
[0040]
[0041] Among them, p i represents the probability of the occurrence of the i-th feature, and n represents Z j The number of features in, log represents the natural logarithmic function, when p i The smaller it is, the larger the absolute value of log is. λ represents the regularization parameter. It means to sum all items from 1 to m, where m represents The number of features in Z j represents the actual observed value of the jth feature in Z, represents the expected value of the j-th feature, and q represents the exponential operation.
[0042] As a preferred solution of the ship track prediction method based on a large model of the present invention, the meta-learning model is constructed by using the meta-learning mechanism and pure features, the pure features are input into the meta-learning model, and the personalized prediction model is output by a sub-model that quickly adapts to the specific ship behavior mode. The specific steps are as follows:
[0043] The meta-learning structure is assumed to consist of a shared base model and a series of sub-models for each task;
[0044] Set each ship as k, and then build a sub-model to quickly adapt to the specific behavior pattern of ship k through historical ship track data;
[0045] Based on the ship and the basic model P, an adaptability measurement function is constructed, and the expression is:
[0046]
[0047] Among them, R k (P,Z k ) indicates that ship k has pure feature Z k and the reliability measure under prediction P, where P represents the prediction result of the basic model, Z k represents the pure feature set of ship k, T represents the total number of time steps, d k (t) represents the deviation at time t, α k represents a constant, β k represents another constant, μ d represents the mean of the deviations, represents the variance of the deviation, N represents the normal distribution function, It means calculating the sum of probability densities of normal distribution in the entire deviation space;
[0048] Using the fitness metric function as the objective function R k , optimize the sub-model;
[0049] When the sub-model is optimized, that is, the personalized prediction model is optimized, the output vector is R k (t).
[0050] As a preferred solution of the ship track prediction method based on a large model described in the present invention, the personalized prediction model is combined with real-time data, the ship track is predicted by the large model, and the predicted track is output. The specific steps are:
[0051] Assume that the real-time data input vector at time t is X(t);
[0052] Based on real-time data X(t) and R k (t), and at the same time combine the pure feature set Z of ship k k Construct a composite function G, the expression is:
[0053]
[0054] Among them, Y k (t) represents the predicted track of ship k at time t, P k(t) represents the output vector of the personalized prediction model of ship k at time t, w i It is used to adjust the importance of each sensor data in the comprehensive prediction. ReLU represents the activation function, K represents the convolution kernel, B represents the matrix for adjusting the output of the personalized prediction model, C represents the matrix for adjusting the pure features, and μ k represents the mean of real-time data, Represents the variance of real-time data.
[0055] As a preferred solution of the ship track prediction method based on a large model of the present invention, wherein: the establishment of a safety decision support system compares the predicted track with the navigation rules, and when the prediction result exceeds the safety threshold, an alarm is triggered and safety recommendations are provided to the ship. The specific steps are:
[0056] Set the navigation rule set to R;
[0057] The safety assessment function is constructed based on the predicted trajectory of the ship, and the expression is:
[0058]
[0059] Among them, S(t) represents the output value of the security assessment function at time t, R i represents the i-th navigation rule, Y k (t) i represents the value of the i-th dimension of the k-th predicted trajectory at time t, n represents the number of navigation rules, γ represents the exponent, τ represents the preset safety threshold, and min represents the minimum value;
[0060] The alarm trigger function A(t) is set based on the safety assessment function and the safety threshold. The expression is:
[0061]
[0062] This formula indicates that when the value of S(t) is greater than the preset safety threshold τ, A(t) returns 1, indicating that an alarm needs to be triggered; otherwise, it returns 0, indicating that an alarm is not triggered;
[0063] Design a security suggestion function to provide security suggestions SG k (t), the expression is:
[0064]
[0065] Among them, v c represents the corrected velocity vector based on the predicted track deviation from the navigation rules, a a represents the adjusted acceleration vector based on the predicted track deviation from the navigation rule, α k and β k They represent the adjustment coefficients for correcting speed and adjusting acceleration respectively.
[0066] In a second aspect, the present invention provides a ship track prediction system based on a large model, comprising:
[0067] The acquisition module is responsible for collecting multimodal data from various sources, including unstructured text descriptions, historical tracks, environmental conditions, radar images, and sonar signals, and preprocessing these data by denoising and filling missing values;
[0068] The feature module is responsible for extracting features from the preprocessed data, converting text data into numerical vectors, extracting the speed, direction, and position change rate of the track, performing spectrum analysis on environmental conditions, performing convolutional neural network processing on sensor data, and finally fusing all features through weighted averaging to form environmental perception features;
[0069] The detection module is responsible for building an anomaly detection model, identifying and filtering out abnormal data, building a meta-learning model using the meta-learning mechanism and pure features, quickly adapting to the behavior patterns of specific ships through sub-models, and outputting a personalized prediction model;
[0070] The prediction module is responsible for combining the personalized prediction model with real-time data, integrating sensor data, environmental conditions, navigation data and communication data using composite functions, predicting the ship's track through a large model, and outputting the predicted track and its confidence level;
[0071] The decision-making module is responsible for establishing a safe decision-making support system, comparing the predicted trajectory with the navigation rules, and calculating the safety level through the safety assessment function. When the predicted result exceeds the safety threshold, an alarm is triggered, and safety recommendations for correcting speed and adjusting acceleration are generated based on the degree of deviation to help ships avoid risks.
[0072] In a third aspect, an embodiment of the present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the ship track prediction method based on a large model as described in the first aspect of the present invention is implemented.
[0073] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the ship track prediction method based on a large model as described in the first aspect of the present invention is implemented.
[0074] The beneficial effects of the present invention are as follows: the present invention collects multimodal data through multiple channels, uses advanced feature extraction technology to fuse environmental perception information, constructs an anomaly detection model to enhance data quality, and uses a meta-learning mechanism to quickly adapt to specific ship behaviors, ultimately achieving high-precision, real-time track prediction and safe decision-making assistance. The present invention focuses on solving the accuracy and safety issues of ship track prediction in complex marine environments, and aims to improve the intelligence level and safety of ship navigation. BRIEF DESCRIPTION OF THE DRAWINGS
[0075] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.
[0076] Figure 1 This is a flow chart of the ship track prediction method based on the large model in Example 1.
[0077] Figure 2 This is a flowchart of alarm triggering judgment in the ship track prediction method based on a large model in Example 1. DETAILED DESCRIPTION
[0078] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.
[0079] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.
[0080] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.
[0081] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, and provides a ship track prediction method based on a large model, comprising the following steps:
[0082] S1 collects multimodal data through multiple channels;
[0083] Collect unstructured text description data, historical track data, and environmental condition data from multiple sources;
[0084] Collect radar images and sonar signals through sensors;
[0085] Preprocess the collected data by denoising and filling missing values;
[0086] Unstructured textual description data, including logbooks and chart annotations;
[0087] Historical track data includes GPS records;
[0088] Environmental conditions data, including meteorological data and ocean current data.
[0089] S2 extracts features from multimodal data respectively and outputs fused environmental perception features;
[0090] The BERT model is used to convert unstructured text descriptions into numerical vectors to obtain text feature vectors, which are expressed as follows:
[0091] T = BERT(wo);
[0092] Where T represents the text features obtained after being processed by the BERT model, wo represents the input text sequence, and its value range is 1 to n, where n represents the number of words;
[0093] The sliding window technology is used to extract the change rate of the speed, direction and position of the track to obtain the track feature vector. The sliding window size is set to W, the step size of the sliding window is s, and the expression is:
[0094]
[0095] Among them, H represents the track characteristics, i represents the starting time point of the window, represents the summation operation starting from time point i and ending at time point i+W-1, v t represents the velocity vector at time t, θ t represents the direction angle at time t, x t represents the position coordinate at time t, x t+s -x t represents the displacement vector between time t and time t+s;
[0096] Discrete Fourier transform is used to perform spectrum analysis on environmental condition data, and the meteorological data at time t is set as m t , the ocean current data at time t is o t , the characteristic expression of environmental conditions is extracted as follows:
[0097] E = (DFT(m), DFT(o));
[0098] Where E represents the environmental condition characteristics, DFT represents discrete Fourier transform;
[0099] Convolutional neural network CNN is used to extract features from sensor data. The two-dimensional matrix of sensor data is set to so, and the expression is:
[0100] S = CNN(so) = ReLU(K*so);
[0101] Among them, S represents sensor features, K represents CNN filter kernel, ReLU represents activation function, and * represents convolution operation;
[0102] The extracted text features, track features, environmental condition features and sensor features are fused by weighted averaging, and the expression is:
[0103] Z=α T T+α H H+α E E+α S S;
[0104] α T +α H +α E +α S =1;
[0105] Among them, Z represents the fused environmental perception features, αT, αH, αE and αS represent the weights of text features, track features, environmental condition features and sensor features respectively, and the sum of the weights of the four features is 1.
[0106] S3 builds an anomaly detection model based on the fused environmental perception features, inputs the fused environmental perception feature vector into the anomaly detection model, identifies and filters abnormal data through the causal reasoning algorithm, and outputs pure features;
[0107] The anomaly detection model is constructed by using multivariate Gaussian distribution, information entropy, and the latent causal variable model LCM in causal reasoning. The expression of the anomaly detection model is:
[0108]
[0109] Among them, A(Z) represents the output of the anomaly detection model, represents the integral symbol, μ represents the mean vector, Σ represents the covariance matrix, N(Z; μ, Σ) represents the multivariate Gaussian distribution function, which is used to describe the distribution of normal data, ι(Z) represents the information filtering function, which is used to evaluate the information entropy and characteristic deviation of the eigenvector Z, exp(-ι(Z)) represents the exponential function, and du is the differential element, which represents a small change in the integral variable μ;
[0110] The expression of information filtering function ι(Z) is:
[0111]
[0112] Among them, p i represents the probability of the occurrence of the i-th feature, and n represents Z j The number of features in, log represents the natural logarithmic function, when p i The smaller it is, the larger the absolute value of log is, indicating that the feature is less likely to appear and carries more information. λ represents the regularization parameter, which is used to balance the weight between information entropy and feature deviation. It means to sum all items from 1 to m, where m represents The number of features in Z j represents the actual observed value of the jth feature in Z, It represents the expected value of the j-th feature, and q represents the exponential operation, which is used to adjust the calculation method of the deviation.
[0113] S4 uses the meta-learning mechanism and pure features to build a meta-learning model, inputs the pure features into the meta-learning model, and outputs a personalized prediction model through a sub-model that quickly adapts to the specific ship behavior pattern;
[0114] The meta-learning structure is set to consist of a shared base model and a series of sub-models for tasks. The base model can make preliminary predictions for any ship's track, while the sub-models are personalized models for specific ships and are used to fine-tune the predictions.
[0115] Set each ship as k, and then build a sub-model to quickly adapt to the specific behavior pattern of ship k through historical ship track data;
[0116] Based on the ship and the basic model P, an adaptability measurement function is constructed, and the expression is:
[0117]
[0118] Among them, R k (P,Z k ) indicates that ship k has pure feature Z k and the reliability measure under prediction P, where P represents the prediction result of the basic model, Z k represents the pure feature set of ship k, T represents the total number of time steps, d k (t) represents the deviation at time t, α k represents a constant, which is the ideal deviation of the model prediction, β k represents another constant that controls the nonlinearity of the effect of the deviation on the reliability measure. kThis means that the deviation has a greater penalty on the reliability measure, μ d represents the mean of the deviations, represents the variance of the deviation, N represents the normal distribution function, It means calculating the sum of probability densities of normal distribution in the entire deviation space;
[0119] Using the fitness metric function as the objective function R k , optimize the sub-model, the goal is to minimize R k , so that the sub-model can better adapt to the track pattern of ship k;
[0120] When the sub-model is optimized, that is, the personalized prediction model is optimized, the output vector is R k (t), used to predict the track of ship k at time t;
[0121] Fitness metric function R k The value range is from 0 to 1. k The value of is close to 1, which indicates that the basic model P has a high adaptability to the ship k and a good prediction effect;
[0122] When R k The value close to 0 indicates poor adaptability and large prediction error. k Approaching 1 can ensure the effectiveness of the personalized prediction model;
[0123] Normal distribution function, expression is:
[0124]
[0125] Among them, e represents the base of natural logarithm, and π represents pi.
[0126] S5 combines the personalized prediction model with real-time data, predicts the ship's track through the large model, and outputs the predicted track; the real-time data input vector at time t is set as X(t), which includes the sensor data, environmental conditions, navigation data, and communication data at time t;
[0127] Based on real-time data X(t) and R k (t), and at the same time combine the pure feature set Z of ship k k Construct a composite function G, the expression is:
[0128]
[0129] Among them, Y k (t) represents the predicted track of ship k at time t, P k (t) represents the output vector of the personalized prediction model of ship k at time t, w iIt is used to adjust the importance of each sensor data in the comprehensive prediction. ReLU represents the activation function. K represents the convolution kernel, which is used to process sensor data. B represents the matrix for adjusting the output of the personalized prediction model. C represents the matrix for adjusting the pure features. μ k represents the mean of real-time data, Indicates the variance of real-time data;
[0130] The value range of Yk(t) is from 0 to 1. When the value of Yk(t) is close to 0, it indicates that the uncertainty of the predicted track is large. When the value of Yk(t) is close to 1, it indicates that the confidence of the predicted track is high.
[0131] S6 establishes a safety decision support system to compare the predicted track with the navigation rules. When the predicted result exceeds the safety threshold, an alarm is triggered and safety recommendations are provided to the ship.
[0132] Set the navigation rule set to R, including but not limited to channel width, center line, minimum safety distance, maximum speed limit and prohibited area;
[0133] The safety assessment function is constructed based on the predicted trajectory of the ship, and the expression is:
[0134]
[0135] Among them, S(t) represents the output value of the security assessment function at time t, R i represents the i-th navigation rule, Y k (t) i represents the value of the i-th dimension of the k-th predicted trajectory at time t, n represents the number of navigation rules, γ represents an exponent used to adjust the degree of influence of the deviation value, a larger γ will make the larger deviation have a more significant impact on the overall safety assessment, τ represents the preset safety threshold, and min represents the minimum value, which is used to ensure that the output of the safety assessment function does not exceed the preset safety threshold τ;
[0136] The alarm trigger function A(t) is set based on the safety assessment function and the safety threshold. The expression is:
[0137]
[0138] This formula means that when the value of S(t) is greater than the preset safety threshold τ, A(t) returns 1 and triggers an alarm; otherwise, it returns 0 and does not trigger an alarm.
[0139] Design a security suggestion function to provide security suggestions SG k (t), the expression is:
[0140]
[0141] Among them, v crepresents the corrected velocity vector based on the predicted track deviation from the navigation rules, a a represents the adjusted acceleration vector based on the predicted track deviation from the navigation rule, α k and β k They represent the adjustment coefficients for correcting speed and adjusting acceleration, respectively, and adjust the strength of the suggestion according to the characteristics of the ship;
[0142] The value range of S(t) is [0,τ], where 0 means that the safety standard is fully met, and τ means that the safety threshold is just reached;
[0143] The value range of A(t) is 0 to 1, where 0 means that the safety threshold is not exceeded and no alarm needs to be triggered; 1 means that the predicted track exceeds the safety threshold and an alarm is triggered.
[0144] This embodiment also provides a ship track prediction system based on a large model, including:
[0145] The acquisition module is responsible for collecting multimodal data from various sources, including unstructured text descriptions, historical tracks, environmental conditions, radar images, and sonar signals, and preprocessing these data by denoising and filling missing values;
[0146] The feature module is responsible for extracting features from the preprocessed data, converting text data into numerical vectors, extracting the speed, direction, and position change rate of the track, performing spectrum analysis on environmental conditions, performing convolutional neural network processing on sensor data, and finally fusing all features through weighted averaging to form environmental perception features;
[0147] The detection module is responsible for building an anomaly detection model, identifying and filtering out abnormal data, building a meta-learning model using the meta-learning mechanism and pure features, quickly adapting to the behavior patterns of specific ships through sub-models, and outputting a personalized prediction model;
[0148] The prediction module is responsible for combining the personalized prediction model with real-time data, integrating sensor data, environmental conditions, navigation data and communication data using composite functions, predicting the ship's track through a large model, and outputting the predicted track and its confidence level;
[0149] The decision-making module is responsible for establishing a safe decision-making support system, comparing the predicted trajectory with the navigation rules, and calculating the safety level through the safety assessment function. When the predicted result exceeds the safety threshold, an alarm is triggered, and safety recommendations for correcting speed and adjusting acceleration are generated based on the degree of deviation to help ships avoid risks.
[0150] This embodiment also provides a computer device, which is suitable for the case of a ship track prediction method based on a large model, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the ship track prediction method based on a large model as proposed in the above embodiment.
[0151] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.
[0152] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for predicting a ship track based on a large model as proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, disk or optical disk.
[0153] In summary, the present invention collects multimodal data through multiple channels, uses advanced feature extraction technology to fuse environmental perception information, builds anomaly detection models to enhance data quality, and uses meta-learning mechanisms to quickly adapt to specific ship behaviors, ultimately achieving high-precision, real-time track prediction and safety decision-making assistance. The present invention focuses on solving the accuracy and safety issues of ship track prediction in complex marine environments, and aims to improve the intelligence level and safety of ship navigation.
[0154] Example 2, referring to Table 1, is the second example of the present invention. To further verify the advancement of the present invention, experimental simulation data of a ship track prediction method based on a large model are provided.
[0155] In order to verify the effectiveness and superiority of the ship track prediction method based on large model proposed in the present invention, a three-month offshore test was carried out. A total of six ships of different types were selected as test objects, including three frigates, two destroyers and one supply ship. During the experiment, the ship's track data, environmental condition data and sensor data were collected every day. The environmental condition data included wind speed, wind direction, current speed and direction, and sea water temperature; the sensor data included radar images and sonar signals. In addition, historical track data and unstructured text description data about each ship were also collected, such as the ship's performance parameters and past navigation logs.
[0156] In the preprocessing stage, the collected data were denoised and missing values were filled. The unstructured text description data was converted into numerical vectors through the BERT model. The historical track data was extracted using the sliding window technology to extract the rate of change of speed, direction and position. The environmental condition data was spectrally analyzed through discrete Fourier transform. The sensor data was feature extracted through convolutional neural network (CNN). Afterwards, the above features were fused to form an environmental perception feature vector.
[0157] Next, an anomaly detection model is built to identify and filter abnormal data through causal reasoning algorithms to ensure the accuracy of subsequent prediction models. The meta-learning mechanism is used to build a meta-learning model based on pure features to quickly adapt to the behavior patterns of each ship and output a personalized prediction model. Finally, the personalized prediction model is combined with real-time data, and the ship's track is predicted through a large model. A safety decision-making support system is established to evaluate the safety of the predicted track.
[0158] The details are shown in Table 1:
[0159] Table 1 Experimental record table
[0160]
[0161] Through the analysis of the above data, it can be clearly seen that the ship track prediction method of the present invention performs well in prediction accuracy, with an average accuracy of 92%, which is much higher than the traditional method of about 80%. The abnormal data filtering rate has also reached a satisfactory level, averaging 95%, which effectively improves the quality of the data and reduces the impact of noise on the prediction model. The feature extraction efficiency is 88% on average, indicating that this method can quickly and accurately extract useful information from multimodal data and provide rich input for the prediction model.
[0162] Prediction latency was kept at a low level, averaging 0.13 seconds, which is critical for real-time prediction and ensures the immediate availability of prediction results. The number of safety threshold exceedances reflects the effectiveness of the safety decision-making assistance system, with an average of only three exceedances per ship, and the safety recommendation execution rate was as high as 97%, indicating that the system can identify risks in a timely and accurate manner and provide effective avoidance measures.
[0163] In summary, the ship track prediction method based on the large model proposed in the present invention shows obvious advantages in prediction accuracy, data processing capability and safety. Compared with the existing technology, it not only improves the accuracy and real-time performance of the prediction, but also enhances the safety of ship navigation. It has important practical value and innovative significance.
[0164] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should be included in the scope of the claims of the present invention.
Claims
1. A ship track prediction method based on a large model, characterized by: include, Collect multimodal data through multiple channels; Extract features from multimodal data separately and output fused environmental perception features; An anomaly detection model is built based on the fused environmental perception features. The fused environmental perception feature vector is input into the anomaly detection model. The abnormal data is identified and filtered through the causal reasoning algorithm to output pure features. A meta-learning model is constructed using the meta-learning mechanism and pure features. Pure features are input into the meta-learning model, and a personalized prediction model is output through a sub-model that quickly adapts to the specific ship behavior pattern. Combine the personalized prediction model with real-time data, use the large model to predict the ship's track, and output the predicted track; A safety decision support system is established to compare the predicted track with the navigation rules. When the predicted result exceeds the safety threshold, an alarm is triggered and safety recommendations are provided to the ship.
2. The ship track prediction method based on a large model as claimed in claim 1, characterized in that: The specific steps of collecting multimodal data through multiple channels are as follows: Collect unstructured text description data, historical track data, and environmental condition data from multiple sources; Collect radar images and sonar signals through sensors; The collected data are preprocessed by denoising and filling missing values.
3. The ship track prediction method based on a large model as claimed in claim 2, characterized in that: The steps of extracting features from the multimodal data and outputting fused environmental perception features are as follows: The BERT model is used to convert unstructured text descriptions into numerical vectors to obtain text feature vectors, which are expressed as follows: T = BERT(wo); Where T represents the text features obtained after being processed by the BERT model, and wo represents the input text sequence; The sliding window technology is used to extract the change rate of the speed, direction and position of the track to obtain the track feature vector. The sliding window size is set to W, the step size of the sliding window is s, and the expression is: Among them, H represents the track characteristics, i represents the starting time point of the window, represents the summation operation starting from time point i and ending at time point i+W-1, v t represents the velocity vector at time t, θ t represents the direction angle at time t, x t represents the position coordinate at time t, x t+s -x t represents the displacement vector between time t and time t+s; Discrete Fourier transform is used to perform spectrum analysis on environmental condition data, and the meteorological data at time t is set as m t , the ocean current data at time t is o t , the characteristic expression of environmental conditions is extracted as follows: E = (DFT(m), DFT(o)); Where E represents the environmental condition characteristics, DFT represents discrete Fourier transform; Convolutional neural network CNN is used to extract features from sensor data. The two-dimensional matrix of sensor data is set to so, and the expression is: S = CNN(so) = ReLU(K*so); Among them, S represents sensor features, K represents CNN filter kernel, ReLU represents activation function, and * represents convolution operation; The extracted text features, track features, environmental condition features and sensor features are fused by weighted averaging, and the expression is: Z=a T T+a H H+a E E+a S S; α T +α H +α E +α S =1; Among them, Z represents the fused environmental perception features, α T , α H , α E and α S They represent the weights of text features, track features, environmental condition features, and sensor features respectively.
4. The ship track prediction method based on a large model as claimed in claim 3, characterized in that: The anomaly detection model is constructed based on the fused environment perception features, the fused environment perception feature vector is input into the anomaly detection model, the abnormal data is identified and filtered through the causal reasoning algorithm, and the pure features are output. The specific steps are as follows: The anomaly detection model is constructed by using multivariate Gaussian distribution, information entropy, and the latent causal variable model LCM in causal reasoning. The expression of the anomaly detection model is: Among them, A(Z) represents the output of the anomaly detection model, represents the integral symbol, μ represents the mean vector, Σ represents the covariance matrix, N(Z; μ, Σ) represents the multivariate Gaussian distribution function, ι(Z) represents the information filtering function, exp(-ι(Z)) represents the exponential function, and du is the differential element, which represents a small change in the integral variable μ; The expression of information filtering function ι(Z) is: Among them, p i represents the probability of the occurrence of the i-th feature, and n represents Z j The number of features in, log represents the natural logarithmic function, when p i The smaller it is, the larger the absolute value of log is. λ represents the regularization parameter. It means to sum all items from 1 to m, where m represents The number of features in Z j represents the actual observed value of the jth feature in Z, represents the expected value of the j-th feature, and q represents the exponential operation.
5. The ship track prediction method based on a large model as claimed in claim 4, characterized in that: The meta-learning model is constructed by using the meta-learning mechanism and pure features, the pure features are input into the meta-learning model, and a personalized prediction model is output through a sub-model that quickly adapts to the specific ship behavior mode. The specific steps are as follows: The meta-learning structure is assumed to consist of a shared base model and a series of sub-models for each task; Set each ship as k, and then build a sub-model to quickly adapt to the specific behavior pattern of ship k through historical ship track data; Based on the ship and the basic model P, an adaptability measurement function is constructed, and the expression is: Among them, R k (P,Z k ) indicates that ship k has pure feature Z k and the reliability measure under prediction P, where P represents the prediction result of the basic model, Z k represents the pure feature set of ship k, T represents the total number of time steps, d k (t) represents the deviation at time t, α k represents a constant, β k represents another constant, μ d represents the mean of the deviations, represents the variance of the deviation, N represents the normal distribution function, It means calculating the sum of probability densities of normal distribution in the entire deviation space; Using the fitness metric function as the objective function R k , optimize the sub-model; When the sub-model is optimized, that is, the personalized prediction model is optimized, the output vector is R k (t).
6. The ship track prediction method based on a large model as claimed in claim 5, characterized in that: The personalized prediction model is combined with real-time data, the ship track is predicted through a large model, and the predicted track is output. The specific steps are as follows: Assume that the real-time data input vector at time t is X(t); Based on real-time data X(t) and R k (t), and at the same time combine the pure feature set Z of ship k k Construct a composite function G, the expression is: Among them, Y k (t) represents the predicted track of ship k at time t, P k (t) represents the output vector of the personalized prediction model of ship k at time t, w i It is used to adjust the importance of each sensor data in the comprehensive prediction. ReLU represents the activation function, K represents the convolution kernel, B represents the matrix for adjusting the output of the personalized prediction model, C represents the matrix for adjusting the pure features, and μ k represents the mean of real-time data, Represents the variance of real-time data.
7. The ship track prediction method based on a large model as claimed in claim 6, characterized in that: The safety decision support system is established to compare the predicted track with the navigation rules. When the predicted result exceeds the safety threshold, an alarm is triggered and safety recommendations are provided to the ship. The specific steps are as follows: Set the navigation rule set to R; The safety assessment function is constructed based on the predicted trajectory of the ship, and the expression is: Among them, S(t) represents the output value of the security assessment function at time t, R i represents the i-th navigation rule, Y k (t) i represents the value of the i-th dimension of the k-th predicted trajectory at time t, n represents the number of navigation rules, γ represents the exponent, τ represents the preset safety threshold, and min represents the minimum value; The alarm trigger function A(t) is set based on the safety assessment function and the safety threshold. The expression is: This formula indicates that when the value of S(t) is greater than the preset safety threshold τ, A(t) returns 1, indicating that an alarm needs to be triggered; otherwise, it returns 0, indicating that an alarm is not triggered; Design a security suggestion function to provide security suggestions SG k (t), the expression is: Among them, v c represents the corrected velocity vector based on the predicted track deviation from the navigation rules, a a represents the adjusted acceleration vector based on the predicted track deviation from the navigation rule, α k and β k They represent the adjustment coefficients for correcting speed and adjusting acceleration respectively.
8. A ship track prediction system based on a large model, based on the ship track prediction method based on a large model according to any one of claims 1 to 7, characterized in that: include: The acquisition module is responsible for collecting multimodal data from various sources, including unstructured text descriptions, historical tracks, environmental conditions, radar images, and sonar signals, and preprocessing these data by denoising and filling missing values; The feature module is responsible for extracting features from the preprocessed data, converting text data into numerical vectors, extracting the speed, direction, and position change rate of the track, performing spectrum analysis on environmental conditions, performing convolutional neural network processing on sensor data, and finally fusing all features through weighted averaging to form environmental perception features; The detection module is responsible for building an anomaly detection model, identifying and filtering out abnormal data, building a meta-learning model using the meta-learning mechanism and pure features, quickly adapting to the behavior patterns of specific ships through sub-models, and outputting a personalized prediction model; The prediction module is responsible for combining the personalized prediction model with real-time data, integrating sensor data, environmental conditions, navigation data and communication data using composite functions, predicting the ship's track through a large model, and outputting the predicted track and its confidence level; The decision-making module is responsible for establishing a safe decision-making support system, comparing the predicted trajectory with the navigation rules, and calculating the safety level through the safety assessment function. When the predicted result exceeds the safety threshold, an alarm is triggered, and safety recommendations for correcting speed and adjusting acceleration are generated based on the degree of deviation to help ships avoid risks.
9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the ship track prediction method based on a large model as described in any one of claims 1 to 7 are implemented.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the ship track prediction method based on a large model as described in any one of claims 1 to 7 are implemented.
Citation Information
Patent Citations
Intelligent ship identification method and system based on multi-modal feature fusion
CN115995032A
Ship trajectory multi-modal prediction method and system
CN117520874A
Multi-modal track prediction method and device, and electronic equipment
CN117782100A
Multimodal unsupervised meta-learning method and apparatus
US20240232648A1
Cited By
Maritime behavior intention prediction method based on dynamic decision tree
CN119719968A
Ship speed missing data complementing method, system and device and medium
CN119884623A
Ship operation efficiency intelligent control system and method based on large model optimization
CN119916692A
Ship navigation hidden danger identification method and system in narrow water area, storage medium and electronic equipment
CN122132929A
Large model-based ship information research and judgment analysis method and platform
CN122173887A
