Ship target classification method, ship target model construction method, ship target classification system, ship target model construction equipment and storage medium

Abstract

The embodiment of the invention provides a ship target classification method, a model construction method, a system, equipment and a storage medium, and belongs to the technical field of artificial intelligence. The method comprises the steps of collecting ship navigation data of each ship navigation task, performing feature extraction on the multiple pieces of ship navigation data to obtain multiple pieces of ship feature data, dividing the multiple pieces of ship feature data into different navigation scenes according to the ship behavior features, and determining the ship navigation tasks according to the ship navigation scenes. And constructing a training data set in the corresponding navigation scene according to the ship feature data in each navigation scene, and constructing a ship target classification model in the corresponding navigation scene according to the training data set. According to the method, the navigation scene to which the ship navigation data belongs is identified through the ship behavior characteristics, and the corresponding classification models are constructed based on the ship navigation data in different navigation scenes, so that the accuracy of ship target classification is improved.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular, to a method for classifying ship targets, a method for constructing a model, a system, a device, and a storage medium. Background Art

[0002] Ship target classification and recognition is based on AIS data and models according to the historical AIS data of real ships, extracts various ship features, and imports the extracted features into a classification model to achieve accurate classification and recognition of ship types or behavior models and other targets. Due to the different water area scenarios where ships are located, their ship behaviors will show different behavioral characteristics. However, the current ship target classification methods do not consider different behavioral characteristics in the ship navigation scenario and uniformly perform ship target classification modeling, resulting in a low classification accuracy of the ship target classification model. Summary of the Invention

[0003] The main purpose of the embodiments of this application is to propose a method for classifying ship targets, a method for constructing a model, a system, a device, and a storage medium, aiming to improve the classification accuracy of the ship target classification model.

[0004] To achieve the above object, on the one hand, an embodiment of this application proposes a method for constructing a ship target classification model, including the following steps:

[0005] Collect ship navigation data for each ship navigation task;

[0006] Extract features from multiple pieces of the ship navigation data respectively to obtain multiple ship feature data, where the ship feature data includes ship behavior features and ship static features;

[0007] Divide the multiple pieces of ship feature data into different navigation scenarios according to the ship behavior features;

[0008] Construct a training data set corresponding to the navigation scenario according to the ship feature data in each navigation scenario;

[0009] Construct a ship target classification model corresponding to the navigation scenario according to the training data set.

[0010] In some embodiments, the ship behavior features include ship dynamic behavior state features and behavior state change features; the ship feature data is subjected to feature extraction through the following steps:

[0011] Screen out ship static data and ship dynamic data from the ship navigation data;

[0012] Perform static feature extraction according to the ship static data to obtain ship static features;

[0013] Extract dynamic features from the ship dynamic data to obtain the ship dynamic behavior state features;

[0014] Analyze the characteristic changes based on the ship dynamic behavior state features to obtain the behavior state change features.

[0015] In some embodiments, each of the ship feature data carries the environmental features corresponding to the ship navigation tasks. The step of dividing the multiple ship feature data into different navigation scenarios according to the ship behavior features includes the following steps:

[0016] Screen each of the ship feature data according to the data screening conditions, where the data screening conditions include static feature screening conditions, dynamic feature screening conditions, and environmental screening conditions;

[0017] Divide the multiple ship feature data obtained after screening into different navigation scenarios according to the ship behavior features.

[0018] In some embodiments, the ship behavior features include the course range and the navigation path. The step of dividing the multiple ship feature data obtained after screening into different navigation scenarios according to the ship behavior features includes the following steps:

[0019] Determine the division parameter values of the ship feature data according to the course range and the number of turning nodes of the navigation path in the ship feature data, where the division parameter values are directly proportional to the course range and inversely proportional to the number of turning nodes;

[0020] When the division parameter value is less than the expected value, divide the corresponding ship feature data into the navigation scenario of the restricted sea area;

[0021] When the division parameter value is greater than the expected value, divide the corresponding ship feature data into the navigation scenario of the open sea area.

[0022] In some embodiments, the step of constructing a ship target classification model for the corresponding navigation scenario according to the training data set includes the following steps:

[0023] Calculate the maximum information coefficient of each ship behavior feature and the target classification according to the training data set in the navigation scenario of the open sea area;

[0024] Select several ship behavior features with a greater contribution to the target classification from the multiple ship behavior features according to the maximum information coefficient;

[0025] Perform target classification modeling according to the selected several ship behavior features to obtain a ship target classification model in the navigation scenario of the open sea area.

[0026] In some embodiments, constructing a ship target classification model for a corresponding navigation scenario based on the training dataset includes the following steps:

[0027] Calculating the degree of association between each ship static feature and ship behavior feature according to the training dataset in the navigation scenario of restricted waters;

[0028] Selecting the ship static feature with the strongest correlation with the ship behavior feature according to the degree of association;

[0029] Performing target classification modeling according to the ship behavior feature and the selected ship static feature to obtain a ship target classification model for the navigation scenario of restricted waters.

[0030] To achieve the above object, another aspect of the embodiments of the present application proposes a ship target classification method, including the following steps:

[0031] Obtaining the ship navigation data of the target ship;

[0032] Performing feature extraction on the ship navigation data to obtain ship feature data, where the ship feature data includes ship behavior features and ship static features;

[0033] Determining the target navigation scenario corresponding to the ship navigation data according to the ship behavior feature;

[0034] Determining the classification result of the target ship according to the ship behavior feature through the ship target classification model of the target navigation scenario.

[0035] To achieve the above object, another aspect of the embodiments of the present application proposes a ship target classification model construction system, including:

[0036] A first module for collecting the ship navigation data of each ship navigation task;

[0037] A second module for respectively performing feature extraction on multiple pieces of the ship navigation data to obtain multiple pieces of ship feature data, where the ship feature data includes ship behavior features and ship static features;

[0038] A third module for dividing multiple pieces of the ship feature data into different navigation scenarios according to the ship behavior feature;

[0039] A fourth module for constructing a training dataset corresponding to the navigation scenario according to the ship feature data in each navigation scenario;

[0040] A fifth module for constructing a ship target classification model for the corresponding navigation scenario according to the training dataset.

[0041] To achieve the above object, on the other hand, an embodiment of the present application provides an electronic device, which includes a memory, a processor, a program stored on the memory and executable on the processor, and a data bus for realizing connection communication between the processor and the memory. When the program is executed by the processor, the method described in the above embodiment is implemented.

[0042] To achieve the above object, on the other hand, an embodiment of the present application provides a storage medium, which is a computer-readable storage medium for computer-readable storage. The storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the method described in the above embodiment.

[0043] The ship target classification method, model construction method, system, device and storage medium provided by the present application collect ship navigation data of each ship navigation task, respectively extract features from multiple ship navigation data to obtain multiple ship feature data. The ship feature data includes ship behavior features and ship static features. Then, according to the ship behavior features, the multiple ship feature data are divided into different navigation scenarios, and then a training data set corresponding to the navigation scenario is constructed according to the ship feature data in each navigation scenario. A ship target classification model corresponding to the navigation scenario is constructed according to the training data set. The present application identifies the navigation scenario to which the ship navigation data belongs through the ship behavior features, and then constructs corresponding classification models respectively based on the ship navigation data under different navigation scenarios, thereby improving the accuracy of ship target classification. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 is a flowchart of the ship target classification model construction method provided by an embodiment of the present application;

[0045] Figure 2 is a flowchart of the ship target classification method provided by an embodiment of the present application;

[0046] Figure 3 is a schematic diagram of the ship target classification model construction system provided by an embodiment of the present application;

[0047] Figure 4 is a schematic diagram of the hardware structure of the electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0048] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, and are not used to limit the present application.

[0049] It should be noted that although the functional modules are divided in the system and the logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order from the module division in the system or the order in the flowchart. The terms "first", "second", etc. in the specification, claims and the above drawings are used to distinguish similar objects and do not necessarily describe a specific order or sequence.

[0050] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0051] First, the following explanations are given for the nouns involved in the embodiments of this application:

[0052] AIS (Automatic Identification System) is a technical system for ship tracking and identification. It automatically sends and receives information related to the static state, dynamic state and voyage of ships through the VHF radio frequency band, helping to exchange information between ships and between ships and shore-based facilities to improve navigation safety, efficiency and regulatory capabilities.

[0053] The embodiments of this application provide a ship target classification method, a model construction method, a system, a device and a storage medium, aiming to improve the classification accuracy of the ship target classification model.

[0054] The ship target classification method, model construction method, system, device and storage medium provided by the embodiments of this application are specifically described through the following embodiments. First, the ship target classification model construction method and ship target classification method in the embodiments of this application are described.

[0055] The ship target classification model construction method and ship target classification method provided by the embodiments of this application relate to the field of artificial intelligence technology. The ship target classification model construction method or ship target classification method provided by the embodiments of this application can be applied to a terminal, can also be applied to a server side, or can also be software running on a terminal or a server side. In some embodiments, the terminal can be a smart phone, a tablet computer, a notebook computer, a desktop computer, etc.; the server side can be configured as an independent physical server, can also be configured as a server cluster or a distributed system composed of multiple physical servers, or can also be configured as a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application implementing the ship target classification model construction method or ship target classification method, etc., but is not limited to the above forms.

[0056] This application can be used in numerous general or specific computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, and so on. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. This application can also be practiced in a distributed computing environment where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0057] Figure 1 is an alternative flowchart of the method for constructing a ship target classification model provided by an embodiment of this application. Figure 1 The method in may include but is not limited to steps S101 to S105.

[0058] Step S101, collect ship navigation data for each ship navigation task;

[0059] Step S102, respectively perform feature extraction on multiple ship navigation data to obtain multiple ship feature data, where the ship feature data includes ship behavior features and ship static features;

[0060] Step S103, divide the multiple ship feature data into different navigation scenarios according to the ship behavior features;

[0061] Step S104, construct a training data set corresponding to the navigation scenario according to the ship feature data in each navigation scenario;

[0062] Step S105, construct a ship target classification model corresponding to the navigation scenario according to the training data set.

[0063] Steps S101 to S105 illustrated in the embodiments of this application identify the navigation scenario to which the ship navigation data belongs through ship behavior features, and then construct corresponding classification models respectively based on the ship navigation data in different navigation scenarios, solving the problems that ship behavior is not considered during ship classification in actual processes and that ship behaviors in navigation scenarios such as restricted waters and open waters are not distinguished, effectively improving ship classification accuracy and realizing accurate classification and recognition of ship targets based on ship behavior in different types of waters.

[0064] In step S101 of some embodiments, the route of the ship in each navigation scenario is defined as a ship navigation task. Combining with the AIS of the Automatic Identification System, the ship navigation data of the ship navigation task can be collected. The ship navigation data includes static data and dynamic data. The static data includes data such as the ship's length and the ship's weight, and the dynamic data includes speed, course, acceleration, etc.

[0065] Further, after obtaining the ship navigation data, data preprocessing operations can be performed on the ship navigation data to improve the usability of the data. The data preprocessing operations include but are not limited to data cleaning module operations, data completion operations, and data segmentation operations. The data cleaning operation is used for deleting duplicate values, outliers, error values, and data repair. The data completion operation is used for interpolation processing of missing values to complete the AIS data. The data segmentation operation is used to segment the data based on the data time and the ship's motion state, so as to delete ships in a non-moving state. The above operations are part of the data preprocessing operations, and are used to screen out the ship navigation data that is not affected by external environmental factors from the original AIS data. Through the data preprocessing operations, the AIS data in navigation scenarios such as open waters and restricted waters can be well processed, laying a good foundation for the subsequent ship classification.

[0066] In step S102 of some embodiments, feature extraction includes static feature extraction and dynamic feature extraction. By performing static feature extraction and dynamic feature extraction on the ship navigation data, the ship behavior features and ship static features of the ship navigation data are correspondingly obtained. The ship behavior features and ship static features together form the ship feature data of the ship navigation data. The ship static features include the static data in the ship navigation data and the data obtained by performing data analysis based on the static data, including but not limited to the draft obtained based on the weight and dimensions, etc. The ship static features include the dynamic data in the ship navigation data and the data obtained by performing data analysis based on the dynamic data, including but not limited to the distance from obstacles obtained based on the position information, etc.

[0067] According to some embodiments of the present application, the ship behavior features include ship dynamic behavior state features and behavior state change features. In step S102, the ship feature data can be subjected to feature extraction through but not limited to the following steps:

[0068] Step S201, screening out the ship static data and ship dynamic data from the ship navigation data;

[0069] Step S202, performing static feature extraction according to the ship static data to obtain ship static features;

[0070] Step S203, performing dynamic feature extraction according to the ship dynamic data to obtain ship dynamic behavior state features;

[0071] Step S204: Analyze the characteristic changes based on the ship's dynamic behavior state characteristics to obtain the characteristic changes of the behavior state.

[0072] In this embodiment, the static feature extraction operation extracts the static features of the ship by analyzing the static data in the ship's navigation data, such as the ship length, ship width, ship draft, and aspect ratio.

[0073] In this embodiment, the ship behavior characteristics include the ship's dynamic behavior state characteristics and the characteristic changes of the behavior state. For the extraction of the ship behavior characteristics, first, the ship's dynamic behavior state is extracted based on the ship's navigation data, such as the heading, speed over ground, course over ground, position information of the ship, and density distribution. On this basis, the analysis of the changes in the ship's dynamic behavior characteristics is carried out to obtain the characteristic changes of the behavior state. Exemplarily, the analysis formula used for extracting the characteristic changes of the behavior state of distance is as follows:

[0074]

[0075] |LON_SPAN| = |LON MAX -LON MIN |;

[0076] |LAT_SPAN| = |LAT MAX -LAT MIN |;

[0077] where d represents the distance between two adjacent track points in the ship's navigation data; r is the radius of the earth, taking 6371.393 km; Δγ′ = |γ′1 - γ′2|, representing the positions of the 1st and 2nd track points, in rad; LON MAX and LAT MAX are the maximum values of longitude and latitude in the data, and LON MIN and LAT MIN are the minimum values of longitude and latitude in the data.

[0078] In step S103 of some embodiments, the different navigation scenarios of the embodiments of the present application may include an open water navigation scenario and a restricted water navigation scenario. Taking the ship behavior characteristics (such as the navigation path, course range, and distance) as the analysis object, thus determining whether each ship navigation task is in the open water navigation scenario or the restricted water navigation scenario, and then classifying the corresponding ship feature data into the corresponding navigation scenario.

[0079] According to some embodiments of the present application, each ship feature data carries the environmental characteristics of the corresponding ship navigation task. The step of classifying multiple ship feature data into different navigation scenarios according to the ship behavior characteristics in step S103 may include, but is not limited to, the following steps:

[0080] Step S301: Screen each piece of ship feature data according to data screening conditions, where the data screening conditions include static feature screening conditions, dynamic feature screening conditions, and environmental screening conditions;

[0081] Step S302: Divide the multiple pieces of ship feature data obtained after screening into different navigation scenarios according to ship behavior characteristics.

[0082] In this embodiment, after extracting various features from ship navigation data, the extracted features and corresponding environmental features can be used as screening conditions to filter out ship feature data that do not meet the screening conditions, and then divide the ship feature data that meets the data screening conditions into navigation scenarios, thereby improving the accuracy of subsequent model construction.

[0083] The data screening conditions can be set according to actual analysis requirements. Exemplarily, the data screening conditions can be at least one of static feature screening conditions, dynamic feature screening conditions, and environmental screening conditions. The static feature screening condition is used to check whether the ship static features in the ship feature data, such as the ship length L, width W, aspect ratio Aspect Ratio = L / W, etc., are reliable. For example, when the ship length in a piece of ship feature data does not meet the first preset screening condition, this piece of ship feature data does not meet the data screening conditions. The dynamic feature screening condition is used to check whether the ship dynamic features (such as heading) in the ship feature data are reliable. The environmental screening condition is used to check whether the environmental description and water area characteristics corresponding to the ship feature data meet the requirements. The environmental screening can specifically include the following aspects: setting a wind speed threshold (such as 8 m / s) to screen out ship feature data with a wind speed lower than this threshold, so as to exclude the influence of wind speed on ship behavior; setting a flow velocity threshold (such as 0.37 m / s) to screen out ship feature data with a flow velocity lower than this threshold, so as to exclude the influence of water flow on ship behavior; setting a visibility threshold (such as 2000 m) to screen out ship feature data with a visibility higher than this threshold to exclude the influence of visibility on ship behavior.

[0084] According to some embodiments of the present application, the ship behavior characteristics include heading range and navigation path. In step S302, the step of dividing the multiple pieces of ship feature data obtained after screening into different navigation scenarios according to ship behavior characteristics may include, but is not limited to, the following steps:

[0085] Step S401: Determine the division parameter value of the ship feature data according to the heading range and the number of turning nodes of the navigation path in the ship feature data, where the division parameter value is directly proportional to the heading range and inversely proportional to the number of turning nodes;

[0086] Step S402: When the division parameter value is less than the expected value, the corresponding ship feature data is divided into the navigation scenario of the restricted sea area;

[0087] Step S403: When the division parameter value is greater than the expected value, the corresponding ship feature data is divided into the navigation scenario of the open sea area.

[0088] In this embodiment, considering that the behavior characteristics of ships are different in different sea areas, this embodiment can select obvious behavior characteristics to judge the sea area where the ship is located. This embodiment takes into account that when sailing in open sea areas without obstacles or when leaving the port, the ship shows a larger course range, and since there is no need to frequently avoid obstacles, the navigation path is relatively smooth. While when sailing in restricted sea areas with obstacles or near the port, the ship shows a smaller course range, and since it needs to frequently avoid obstacles, the navigation path is relatively tortuous. Based on this, the course range and the number of turning nodes of the navigation path can be considered to judge the corresponding navigation scenario. Further, the course range and the number of turning nodes can be input into the fitted mapping formula to obtain the division parameter value. In this mapping formula, the output division parameter value is proportional to the input course range and inversely proportional to the number of turning nodes. The formula also includes the coefficient of the course range and the coefficient of the number of turning nodes, and the coefficients can be obtained by machine learning fitting.

[0089] It can be understood that in the navigation path, when the steering angle of the steering operation is greater than the preset value, it is recorded as the number of one node.

[0090] In step S104 of some embodiments, after dividing multiple ship feature data into the navigation scenarios of the open sea area and the restricted sea area respectively, classification labels can be added to the ship feature data in different navigation scenarios according to the actual classification target. For example, if the classification target is to identify the ship type, the classification labels are oil tanker, cargo ship, etc.; if the classification target is the ship behavior mode, the classification labels are obstacle avoidance mode, detection mode, etc. It can be understood that labels for multiple classification targets can be added to the data.

[0091] In step S105 of some embodiments, a ship target classification model for the corresponding navigation scenario is constructed according to the training data set. Exemplarily, the embodiments of the present application can construct a ship target classification model under the open sea area and a ship target classification model under the restricted sea area. The ship target classification models for different navigation scenarios can adopt the same classification algorithm or different classification algorithms, which can be selected according to the actual analysis requirements.

[0092] According to some embodiments of the present application, for the target classification model of the open sea area, it can be constructed by including but not limited to the following steps:

[0093] Step S501: Calculate the maximum information coefficient of each ship behavior feature and the target classification based on the training dataset in the navigation scenario of open sea areas.

[0094] Step S502: Select several ship behavior features with a relatively large contribution to the target classification from multiple ship behavior features according to the maximum information coefficient.

[0095] Step S503: Perform target classification modeling based on the selected several ship behavior features to obtain a ship target classification model in the navigation scenario of open sea areas.

[0096] In this embodiment, after calculating the maximum information coefficient of features and classification, the importance selection of ship behavior features can be performed to determine the ship behavior features with a relatively large contribution to the ship classification task. Specifically, the maximum information coefficient (MIC) algorithm is used to measure the correlation degree between each behavior feature and the ship class. The value range of MIC is between 0 and 1, and the larger the value, the stronger the correlation. By calculating the maximum information coefficient of each feature, the ranking of the contribution of ship behavior features to ship classification is obtained, so as to screen out the features with a relatively high contribution to classification and eliminate the features with a small contribution to classification, which helps to speed up the classification speed and reduce overfitting, simplify the calculation complexity, and on this basis, construct and train the classification model.

[0097] Exemplarily, according to the MIC algorithm, calculate the MIC value of each ship feature. Each ship feature value is X, and the ship class is Y. The mutual information I(X; Y) measures the dependence relationship between the two. The specific calculation formula is as follows:

[0098]

[0099] Among them, p(x,y) is the joint distribution, and p(x) and p(y) are the marginal distributions.

[0100] Then perform normalization processing on all features, select the maximum value after normalization as the MIC, and finally sort in descending order according to the MIC value, and select the top k features to form a new feature set for ship classification. The specific calculation formula is as follows:

[0101]

[0102] Among them, I * is the current maximum mutual information coefficient, B is the upper limit value of the current x×y partition, n is the total amount of data, and B(n)=n 0.6 Restrict the total number of calculations to avoid overfitting.

[0103] After determining several ship behavior characteristics with greater contributions, using these several ship behavior characteristics as model inputs and ship classification as model outputs, train a ship target classification model based on the training data set in open sea areas. The ship target classification model can be constructed based on BP-AdaBoost, and the classification formula applied is:

[0104]

[0105] where Sign represents the sign function, which is used to convert the result of the weighted sum into the final classification decision; M is the total number of base classifiers; a n is the weight of the nth base classifier, indicating the importance of this classifier; G n (x) is the classification result generated by the nth base classifier for the input x.

[0106] Use the training set data to train the model, and the test set data can be used for classification prediction to verify the model.

[0107] According to some embodiments of the present application, for the target classification model in restricted sea areas, it can be constructed by including but not limited to the following steps:

[0108] Step S601, calculate the correlation degree between each ship static feature and ship behavior feature according to the training data set in the navigation scenario of the restricted sea area;

[0109] Step S602, select the ship static feature with the strongest correlation with the ship behavior feature according to the correlation degree;

[0110] Step S603, perform target classification modeling according to the ship behavior feature and the selected ship static feature to obtain a ship target classification model in the navigation scenario of the restricted sea area.

[0111] In this embodiment, after obtaining the training data set in the navigation scenario of the restricted sea area, the feature analysis of the ship feature data can be performed first, and then the target classification modeling can be performed according to the feature analysis result to obtain a ship target classification model in the navigation scenario of the restricted sea area. The feature analysis includes the correlation analysis between static features and ship behavior features and the clustering analysis of ship behavior features.

[0112] The clustering analysis of ship behavior characteristics uses the k-means clustering algorithm. Taking ship speed and trajectory as the main ship behavior attributes, it conducts clustering analysis on ship behavior to identify different ship behavior patterns. The clustering analysis mainly includes cluster confirmation operation, initial cluster center initialization operation, and clustering stop operation. Specifically, the cluster confirmation operation gradually increases the number of clusters and uses a statistical t-test to determine the number of clusters to ensure that the ship behavior data in different clusters are significantly different in all cross-sections; the initial cluster center initialization operation defines the initial cluster center as the minimum value, maximum value, and other percentile values of the average value of ship behavior characteristics in all cross-sections to avoid the influence of random initialization; the clustering stop operation is to stop clustering when the maximum change in the cluster center is less than the set threshold (position 0.1 m, SOG 0.01 knot), and it is considered that the clusters at this time represent the ship behavior patterns in the area. Through the above operations, the confirmation of the number of ship behavior clusters, the calculation of the distance between ship behavior and the cluster center, and the clustering stop criterion can be completed. The cluster center of the ship behavior is expressed as follows:

[0113]

[0114] Among them, B n(j) represents the behavior attribute on the cross-section j of the trajectory, and B c (i,j) represents the center of cluster i of this behavior attribute on cross-section j, and m is the total number of cross-sections.

[0115] The correlation analysis between static characteristics and ship behavior characteristics mainly analyzes AIS data, screens out ship behavior data under unobstructed conditions, and based on ship static characteristics, analyzes the degree of correlation between them and ship behavior characteristics.

[0116] For the ship characteristic data under different ship behavior patterns, the ship behavior characteristics among them are discretized. For example, characteristic data such as ship speed and trajectory position are discretized according to the longitudinal position of the ship. By analyzing AIS data, screening out ship behavior data under unobstructed conditions, and based on ship static characteristics, the degree of correlation between them and ship behavior characteristics is analyzed. Then, combined with the analysis results of the correlation between ship static characteristics and ship behavior, the ship static characteristics with the strongest correlation with ship behavior characteristics are selected, and the Naive Bayes algorithm is used to construct a classification model to realize the classification and recognition of ship behavior based on ship static characteristics in restricted waters. Among them, the classification formula applied by the Naive Bayes ship recognition classification model is:

[0117]

[0118] Among them, X represents ship characteristics, and C i represents ship behavior clustering.

[0119] In some embodiments of the present application, after constructing the ship target classification models for different navigation scenarios, the classification effects of each ship target classification model can be verified. The description of the classification effect is as follows:

[0120]

[0121] Among them, TP, FP, FN, and TN respectively represent true positive, false positive, false negative, and true negative. Accuracy is the accuracy rate, Precision is the classification accuracy of a single type of ship, Recall is the recall rate, and F1-Score is the F1 score.

[0122] The specific formula for verifying the classification effect is:

[0123]

[0124] Among them, k is the number of classes; A ij is the number of data objects in the i-th interval, the j-th class; E ij is the expected frequency of A ij and E ij = R i * C j / N; R i is the number of data objects in the i-th interval C j is the number of data objects in the j-th class N is the total number of data objects

[0125]

[0126] Among them, N is the number of samples, y i is the true label (0 or 1) of the i-th sample, is the probability that the model predicts the i-th sample as the positive class.

[0127] According to some embodiments of the present application, the ship target classification models for open waters and restricted waters are specifically described as follows:

[0128] S1. Data processing.

[0129] Process the collected AIS navigation data, including data segmentation, anomaly repair, key value processing, detection of duplicate values, etc., and screen the wind speed, flow rate, visibility, and ship encounter situation that meet the conditions through the data, laying a good foundation for the subsequent ship classification.

[0130] S2. Feature extraction.

[0131] Ship feature extraction is mainly divided into ship static feature extraction and ship behavior feature extraction. Ship features (including static features, heading, speed, navigation distance, path, etc.) are extracted to assist in ship behavior analysis and the construction of target recognition and classification models. For ship classification in open waters, since the navigation range of ships is not restricted by water areas, the trajectory characteristics in ship behavior need to consider their distance characteristics. Therefore, in addition to static feature extraction, ship behavior features include the extraction of heading and speed features, as well as distance feature extraction; in restricted waters, ship behavior is restricted by the boundaries of navigable waters, and the heading can only be adjusted within a limited range. Therefore, the ship behavior features required for ship classification can only include path features and speed features.

[0132] S3. Classification and screening.

[0133] Based on ship behavior features, taking the heading range, navigation distance, and path as the basis, it can be judged whether the ship's navigation waters are open waters or restricted waters. Then, according to the maximum information coefficient of the ship, the importance of ship features for classification is determined, so as to understand and compare the contribution degrees of each feature in classification and recognition, eliminate the features with small contribution degrees to classification, speed up the classification speed and reduce overfitting. Finally, the important ship features with large contribution degrees are selected.

[0134] S4. Ship behavior analysis and target recognition and classification in open waters.

[0135] Based on AIS data, ship feature extraction, and the selection of ship feature importance, ship behavior analysis and target classification and recognition in open waters combine the BP neural network and the AdaBoost algorithm to construct a BP-AdaBoost classification model, and then use AIS data to conduct classification experiments on ship types, including cargo ships, oil tankers, fishing boats, passenger ships, etc.

[0136] S5. Ship behavior analysis and target recognition and classification in restricted waters.

[0137] Ship behavior analysis and target classification and recognition in restricted waters use the K-means clustering algorithm to conduct clustering analysis on the ship behavior in restricted waters, mainly including its speed and trajectory distribution position, to identify different ship behavior patterns, including speed intervals, lateral positions of ships in the channel, speed and trajectory change characteristics, etc. On this basis, analyze the laws and characteristics of ship static features under different behavior patterns, and based on the static features of the ship (such as ship length, ship width, draft, etc.), select the static features with the strongest correlation with ship behavior features, develop a naive Bayes classifier, judge and predict the ship's behavior pattern, so as to realize the identification of ships in restricted waters and the prediction of their behavior pattern categories.

[0138] S6. Verification of classification effect.

[0139] Through various verification methods such as accuracy rate, classification accuracy, recall rate, F1-score, cross-entropy verification, and chi-square verification, the ship classification effect is described in multiple aspects, and high-precision ship target classification and recognition are achieved.

[0140] The ship target classification and recognition method in this embodiment combines the characteristics of the water area and considers ship behavior. Through the extraction of ship behavior characteristics, the differences between ship behavior patterns among various types of ships can be characterized with the cooperation of other ship characteristics, and the characteristics of the water area can be reflected according to the feedback of ship behavior characteristics. An appropriate classifier is automatically selected for data screening, and various types of ships in open waters and ships entering and leaving ports in restricted waters are accurately recognized through the ship classification and recognition module. Finally, through the verification module, through various verification methods such as cross-entropy verification and chi-square verification, the ship classification effect is described in multiple aspects, and high-precision ship target classification and recognition are achieved.

[0141] Please refer to Figure 2 , this application embodiment also proposes a ship target classification method, which may include but is not limited to the following steps:

[0142] Step S701, obtaining the ship navigation data of the target ship;

[0143] Step S702, performing feature extraction on the ship navigation data to obtain ship feature data, where the ship feature data includes ship behavior features and ship static features;

[0144] Step S703, determining the target navigation scenario corresponding to the ship navigation data according to the ship behavior features;

[0145] Step S704, through the ship target classification model of the target navigation scenario, determining the classification result of the target ship according to the ship behavior features.

[0146] In this embodiment, the target ship is the ship that needs to be classified and recognized. When obtaining the ship navigation data of the target ship, through the same data preprocessing process, feature extraction process, and navigation scenario division process based on ship behavior features above, the navigation scenario where the collected ship navigation data of the target ship is located can be determined, and then the ship target classification model in the corresponding navigation scenario is used to classify and recognize it to obtain an accurate classification result. The ship target classification model in the navigation scenario in this embodiment is obtained through the above ship target classification model construction method.

[0147] Please refer to Figure 3 , this application embodiment also proposes a ship target classification model construction system, including:

[0148] The first module is used to collect the ship navigation data of each ship navigation task;

[0149] A second module, configured to extract features from multiple ship navigation data respectively to obtain multiple ship feature data, where the ship feature data includes ship behavior features and ship static features;

[0150] A third module, configured to divide the multiple ship feature data into different navigation scenarios according to the ship behavior features;

[0151] A fourth module, configured to construct a training data set corresponding to the navigation scenario according to the ship feature data in each navigation scenario;

[0152] A fifth module, configured to construct a ship target classification model corresponding to the navigation scenario according to the training data set.

[0153] It can be understood that the content in the above embodiments of the ship target classification model construction method is applicable to the embodiments of this system. The functions specifically implemented by the embodiments of this system are the same as those of the above embodiments of the ship target classification model construction method, and the beneficial effects achieved are also the same as those of the above embodiments of the ship target classification model construction method.

[0154] An embodiment of this application further provides an electronic device, which includes: a memory, a processor, a program stored on the memory and executable on the processor, and a data bus for realizing connection communication between the processor and the memory. When the program is executed by the processor, it realizes the above ship target classification model construction method or ship target classification method. The electronic device can be any intelligent terminal including a tablet computer, an in-vehicle computer, etc.

[0155] Please refer to Figure 4 , Figure 4 which schematically shows the hardware structure of an electronic device according to another embodiment. The electronic device includes:

[0156] A processor 901, which can be implemented in ways such as a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is configured to execute relevant programs to implement the technical solutions provided by the embodiments of this application;

[0157] The memory 902 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM), etc. The memory 902 can store an operating system and other application programs. When implementing the technical solutions provided in the embodiments of this specification through software or firmware, the relevant program codes are stored in the memory 902 and are called by the processor 901 to execute the ship target classification model construction method or the ship target classification method of the embodiments of this application;

[0158] The input / output interface 903 is used to implement information input and output;

[0159] The communication interface 904 is used to implement communication and interaction between this device and other devices. It can communicate through a wired method (such as USB, network cable, etc.) or through a wireless method (such as mobile network, WIFI, Bluetooth, etc.);

[0160] The bus 905 transmits information between the various components of the device (such as the processor 901, the memory 902, the input / output interface 903, and the communication interface 904);

[0161] Among them, the processor 901, the memory 902, the input / output interface 903, and the communication interface 904 are communicatively connected to each other inside the device through the bus 905.

[0162] The embodiments of this application also provide a storage medium. The storage medium is a computer-readable storage medium for computer-readable storage. The storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the above-mentioned ship target classification model construction method or ship target classification method.

[0163] As a non-transitory computer-readable storage medium, the memory can be used to store non-transitory software programs and non-transitory computer-executable programs. In addition, the memory can include high-speed random access memory, and can also include non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state storage devices. In some embodiments, the memory can optionally include a memory remotely set relative to the processor, and these remote memories can be connected to the processor through a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof.

[0164] The embodiments described in the embodiments of the present application are to more clearly illustrate the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions provided by the embodiments of the present application. Those skilled in the art can know that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of the present application are also applicable to similar technical problems.

[0165] Those skilled in the art can understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than shown in the figures, or combine some steps, or different steps.

[0166] The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0167] Those of ordinary skill in the art can understand that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware and their appropriate combinations.

[0168] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily need to be used to describe a specific order or sequence. It should be understood that such used data can be interchanged under appropriate circumstances, so that the embodiments of the present application described here can be implemented in an order other than those illustrated or described here. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0169] It should be understood that in this application, "at least one (item)" means one or more, and "a plurality" means two or more. "And / or" is used to describe the association relationship of associated objects and indicates that three relationships can exist. For example, "A and / or B" can mean: only A exists, only B exists, and both A and B exist at the same time. Among them, A and B can be singular or plural. The character " / " generally indicates that the associated objects before and after are in an "or" relationship. "At least one (item) of the following" or a similar expression means any combination of these items, including any combination of single item (item) or plural items (items). For example, at least one (item) of a, b, or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0170] In several embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For example, the division of the above units is only a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection between each other can be through some interfaces. The indirect coupling or communication connection of the system or unit can be in an electrical, mechanical or other form.

[0171] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place, or they can be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0172] In addition, each functional unit in various embodiments of this application can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

[0173] When an integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of various embodiments of this application. The aforementioned storage medium includes: various media that can store programs, such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs.

[0174] The preferred embodiments of the embodiments of this application have been described above with reference to the accompanying drawings, which does not limit the scope of the rights of the embodiments of this application. Any modifications, equivalent replacements, and improvements made by those skilled in the art without departing from the scope and essence of the embodiments of this application shall be within the scope of the rights of the embodiments of this application.

Claims

1. A method for constructing a ship target classification model, characterized in that, It includes the following steps: Collect the ship navigation data of each ship navigation task; Extract features from multiple pieces of the ship navigation data respectively to obtain multiple ship feature data, where the ship feature data includes ship behavior features and ship static features; Divide the multiple pieces of ship feature data into different navigation scenarios according to the ship behavior features; Construct a training data set corresponding to the navigation scenario according to the ship feature data in each navigation scenario; Construct a ship target classification model corresponding to the navigation scenario according to the training data set.

2. The method for constructing a ship target classification model according to claim 1, wherein The ship behavior features include ship dynamic behavior state features and behavior state change features; the ship feature data is subjected to feature extraction through the following steps: Screen out ship static data and ship dynamic data from the ship navigation data; Perform static feature extraction according to the ship static data to obtain ship static features; Perform dynamic feature extraction according to the ship dynamic data to obtain ship dynamic behavior state features; Perform characteristic change analysis according to the ship dynamic behavior state features to obtain behavior state change features.

3. The method for constructing a ship target classification model according to claim 1, wherein, Each piece of ship feature data carries the environmental features of the corresponding ship navigation task. The step of dividing the multiple pieces of ship feature data into different navigation scenarios according to the ship behavior features includes the following steps: Screen each piece of ship feature data according to data screening conditions, where the data screening conditions include static feature screening conditions, dynamic feature screening conditions, and environmental screening conditions; Divide the multiple pieces of ship feature data obtained after screening into different navigation scenarios according to the ship behavior features.

4. The method for constructing a ship target classification model according to claim 1, wherein, The ship behavior features include the course range and the navigation path. The step of dividing the multiple pieces of ship feature data obtained after screening into different navigation scenarios includes the following steps: Determine the division parameter value of the ship feature data according to the course range and the number of turning nodes of the navigation path in the ship feature data, where the division parameter value is proportional to the course range and inversely proportional to the number of turning nodes; When the division parameter value is less than the expected value, divide the corresponding ship feature data into the navigation scenario of the restricted sea area; When the division parameter value is greater than the expected value, divide the corresponding ship feature data into the navigation scenario of the open sea area.

5. The method for constructing a ship target classification model according to claim 4, wherein The step of constructing a ship target classification model corresponding to the navigation scenario according to the training data set includes the following steps: Calculate the maximum information coefficient between each ship behavior feature and the target classification according to the training data set in the navigation scenario of the open sea area; Select several ship behavior features with a greater contribution to the target classification from the multiple ship behavior features according to the maximum information coefficient; Perform target classification modeling according to the selected several ship behavior features to obtain a ship target classification model in the navigation scenario of the open sea area.

6. The method for constructing a ship target classification model according to claim 4, wherein The step of constructing a ship target classification model corresponding to the navigation scenario according to the training data set includes the following steps: Calculate the correlation degree between each ship static feature and the ship behavior feature according to the training data set in the navigation scenario of the restricted sea area; Select the ship static feature with the strongest relevance to the ship behavior feature according to the relevance degree; Perform target classification modeling based on the ship behavior feature and the selected ship static feature to obtain a ship target classification model for the navigation scenario in the restricted sea area.

7. A method for classifying ship targets, characterized in that, It includes the following steps: Obtain the ship navigation data of the target ship; Extract features from the ship navigation data to obtain ship feature data, where the ship feature data includes ship behavior features and ship static features; Determine the target navigation scenario corresponding to the ship navigation data according to the ship behavior feature; Through the ship target classification model of the target navigation scenario, determine the classification result of the target ship according to the ship behavior feature.

8. A ship target classification model construction system, characterized in that It includes: The first module is used to collect the ship navigation data of each ship navigation task; The second module is used to extract features from multiple pieces of the ship navigation data respectively to obtain multiple pieces of ship feature data, where the ship feature data includes ship behavior features and ship static features; The third module is used to divide multiple pieces of the ship feature data into different navigation scenarios according to the ship behavior feature; The fourth module is used to construct a training data set for the corresponding navigation scenario according to the ship feature data in each navigation scenario; The fifth module is used to construct a ship target classification model for the corresponding navigation scenario according to the training data set.

9. An electronic device, characterized in that, The electronic device includes a memory, a processor, a program stored on the memory and executable on the processor, and a data bus for realizing the connection and communication between the processor and the memory. When the program is executed by the processor, the steps of the method according to any one of claims 1 to 7 are realized.

10. A storage medium, the storage medium being a computer-readable storage medium for computer-readable storage, characterized in that, The storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to realize the steps of the method according to any one of claims 1 to 7.

Images