CN113177264A - Sea area target object multi-dimensional data simulation method and system based on generation countermeasure network - Google Patents

Sea area target object multi-dimensional data simulation method and system based on generation countermeasure network Download PDF

Info

Publication number
CN113177264A
CN113177264A CN202110511422.5A CN202110511422A CN113177264A CN 113177264 A CN113177264 A CN 113177264A CN 202110511422 A CN202110511422 A CN 202110511422A CN 113177264 A CN113177264 A CN 113177264A
Authority
CN
China
Prior art keywords
data
track
simulation
sea area
target object
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202110511422.5A
Other languages
Chinese (zh)
Other versions
CN113177264B (en
Inventor
程渤
赵帅
杨芳芳
陈俊亮
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing University of Posts and Telecommunications
Original Assignee
Beijing University of Posts and Telecommunications
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing University of Posts and Telecommunications filed Critical Beijing University of Posts and Telecommunications
Priority to CN202110511422.5A priority Critical patent/CN113177264B/en
Publication of CN113177264A publication Critical patent/CN113177264A/en
Application granted granted Critical
Publication of CN113177264B publication Critical patent/CN113177264B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/10—Geometric CAD
    • G06F30/15—Vehicle, aircraft or watercraft design
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/10—Terrestrial scenes
    • G06V20/13—Satellite images
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/08—Probabilistic or stochastic CAD

Abstract

The invention provides a sea area target object multi-dimensional data simulation method and system based on a generation countermeasure network, and the method comprises the following steps: acquiring target track space data of a sea area target object; inputting the target track space data into a trained track image data generation model, and outputting to obtain track image data of the sea area target object, wherein the trained track image data generation model is obtained by training a countermeasure network through sample track space data and sample track gray level image data corresponding to the sample track space data; and simulating to obtain multi-dimensional data of the sea area target object according to the flight path image data, the preset simulation time period and the static attribute historical data of the automatic ship identification system. The sea area target object multi-dimensional simulation data under various sensors is comprehensively simulated by utilizing a small amount of sea area target object track sample data, generating target object track information based on the generated countermeasure network, and simulating the behavior of the sea area target object.

Description

Sea area target object multi-dimensional data simulation method and system based on generation countermeasure network
Technical Field
The invention relates to the technical field of sea area monitoring, in particular to a sea area target object multi-dimensional data simulation method and system based on a generation countermeasure network.
Background
The automatic ship identification system, radar, photoelectric tracker, sonar and the like belong to important sensors for monitoring sea area targets, and the data of the devices or systems are utilized to be combined with a neural network to carry out the research of target data association and accumulated identification, so that the system and the method have important significance for the development of maritime trade and the safety of marine defense.
In the existing research model of sea area data, the accuracy of distinguishing sea area target objects depends on the quantity and quality of training data to a great extent. On the one hand, however, this type of training data is limited in its public volume due to security issues and privacy protection; on the other hand, a certain difficulty exists in acquiring massive real data in a short time, the acquired original data needs decryption, extraction and massive post-conversion processing, and finally, the sparsity of the data and whether the target characteristics contained in the data meet the algorithm requirements are difficult to judge, so that the difficulty in acquiring massive sample data comprehensively containing sea area target characteristics in a short time is high. In addition, the traditional sea area target object data simulation work mostly focuses on data simulation with a single means, even if sea area target object feature sample data is obtained, the motion complexity of the sea area target object is not comprehensively considered, the whole image of the sea area target object is difficult to be drawn by using the simulation data, and the effects of data association and accumulated identification work are greatly reduced.
Therefore, there is a need for a sea area target object multidimensional data simulation method and system based on generation of an adversarial network to solve the above problems.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a sea area target object multi-dimensional data simulation method and system based on a generation countermeasure network.
The invention provides a sea area target object multi-dimensional data simulation method based on a generation countermeasure network, which comprises the following steps:
acquiring target track space data of a sea area target object;
inputting the target track space data into a trained track image data generation model, and outputting to obtain track image data of the sea area target object, wherein the trained track image data generation model is obtained by training a countermeasure network through sample track space data and sample track gray level image data corresponding to the sample track space data;
and simulating to obtain multi-dimensional data of the sea area target object according to the track image data, the preset simulation time period and the static attribute historical data of the automatic ship identification system.
According to the sea area target object multi-dimensional data simulation method based on the generation countermeasure network provided by the invention, the trained flight path image data generation model is obtained by training through the following steps:
obtaining sample track space data of a sea area target object and sample track gray image data corresponding to the sample track space data, and converting the sample track space data through a rasterization method to obtain a track three-dimensional image matrix;
training a generator network according to the track three-dimensional image matrix to obtain sample track gray image prediction data;
training a discriminator network according to the sample track gray level image data and the sample track gray level image prediction data, and if a training result meets a preset condition, obtaining a trained track image data generation model; and the generation countermeasure network formed by the generator network and the discriminator network is provided with a constraint condition for marking the samples after random sequencing and representing the marked samples through one-hot coding.
According to the sea area target object multi-dimensional data simulation method based on the generation countermeasure network provided by the invention, after the sample track space data and the sample track gray scale image data corresponding to the sample track space data are obtained, the method further comprises the following steps:
filtering the sample track space data according to a preset sea area sandbox range;
and slicing the filtered sample track spatial data according to a preset time period, and converting the sliced sample track spatial data through a rasterization method to obtain a track three-dimensional image matrix.
According to the sea area target object multi-dimensional data simulation method based on the generation countermeasure network provided by the invention, the sea area target object multi-dimensional data comprises the following steps: simulation data of the automatic ship identification system, sonar simulation data, radar simulation data, photoelectric image synthesis data and abnormal event simulation.
According to the sea area target object multi-dimensional data simulation method based on the generation countermeasure network provided by the invention, the sea area target object multi-dimensional data is obtained through simulation according to the flight path image data, the preset simulation time period and the static attribute historical data of the automatic ship identification system, and the method comprises the following steps:
acquiring track longitude and latitude point data of the track image data at each moment in a preset simulation period, and obtaining corresponding timestamp data according to the track longitude and latitude point data through interpolation operation;
generating dynamic simulation data of the automatic ship identification system according to the timestamp data, generating static attribute simulation data of the automatic ship identification system according to the dynamic simulation data of the automatic ship identification system and the static attribute historical data of the automatic ship identification system, and obtaining simulation data of the automatic ship identification system according to the dynamic simulation data of the automatic ship identification system and the static simulation data of the automatic ship identification system;
generating sonar simulation data of an underwater target object in the sea area based on dynamic simulation data of the automatic ship identification system according to the detection frequency and the positive direction angle of the detection section of the sonar equipment;
according to the scanning angle range and the scanning frequency of the radar equipment, generating radar simulation data of a sea area target object based on the dynamic simulation data of the automatic ship identification system;
based on an image ranging technology, carrying out distance measurement marking on the flight path image data; acquiring target track longitude and latitude point data according to track image data marked by distance measurement and the shooting frequency of photoelectric shooting equipment, and acquiring the shooting distance between a sea area target object and the photoelectric shooting equipment according to the target track longitude and latitude point data;
acquiring target track image data according to the shooting distance, and adding a timestamp to the target track image data to obtain photoelectric image synthetic data of a sea area target object;
and performing abnormal event simulation based on the type and the motion process of the sea area target object, calculating the occurrence probability of the abnormal event, and modifying the simulation data of the automatic ship identification system, the sonar simulation data, the radar simulation data and the photoelectric image synthetic data if the occurrence probability of the abnormal event meets a preset threshold value so as to obtain the abnormal event simulation data of the sea area target object.
The invention also provides a sea area target object multi-dimensional data simulation system based on the generation countermeasure network, which is characterized by comprising the following components:
the system comprises a track space data acquisition module, a data acquisition module and a data processing module, wherein the track space data acquisition module is used for acquiring target track space data of a sea area target object;
the track data generation module is used for inputting the target track space data into a trained track image data generation model and outputting the trained track image data to obtain the track image data of the sea area target object, wherein the trained track image data generation model is obtained by training a countermeasure network through sample track space data and sample track gray level image data corresponding to the sample track space data;
and the multi-dimensional data simulation module is used for simulating to obtain multi-dimensional data of the sea area target object according to the track image data, the preset simulation time period and the static attribute historical data of the automatic ship identification system.
According to the sea area target object multidimensional data simulation system based on the generation countermeasure network provided by the invention, the system further comprises:
the system comprises a sample processing module, a data processing module and a data processing module, wherein the sample processing module is used for acquiring sample track space data of a sea area target object and sample track gray level image data corresponding to the sample track space data, and converting the sample track space data through a rasterization method to obtain a track three-dimensional image matrix;
the generator training module is used for training a generator network according to the track three-dimensional image matrix to obtain sample track gray image prediction data;
the discriminator training module is used for training a discriminator network according to the sample track gray level image data and the sample track gray level image prediction data, and obtaining a trained track image data generation model if a training result meets a preset condition; and the generation countermeasure network formed by the generator network and the discriminator network is provided with a constraint condition for marking the samples after random sequencing and representing the marked samples through one-hot coding.
According to the sea area target object multi-dimensional data simulation system based on the generation countermeasure network provided by the invention, the multi-dimensional data simulation module comprises:
the navigation track image data interpolation processing unit is used for acquiring navigation track longitude and latitude data of the navigation track image data at each moment in a preset simulation period, and obtaining corresponding timestamp data according to the navigation track longitude and latitude data through interpolation operation;
the automatic ship identification system data simulation unit is used for generating dynamic simulation data of an automatic ship identification system according to the timestamp data, generating static attribute simulation data of the automatic ship identification system according to the dynamic simulation data of the automatic ship identification system and the static attribute historical data of the automatic ship identification system, and obtaining simulation data of the automatic ship identification system according to the dynamic simulation data of the automatic ship identification system and the static simulation data of the automatic ship identification system;
the sonar data simulation unit is used for generating sonar simulation data of the underwater target object in the sea area based on the dynamic simulation data of the automatic ship identification system according to the detection frequency and the positive direction angle of the detection section of the sonar equipment;
the radar data simulation unit is used for generating radar simulation data of a sea area target object based on the dynamic simulation data of the automatic ship identification system according to the scanning angle range and the scanning frequency of the radar equipment;
the first photoelectric synthesis image processing unit is used for carrying out distance measurement marking on the flight path image data based on an image ranging technology; acquiring target track longitude and latitude point data according to track image data marked by distance measurement and the shooting frequency of photoelectric shooting equipment, and acquiring the shooting distance between a sea area target object and the photoelectric shooting equipment according to the target track longitude and latitude point data;
the photoelectric synthetic image second processing unit is used for acquiring target track image data according to the shooting distance and adding a timestamp to the target track image data to obtain photoelectric image synthetic data of the sea area target object;
and the abnormal event simulation unit is used for performing abnormal event simulation based on the type and the motion process of the sea area target object, calculating the occurrence probability of the abnormal event, and modifying the simulation data of the automatic ship identification system, the sonar simulation data, the radar simulation data and the photoelectric image synthetic data if the occurrence probability of the abnormal event meets a preset threshold value so as to obtain the abnormal event simulation data of the sea area target object.
The invention also provides an electronic device, which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor executes the program to realize the steps of any one of the above sea area target object multi-dimensional data simulation methods based on the generation countermeasure network.
The invention also provides a non-transitory computer-readable storage medium on which a computer program is stored, which computer program, when being executed by a processor, carries out the steps of the method for multi-dimensional data simulation of a sea area target object based on generation of a countermeasure network as described in any one of the above.
According to the sea area target object multi-dimensional data simulation method and system based on the generated countermeasure network, a small amount of sea area target object track sample data is utilized, the target object track information is generated based on the generated countermeasure network and used for simulating the behavior of a sea area target object, sea area target object multi-dimensional simulation data under various sensors are simulated comprehensively, actual data support is provided for data association and sea area accumulation identification work, and usability and effectiveness are achieved.
Drawings
In order to more clearly illustrate the technical solutions of the present invention or the prior art, the drawings needed for the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and those skilled in the art can also obtain other drawings according to the drawings without creative efforts.
FIG. 1 is a schematic flow chart of a sea area target object multi-dimensional data simulation method based on a generation countermeasure network provided by the invention;
FIG. 2 is a schematic diagram of the improved generation of a countermeasure network provided by the present invention;
FIG. 3 is a schematic structural diagram of a sea area target object multidimensional data simulation system based on a generative confrontation network provided by the invention;
FIG. 4 is a schematic diagram of a frame of a multi-dimensional data simulation system for a sea area target object provided by the present invention;
FIG. 5 is a schematic structural diagram of a track data generation module provided in the present invention;
FIG. 6 is a schematic structural diagram of a multi-dimensional data simulation module provided in the present invention;
FIG. 7 is a simulation timing diagram of the multi-dimensional data simulation system based on the sea area target object provided by the invention;
fig. 8 is a schematic structural diagram of an electronic device provided in the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is obvious that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The data simulation or generation technology is continuously developed along with the new characteristics of data and the new requirements of data science research, and has very important application in the field of machine learning. Aiming at the problems of small data volume, complex motion characteristics and the like of sea area target objects, the invention adopts a generated countermeasure network (GAN for short) to predict flight path information for the sea area target objects, wherein the generated countermeasure network consists of a generator G and a discriminator D, the generator G can generate new sample data by learning the characteristics and probability distribution of a preset data set and following the distribution, and the discriminator D continuously feeds back through the judgment of the new data to promote the generator G to improve the performance. In the present invention, the sea object mainly includes a ship, reef, underwater creature, and the like in the sea area.
Fig. 1 is a schematic flow chart of a sea area target object multidimensional data simulation method based on generation of an antagonistic network, as shown in fig. 1, the sea area target object multidimensional data simulation method based on generation of the antagonistic network includes:
step 101, obtaining target track space data of a sea area target object.
In the invention, preprocessing such as filtering and slicing is carried out on the track space data of the sea area target object with a small amount of data to obtain a data format which accords with the subsequently generated countermeasure network input, and the target track space data of the sea area target object is obtained.
102, inputting the target track space data into a trained track image data generation model, and outputting to obtain track image data of the sea area target object, wherein the trained track image data generation model is obtained by training a countermeasure network through sample track space data and sample track gray level image data corresponding to the sample track space data;
in the invention, target track space data is input into a trained track image data generation model, and the model generates track image data corresponding to a sea area target object according to the target track space data. Preferably, the method screens the flight path of the generated flight path image data based on the simulation sandbox range through a depth traversal algorithm, and stores the flight path image data meeting the range so as to be used as a data source basis for subsequent sea area target object multi-dimensional data simulation. Preferably, in an optional embodiment, the grayscale image data (grayscale image data) generated by the model is sliced according to a preset time period, a flight path meeting requirements in the grayscale image data slice is screened, and the image data is digitized into the flight path longitude and latitude data and then persisted into a data storage layer according to a preset simulation sandbox to be simulated to be stored as a graph data set.
103, simulating to obtain multi-dimensional data of the sea area target object according to the track image data, the preset simulation time period and the static attribute historical data of the automatic ship identification system.
In the invention, final data output by a track image data generation model is track space data of a sea area target object, a corresponding image data set is constructed based on the track space data, static data and dynamic data simulation of an Automatic Identification System (AIS) of a ship is carried out according to a preset simulation time period and by combining time data of the data set and static attribute historical data of the sea area target object, and the working principle of radar and sonar is simulated. Specifically, radar data and sonar data simulation is carried out according to radar parameters and sonar parameters through AIS dynamic simulation data; marking an existing target image data set by using an image ranging technology, and combining the image data set to form photoelectric data of the sea area target object at a certain moment according to the distance between the photoelectric equipment and the sea area target object; further, abnormal event simulation is carried out based on the type and the motion process of the sea area target object (relevant data obtained through simulation, such as radar simulation data and sonar simulation data, and the like). Preferably, the characteristic map of the sea area target object can be constructed in the process of simulating data so as to visually reflect the data generation condition, all the simulation data are stored in a structured data table of a relational database, and the characteristic map of the sea area target object is stored in a map database.
According to the sea area target object multi-dimensional data simulation method based on the generated countermeasure network, a small amount of sea area target object track sample data is utilized, the target object track information is generated based on the generated countermeasure network and used for simulating the behavior of a sea area target object, sea area target object multi-dimensional simulation data under various sensors are simulated comprehensively, actual data support is provided for data association and sea area accumulation identification work, and the sea area target object multi-dimensional data simulation method based on the generated countermeasure network has usability and effectiveness.
On the basis of the above embodiment, the trained flight path image data generation model is obtained by training through the following steps:
step 201, obtaining sample track space data of a sea area target object and sample track gray level image data corresponding to the sample track space data, and converting the sample track space data through a rasterization method to obtain a track three-dimensional image matrix.
According to the method, for a small amount of sample track space data (spatial dimension data of historical tracks of sea area target objects), the sample track space data is filtered according to a preset sea area sandbox range, and the filtering processing comprises data cleaning and data sorting; and then, slicing the filtered sample track space data according to a preset time period, and converting the sliced sample track space data through a rasterization method to obtain a track three-dimensional image matrix. In the invention, the spatial dimension data of the flight path is converted into a three-dimensional image matrix format which accords with the input of a generated countermeasure network through a rasterization modeling method.
Specifically, in the invention, sample track space data is filtered according to a preset sandbox range, then sliced according to a preset time period (day), sea areas in the sandbox are divided by adopting a rasterization modeling mode, a two-dimensional matrix formed by longitude and latitude coordinate points falling in each grid is counted, finally normalization processing is carried out, matrix elements are quantized to be 0 or 1, so that a three-dimensional gray map matrix, namely a track three-dimensional image matrix, is formed, and training data of a model is formed according to all the three-dimensional gray map matrices.
And 202, training a generator network according to the track three-dimensional image matrix to obtain sample track gray image prediction data.
Step 203, training a discriminator network according to the sample track gray level image data and the sample track gray level image prediction data, and if a training result meets a preset condition, obtaining a trained track image data generation model; and the generation countermeasure network formed by the generator network and the discriminator network is provided with a constraint condition for marking the samples after random sequencing and representing the marked samples through one-hot coding.
In the present invention, the generation of the countermeasure network WGAN-GP is trained by the training data obtained in the above embodiment, and according to the generation of the countermeasure network:
gen_cost＝tf.reduce_mean(disc_fake)；
disc_cost＝-tf.reduce_mean(disc_fake)+tf.reduce_mean(disc_real)；
calculating gradient penalty by using Euclidean distance, wherein lambda is a proper value (10 is taken as an example in the invention);
after the Adam optimizer is used for calculating loss function values g _ loss and d _ loss by using gen _ cost and disc _ cost', a generator and a discriminator are updated, after a plurality of (200 in the invention) epochs are trained, trends of g _ loss and d _ loss are observed, a g _ loss curve is close to 0, a d _ loss curve is oscillated near 0, the ideal loss curve changes, and the model is stable. And storing parameters for stably generating the model, so that track gray level image data of the specified sea area target object can be directly generated through the trained model in the actual simulation process. Optionally, in an embodiment, a Network structure of WGAN-GP (Wasserstein generated adaptive Network-Gradient Penalty) is modified, and a constraint Condition is added to the Network: the randomly ordered targets are label labeled and expressed by One-Hot coding to become C-WGAN-GP (Conditional WGAN-GP), thereby obtaining an improved generation countermeasure network. Then, the network is trained by using the training data obtained in the above embodiment.
Fig. 2 is a schematic structural diagram of the improved generation countermeasure network provided by the present invention, and as shown in fig. 2, Label y is a Label matrix formed by uniquely numbering sea area targets in a training sample set by using One-Hot codes, and is added to the network structure as a constraint condition. And the generated data is a gray level image matrix, and then depth traversal screening is performed in the gray level image matrix. Table 1 shows the track screening algorithm provided according to the present invention:
TABLE 1
Figure BDA0003060534280000111
Referring to table 1, the screened grayscale image matrix data is restored to actual longitude and latitude data according to the range of the preset simulation sandbox, and the data and the available track number generated by each sea area target object are stored in the data table.
On the basis of the above embodiment, the multi-dimensional data of the sea area target object includes: simulation data of the automatic ship identification system, sonar simulation data, radar simulation data, photoelectric image synthesis data and abnormal event simulation.
On the basis of the above embodiment, the simulating to obtain multi-dimensional data of the sea area target object according to the track image data, the preset simulation time period and the static attribute historical data of the automatic ship identification system includes:
acquiring track longitude and latitude point data of the track image data at each moment in a preset simulation period, and obtaining corresponding timestamp data according to the track longitude and latitude point data through interpolation operation;
generating dynamic simulation data of the automatic ship identification system according to the timestamp data, generating static attribute simulation data of the automatic ship identification system according to the dynamic simulation data of the automatic ship identification system and the static attribute historical data of the automatic ship identification system, and obtaining simulation data of the automatic ship identification system according to the dynamic simulation data of the automatic ship identification system and the static simulation data of the automatic ship identification system.
According to the method, firstly, a model is generated based on trained track image data according to a preset time period, AIS static data and dynamic data of a sea area target object are automatically generated by taking each day as unit time, whether the quantity of currently unused track space data generated by the model meets a preset quantity or not is judged, and if not, the trained track image data generation model is needed to generate the data. In the present invention, the AIS static data mainly includes: MMSI (i.e., marine mobile service identification number), call and ship names, IMO number, length and width, etc. The AIS dynamic data mainly comprises: the ship position, the positioning timestamp, the ground course, the ground speed, the heading direction, the navigation state and the like with the precision identification and the complete state.
And generating sonar simulation data of the underwater target object in the sea area based on the dynamic simulation data of the automatic ship identification system according to the detection frequency and the positive direction angle of the detection section of the sonar equipment.
According to the invention, the AIS dynamic simulation data and the longitude and latitude data of the position where the preset marine organism is located are obtained by combining sonar equipment parameters and utilizing the embodiment, and sonar detection simulation data of the marine organism (namely, a marine underwater target object) is generated.
And generating radar simulation data of the sea area target object based on the dynamic simulation data of the automatic ship identification system according to the scanning angle range and the scanning frequency of the radar equipment.
In the invention, the AIS dynamic simulation data and the related data (such as water clutter data or reef data) of the preset sea area target object are obtained by simulating the radar scanning principle and combining the radar equipment parameters by using the embodiment, so as to generate the radar scanning simulation data of the sea area target object.
Based on an image ranging technology, carrying out distance measurement marking on the flight path image data; acquiring target track longitude and latitude point data according to track image data marked by distance measurement and the shooting frequency of photoelectric shooting equipment, and acquiring the shooting distance between a sea area target object and the photoelectric shooting equipment according to the target track longitude and latitude point data;
and acquiring target track image data according to the shooting distance, and adding a timestamp to the target track image data to obtain photoelectric image synthetic data of the sea area target object.
In the invention, a data set consisting of flight path image data is marked by utilizing an image ranging technology, and photoelectric image data of a sea area target object at a preset moment is synthesized according to the position relation between the sea area target object and photoelectric equipment.
And performing abnormal event simulation based on the type and the motion process of the sea area target object, calculating the occurrence probability of the abnormal event, and modifying the simulation data of the automatic ship identification system, the sonar simulation data, the radar simulation data and the photoelectric image synthetic data if the occurrence probability of the abnormal event meets a preset threshold value so as to obtain the abnormal event simulation data of the sea area target object.
In the invention, abnormal data is generated by simulating abnormal events of a sea area target object in the motion process, firstly, whether the sea area target object is an overwater target or an underwater target is judged, and the abnormal events are classified according to the sea area target object, wherein when the sea area target object is an overwater target, the abnormal events comprise 4 types: AIS silence, AIS deception, approaching or entering a protection area and whether the weather conditions are severe or not; for underwater targets, the exceptional events include 3 types: the water stays still, approaches or enters or exits a protection area (three-dimensional space) and frequently floats out of the water, and the occurrence of certain abnormal events is represented by the difference of data.
Spring is a layered full-stack (full-stack) lightweight framework, and the core of Spring is Control Inversion (IOC) and Aspect-Oriented Programming (AOP), and simple javabeans can be used to replace EJBs to complete work. The Spring Boot is actually a package for Spring, and based on the Spring4.0 design, besides inheriting the excellent characteristics of the Spring framework, the Spring Boot also simplifies the construction and development processes of the application program by simplifying the configuration. In addition, the Spring Boot integrates a large number of frameworks (such as Mybatis, Spring MVC and the like), and solves the problems of version conflict and reference instability of the dependent package. Compared with the existing SSM (Struts2+ Spring + Hibernate3) project, the construction and management are simpler, and therefore the framework is adopted to construct the sea area target object multidimensional data simulation system based on the generation countermeasure network.
Fig. 3 is a schematic structural diagram of a sea area target multi-dimensional data simulation system based on a generation countermeasure network, as shown in fig. 3, the sea area target multi-dimensional data simulation system based on a generation countermeasure network includes a track spatial data acquisition module 301, a track data generation module 302 and a multi-dimensional data simulation module 303, where the track spatial data acquisition module 301 is used to acquire target track spatial data of a sea area target; the track data generation module 302 is configured to input the target track spatial data into a trained track image data generation model, and output to obtain track image data of the sea area target object, where the trained track image data generation model is obtained by training a countermeasure network through sample track spatial data and sample track grayscale image data corresponding to the sample track spatial data; and the multi-dimensional data simulation module 303 is used for simulating to obtain multi-dimensional data of the sea area target object according to the track image data, a preset simulation time period and static attribute historical data of the automatic ship identification system.
In the invention, a sea area target object multi-dimensional data simulation system is divided into a data display layer, a data processing layer and a data storage layer, fig. 4 is a frame schematic diagram of the sea area target object multi-dimensional data simulation system provided by the invention, and can refer to fig. 4, the data display layer is mainly used for displaying historical simulation data and receiving a man-machine interaction service function request, and a track space data acquisition module 301 is arranged in the layer; the data storage layer is mainly used for storing data generated by the track data generation module 302 and the multidimensional data simulation module 303; the data simulation processing layer bears all simulation logics of the sea area target object multi-dimensional data simulation system, and comprises a track data generation module 302 based on generation of a countermeasure network and realized based on Python and a multi-dimensional data simulation module 303 based on Java.
Further, the track data generation module 302 based on the generation countermeasure network based on the Python is implemented by taking a result obtained by preprocessing a small amount of sea area target object track space data set as a training sample, inputting the training sample into the modified generation countermeasure network for model training, outputting track image data by the module, then performing track screening by a deep traversal algorithm, finally restoring the track image data to the range of a simulation sandbox and persisting the track image data to a data storage layer, and using the track data as a data source basis for sea area target object multi-dimensional data simulation. Fig. 5 is a schematic structural diagram of a track data generation module provided by the present invention, and as shown in fig. 5, the track data generation module based on the generation of the countermeasure network C-WGAN-GP mainly includes three functions, namely, a data preprocessing function, a data generation function of the countermeasure network, and a generated data digitizing function, where 1 denotes the data preprocessing function, and inputs a data set into a network after processing the data set; 2, after the generation of the confrontation network generation data, performing a data numeralization function; 3, data persistence; and 4, calling generation of the countermeasure network generation data by the server side. In the invention, a sample data set is utilized, parameters are saved after a generation countermeasure network is trained, the trained generation countermeasure network is used as a provider of basic track data (spatial data), a server side monitoring service is started, a data generation request and a target label are received, the generation countermeasure network is utilized to generate data and carry out digitization, and a completed message is fed back to a requester after the data is durably stored in a data storage layer.
A Java-based multidimensional data simulation module 303, as shown in fig. 4, using the final data of the track data generation module 302 as the track space data of the sea area target object, combining the track space data to construct the time data of the data set and the AIS static attribute historical data of the sea area target object, performing AIS static data simulation and dynamic data simulation, simulating the working principle of radar and sonar, and performing radar and sonar data simulation according to the AIS dynamic simulation data, radar equipment and sonar equipment parameters; the existing sea area target object image data set can be marked by utilizing an image ranging technology, and the photoelectric data of the sea area target object at a certain moment is synthesized by combining the sea area target object image data set according to the distance between the photoelectric equipment and the sea area target object; generating abnormal data by an abnormal event generator; providing an interface for data custom processing (adding, deleting, modifying and checking) and realizing data display layer access; in the process of simulating data, a target characteristic map is constructed to visually reflect the data generation condition. In the invention, all simulation data is stored in a structured data table of a relational database, and the target feature map is stored in a map database.
Specifically, as shown in fig. 4, in the present invention, the multidimensional data simulation module 303 uses the concepts of object-oriented programming and multithreading to implement various functions, and mainly adopts a SpringBoot framework for development. The sea area target object and data of each dimension are packaged in an object form and stored by using a structured relation table, each function is realized in a multi-thread mode, shared variables are used for communication among threads, a Spring MVC mode is upwards used for being connected with a data display layer, and a Mybatis frame is downwards used for being connected with a data storage layer. The module uses socket to perform data communication and function call with the trajectory data generation module 302 based on generation of the countermeasure network.
Further, MVC (Model-View-Controller) is a design Model widely used in the field of Web applications, and its main idea is to decouple the module containing service data from the View of the display module, i.e. C (Controller) separates V (View, user client) and M (JavaBean: encapsulated data) to constitute MVC. Spring Web MVC is a request-driven lightweight Web framework adopting an MVC architecture idea, and is realized on the basis of Java, the responsibility of a Web layer is decoupled, and a request-response model is used for data transmission, so that the development process is simplified. The invention adopts Spring MVC mode to communicate with the data display layer, realizes the functions of increasing, deleting, modifying and checking data and has the appearance of simulation data.
Mybatis is one of open-source data persistence layer frameworks, and the inside of the framework encapsulates operations for accessing a database through JDBC, thereby providing support for commonly used SQL statement query, advanced mapping and storage processes, and also subtracting complex work such as manually writing JDBC codes, setting parameters and retrieving result sets. Mybatis has the advantages of flexible configuration and use, SQL is compiled by XML, program codes and SQL sentences are completely decoupled, unified management and optimization are facilitated, reuse can be realized, and the design of the whole system is simpler, more convenient and clearer, and more convenient for code maintenance and unit testing. To achieve code reuse and low module coupling, the present invention employs the framework to enable data communication with a data store layer.
In the sea area target object multidimensional data simulation system, simulation data realize automatic simulation in a timing task mode in multiple threads. Setting a daily generation task module for checking available basic track data (namely space data corresponding to track image data) of each sea area target once a day based on a daily generation task thread, if the basic track data is unavailable, requesting to generate based on a track data generation module 302 for generating an antagonistic network, specifically, setting a daily generation task functional module to automatically execute a data simulation task once a day at 0 point, calling an AIS data simulation unit to generate data, detecting unused track longitude and latitude data of each sea area target, and if the number is 0, calling a target label to call a track data generation module for generating the track longitude and latitude data based on the antagonistic network. The AIS and sonar threads automatically generate data once a day, abnormal events are simulated to occur simultaneously in the period, and the generated data are stored in a data storage layer and a global cache; and the running state of the radar thread is determined according to the set time, the global cache is circularly checked, whether scanning is performed or not is judged, and the photoelectric image data is simultaneously synthesized in the process of generating the radar data. Data nodes of the target feature map are automatically added once a day, and event nodes are added in real time by using asynchronous threads. The man-machine interaction service and the automatic data simulation can be carried out simultaneously.
According to the sea area target object multi-dimensional data simulation system based on the generated countermeasure network, a small amount of sea area target object track sample data is utilized, the target object track information is generated based on the generated countermeasure network and used for simulating the behavior of a sea area target object, sea area target object multi-dimensional simulation data under various sensors are simulated comprehensively, actual data support is provided for data association and sea area accumulation identification work, and the sea area target object multi-dimensional data simulation system based on the generated countermeasure network has usability and effectiveness.
On the basis of the above embodiment, the system further includes:
the system comprises a sample processing module, a data processing module and a data processing module, wherein the sample processing module is used for acquiring sample track space data of a sea area target object and sample track gray level image data corresponding to the sample track space data, and converting the sample track space data through a rasterization method to obtain a track three-dimensional image matrix;
the generator training module is used for training a generator network according to the track three-dimensional image matrix to obtain sample track gray image prediction data;
the discriminator training module is used for training a discriminator network according to the sample track gray level image data and the sample track gray level image prediction data, and obtaining a trained track image data generation model if a training result meets a preset condition; and the generation countermeasure network formed by the generator network and the discriminator network is provided with a constraint condition for marking the samples after random sequencing and representing the marked samples through one-hot coding.
In the present invention, the sample processing module includes a data preprocessing function unit, and is specifically configured to: filtering the data set according to the range of the simulation sandbox, slicing according to time (day), dividing sea areas in the sandbox in a space rasterization modeling mode, counting longitude and latitude coordinate points falling in each grid to form a two-dimensional matrix, finally performing normalization processing, enabling matrix elements to be valued to be 0 or 1 to form a three-dimensional gray scale map matrix, and enabling all gray scale matrixes to form training data of a model. In the invention, a generator training module and a discriminator training module are used for training a model, preferably, a network structure of WGAN-GP is reformed, and a constraint Condition is added into the network: and (4) label labeling the randomly ordered targets and representing the targets by using One-Hot coding, thereby obtaining an improved generation countermeasure network.
On the basis of the foregoing embodiment, fig. 6 is a schematic structural diagram of a multidimensional data simulation module provided by the present invention, and as shown in fig. 6, 1 indicates that when a daily generation task module detects track basic data that needs to generate a marine target, the daily generation task module requests a track data generation module to generate corresponding data; 2, the track data generation module generates data and then persists the data, and feeds back the result to the daily task generation module; 3, a daily generation task module calls an AIS data simulation unit, a sonar data simulation unit, and combines track basic data, ocean target static attributes (historical data) and a time set to synthesize AIS static information, dynamic information and sonar information detected by an underwater target, wherein the AIS static information, the dynamic information and the sonar information comprise sparse track data interpolation enriching operation; 4, AIS data simulation unit price stores AIS data into an AIS static information table and a dynamic information table, a sonar data simulation unit stores sonar information of sea area targets detected by simulated sonar equipment into the sonar information table and stores the sonar information into a global cache, and meanwhile, an abnormal event simulation unit can simulate abnormal data of some sea area targets according to probability, update AIS information and the sonar information, and simultaneously store some information fields of abnormal events into an abnormal event table; 5, globally caching the radar data generated when the analog radar works by the radar data simulation unit and storing the radar data into a radar information table; 6, the photoelectric image synthesis unit matches the target images with the closest image data set according to the position and time information; and 7, asynchronously constructing a target feature map, and storing data nodes and event nodes of a target into a map database.
Further, the multidimensional data simulation module comprises:
the navigation track image data interpolation processing unit is used for acquiring navigation track longitude and latitude data of the navigation track image data at each moment in a preset simulation period, and obtaining corresponding timestamp data according to the navigation track longitude and latitude data through interpolation operation;
and the data simulation unit of the automatic ship identification system is used for generating dynamic simulation data of the automatic ship identification system according to the timestamp data, generating static attribute simulation data of the automatic ship identification system according to the dynamic simulation data of the automatic ship identification system and the static attribute historical data of the automatic ship identification system, and obtaining the simulation data of the automatic ship identification system according to the dynamic simulation data of the automatic ship identification system and the static simulation data of the automatic ship identification system.
In the present invention, the AIS data simulation unit is specifically configured to: for each sea area target object, extracting a time set consisting of starting time points of tracks appearing in each corresponding slice in the image data set, randomly selecting one time point of the time set as a time starting point of the secondary generated tracks, matching track longitude and latitude point data generated by a track data generation module based on a generated countermeasure network, adopting interpolation operation, calculating time stamp data of all points in the tracks through t ═ s/v, and generating AIS dynamic simulation data by combining with static attribute historical data (MMSI (man-machine-specification) number, call number, state, type of book, ship length, ship width, speed v and the like) of the sea area target object.
And the sonar data simulation unit is used for generating sonar simulation data of the underwater target object in the sea area based on the dynamic simulation data of the automatic ship identification system according to the detection frequency and the detection positive direction angle of the cross section of the sonar equipment.
In the present invention, the sonar data simulation unit is specifically configured to: according to the frequency detected by sonar equipment and the positive direction angle of a sonar detection section, partial data of a preset sea area target object in AIS dynamic simulation data is used as track data of an underwater target projected on the sea level, for the data detected by each sonar equipment, the underwater depth sDeep of the target is randomly generated according to a certain rule, the deviation angle between the sea area target object and the positive direction is calculated by using a trigonometric function principle, and therefore the sonar detection data of the sea area target object is generated, and sonar simulation data are obtained.
And the radar data simulation unit is used for generating radar simulation data of the sea area target object based on the dynamic simulation data of the automatic ship identification system according to the scanning angle range and the scanning frequency of the radar equipment.
In the present invention, the radar data simulation unit is specifically configured to: according to the scanning angle range and frequency of the radar equipment, the AIS dynamic longitude and latitude data points of all sea area targets are scanned and traversed in a circulating mode, every time when the AIS dynamic longitude and latitude data points move by one step length (a small sector range is moved), whether the AIS dynamic longitude and latitude data points accord with the scanning time of the current radar equipment and are within the scanning range is judged, and if the AIS dynamic longitude and latitude data points accord with the scanning time of the current radar equipment, one piece of radar scanning data is correspondingly generated. In order to avoid repeated scanning, a hash table indexMap is set, keys are sea area target object identifiers, the values are index values of longitude and latitude data points in tracks, the number of track points scanned by each sea area target object in the scanning is recorded, after one round of scanning, the angle of a radar is moved, the indexMap is modified, and then the next round of scanning is continued.
The first photoelectric synthesis image processing unit is used for carrying out distance measurement marking on the flight path image data based on an image ranging technology; acquiring target track longitude and latitude point data according to track image data marked by distance measurement and the shooting frequency of photoelectric shooting equipment, and acquiring the shooting distance between a sea area target object and the photoelectric shooting equipment according to the target track longitude and latitude point data;
and the photoelectric composite image second processing unit is used for acquiring target track image data according to the shooting distance and adding a time stamp to the target track image data to obtain the photoelectric image composite data of the sea area target object.
In the invention, an image ranging technology is utilized to carry out maximum and minimum distance measurement and marking on each image in an image data set, the farthest distance which can be shot by photoelectric equipment is taken as a threshold value, partial longitude and latitude data of a sea area target object track are extracted according to the shooting frequency of the equipment, the distance between the sea area target object and the photoelectric equipment is calculated, a proper sea area target object image is selected, a time stamp is added, and therefore photoelectric image data of the sea area target object is obtained through synthesis.
And the abnormal event simulation unit is used for performing abnormal event simulation based on the type and the motion process of the sea area target object, calculating the occurrence probability of the abnormal event, and modifying the simulation data of the automatic ship identification system, the sonar simulation data, the radar simulation data and the photoelectric image synthetic data if the occurrence probability of the abnormal event meets a preset threshold value so as to obtain the abnormal event simulation data of the sea area target object.
In the present invention, the abnormal event simulation unit is specifically configured to: according to the definition of the abnormal event and the data change characteristics after the abnormal event occurs, the occurrence of the abnormal event of the sea area target object is simulated, the joint probability is calculated according to the occurrence probability configuration of various abnormal events and the probability configuration of a certain type of event occurring on each sea area target object, whether the certain event occurs on the sea area target object on the same day is judged, and if the certain event occurs, AIS simulation data, sonar simulation data, radar simulation data and photoelectric image synthesis data generated on the day in the data storage layer are modified.
Optionally, the multidimensional data simulation module further includes the following functional units, and the data customization (add, delete, modify, and check) processing unit is specifically configured to: providing a data query interface for displaying by a data display layer; providing an equipment parameter modification interface to modify parameters of the sonar and the radar; providing a custom track interface; after the sea area target object identification, the track point longitude and latitude and the time data are input, an AIS data simulation unit is called to carry out interpolation processing and static and dynamic data generation of the sea area target object. The target feature map construction function module is specifically configured to: the characteristic map of each sea area target object is established in a map database, a plurality of dimensionality data nodes in the previous day are added before data are generated every day, event nodes are added in the map in real time every time an abnormal event occurs, and the condition of data simulation can be intuitively reflected.
In another embodiment, a simulation process of the sea area target object multidimensional data simulation system of the present invention is generally described, fig. 7 is a simulation timing chart of the sea area target object multidimensional data simulation system provided by the present invention, and as shown in fig. 7, a cycle 1 indicates that multidimensional data of one slice (one day) is generated every day, and the operation is repeated; the circulation 2 means that after the space track information of a plurality of sea area target objects generated by the countermeasure network is generated, the track image of each sea area target object is circularly processed through track screening, and the data restored into the simulation sand box is durably stored into a data table; cycle 3 represents an AIS data simulation unit or a sonar data simulation unit, and generates AIS data or sonar information of each sea area target object in a cycle manner; cycle 4 means that the radar repeatedly scans the radar information generated according to the AIS and sonar information within a specified time period and persists in a database table; cycle 5 represents the generation of abnormal data by setting the occurrence of abnormal events according to the past behaviors and probabilities of the sea area target objects every day.
Fig. 8 is a schematic structural diagram of an electronic device provided in the present invention, and as shown in fig. 8, the electronic device may include: a processor (processor)801, a communication interface (communication interface)802, a memory (memory)803 and a communication bus 804, wherein the processor 801, the communication interface 802 and the memory 803 complete communication with each other through the communication bus 804. The processor 801 may call logic instructions in the memory 803 to execute a sea area target multi-dimensional data simulation method based on generation of a countermeasure network, the method comprising: acquiring target track space data of a sea area target object; inputting the target track space data into a trained track image data generation model, and outputting to obtain track image data of the sea area target object, wherein the trained track image data generation model is obtained by training a countermeasure network through sample track space data and sample track gray level image data corresponding to the sample track space data; and simulating to obtain multi-dimensional data of the sea area target object according to the track image data, the preset simulation time period and the static attribute historical data of the automatic ship identification system.
In addition, the logic instructions in the memory 803 may be implemented in the form of software functional units and stored in a computer readable storage medium when the logic instructions are sold or used as independent products. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-only memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.
In another aspect, the present invention further provides a computer program product, the computer program product includes a computer program stored on a non-transitory computer-readable storage medium, the computer program includes program instructions, when the program instructions are executed by a computer, the computer can execute the sea area target object multidimensional data simulation method based on generation of an antagonistic network provided by the above methods, the method includes: acquiring target track space data of a sea area target object; inputting the target track space data into a trained track image data generation model, and outputting to obtain track image data of the sea area target object, wherein the trained track image data generation model is obtained by training a countermeasure network through sample track space data and sample track gray level image data corresponding to the sample track space data; and simulating to obtain multi-dimensional data of the sea area target object according to the track image data, the preset simulation time period and the static attribute historical data of the automatic ship identification system.
In still another aspect, the present invention further provides a non-transitory computer-readable storage medium, on which a computer program is stored, where the computer program is implemented by a processor to execute the sea area target object multidimensional data simulation method based on generation of an adversarial network provided in the foregoing embodiments, and the method includes: acquiring target track space data of a sea area target object; inputting the target track space data into a trained track image data generation model, and outputting to obtain track image data of the sea area target object, wherein the trained track image data generation model is obtained by training a countermeasure network through sample track space data and sample track gray level image data corresponding to the sample track space data; and simulating to obtain multi-dimensional data of the sea area target object according to the track image data, the preset simulation time period and the static attribute historical data of the automatic ship identification system.
The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. With this understanding in mind, the above-described technical solutions may be embodied in the form of a software product, which can be stored in a computer-readable storage medium such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in the embodiments or some parts of the embodiments.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.

Claims (10)

1. A sea area target object multi-dimensional data simulation method based on a generation countermeasure network is characterized by comprising the following steps:
acquiring target track space data of a sea area target object;
inputting the target track space data into a trained track image data generation model, and outputting to obtain track image data of the sea area target object, wherein the trained track image data generation model is obtained by training a countermeasure network through sample track space data and sample track gray level image data corresponding to the sample track space data;
and simulating to obtain multi-dimensional data of the sea area target object according to the track image data, the preset simulation time period and the static attribute historical data of the automatic ship identification system.
2. The sea area target object multidimensional data simulation method based on generation of the countermeasure network as claimed in claim 1, wherein the trained flight path image data generation model is obtained by training through the following steps:
obtaining sample track space data of a sea area target object and sample track gray image data corresponding to the sample track space data, and converting the sample track space data through a rasterization method to obtain a track three-dimensional image matrix;
training a generator network according to the track three-dimensional image matrix to obtain sample track gray image prediction data;
training a discriminator network according to the sample track gray level image data and the sample track gray level image prediction data, and if a training result meets a preset condition, obtaining a trained track image data generation model; and the generation countermeasure network formed by the generator network and the discriminator network is provided with a constraint condition for marking the samples after random sequencing and representing the marked samples through one-hot coding.
3. The sea area target object multidimensional data simulation method based on the generation countermeasure network as claimed in claim 2, wherein after the obtaining of the sample track space data and the sample track gray scale image data corresponding to the sample track space data, the method further comprises:
filtering the sample track space data according to a preset sea area sandbox range;
and slicing the filtered sample track spatial data according to a preset time period, and converting the sliced sample track spatial data through a rasterization method to obtain a track three-dimensional image matrix.
4. The sea area target object multi-dimensional data simulation method based on generation of the countermeasure network according to claim 1, wherein the sea area target object multi-dimensional data comprises: simulation data of the automatic ship identification system, sonar simulation data, radar simulation data, photoelectric image synthesis data and abnormal event simulation.
5. The sea area target object multi-dimensional data simulation method based on the generation countermeasure network of claim 4, wherein the simulation of the sea area target object multi-dimensional data according to the track image data, the preset simulation time period and the static attribute historical data of the automatic ship identification system comprises:
acquiring track longitude and latitude point data of the track image data at each moment in a preset simulation period, and obtaining corresponding timestamp data according to the track longitude and latitude point data through interpolation operation;
generating dynamic simulation data of the automatic ship identification system according to the timestamp data, generating static attribute simulation data of the automatic ship identification system according to the dynamic simulation data of the automatic ship identification system and the static attribute historical data of the automatic ship identification system, and obtaining simulation data of the automatic ship identification system according to the dynamic simulation data of the automatic ship identification system and the static simulation data of the automatic ship identification system;
generating sonar simulation data of an underwater target object in the sea area based on dynamic simulation data of the automatic ship identification system according to the detection frequency and the positive direction angle of the detection section of the sonar equipment;
according to the scanning angle range and the scanning frequency of the radar equipment, generating radar simulation data of a sea area target object based on the dynamic simulation data of the automatic ship identification system;
based on an image ranging technology, carrying out distance measurement marking on the flight path image data; acquiring target track longitude and latitude point data according to track image data marked by distance measurement and the shooting frequency of photoelectric shooting equipment, and acquiring the shooting distance between a sea area target object and the photoelectric shooting equipment according to the target track longitude and latitude point data;
acquiring target track image data according to the shooting distance, and adding a timestamp to the target track image data to obtain photoelectric image synthetic data of a sea area target object;
and performing abnormal event simulation based on the type and the motion process of the sea area target object, calculating the occurrence probability of the abnormal event, and modifying the simulation data of the automatic ship identification system, the sonar simulation data, the radar simulation data and the photoelectric image synthetic data if the occurrence probability of the abnormal event meets a preset threshold value so as to obtain the abnormal event simulation data of the sea area target object.
6. A sea area target object multi-dimensional data simulation system based on a generation countermeasure network is characterized by comprising:
the system comprises a track space data acquisition module, a data acquisition module and a data processing module, wherein the track space data acquisition module is used for acquiring target track space data of a sea area target object;
the track data generation module is used for inputting the target track space data into a trained track image data generation model and outputting the trained track image data to obtain the track image data of the sea area target object, wherein the trained track image data generation model is obtained by training a countermeasure network through sample track space data and sample track gray level image data corresponding to the sample track space data;
and the multi-dimensional data simulation module is used for simulating to obtain multi-dimensional data of the sea area target object according to the track image data, the preset simulation time period and the static attribute historical data of the automatic ship identification system.
7. The sea area target object multidimensional data simulation system based on generation of the countermeasure network as recited in claim 6, further comprising:
the system comprises a sample processing module, a data processing module and a data processing module, wherein the sample processing module is used for acquiring sample track space data of a sea area target object and sample track gray level image data corresponding to the sample track space data, and converting the sample track space data through a rasterization method to obtain a track three-dimensional image matrix;
the generator training module is used for training a generator network according to the track three-dimensional image matrix to obtain sample track gray image prediction data;
the discriminator training module is used for training a discriminator network according to the sample track gray level image data and the sample track gray level image prediction data, and obtaining a trained track image data generation model if a training result meets a preset condition; and the generation countermeasure network formed by the generator network and the discriminator network is provided with a constraint condition for marking the samples after random sequencing and representing the marked samples through one-hot coding.
8. The sea area target object multidimensional data simulation system based on generation of countermeasure network as claimed in claim 6, wherein the multidimensional data simulation module comprises:
the navigation track image data interpolation processing unit is used for acquiring navigation track longitude and latitude data of the navigation track image data at each moment in a preset simulation period, and obtaining corresponding timestamp data according to the navigation track longitude and latitude data through interpolation operation;
the automatic ship identification system data simulation unit is used for generating dynamic simulation data of an automatic ship identification system according to the timestamp data, generating static attribute simulation data of the automatic ship identification system according to the dynamic simulation data of the automatic ship identification system and the static attribute historical data of the automatic ship identification system, and obtaining simulation data of the automatic ship identification system according to the dynamic simulation data of the automatic ship identification system and the static simulation data of the automatic ship identification system;
the sonar data simulation unit is used for generating sonar simulation data of the underwater target object in the sea area based on the dynamic simulation data of the automatic ship identification system according to the detection frequency and the positive direction angle of the detection section of the sonar equipment;
the radar data simulation unit is used for generating radar simulation data of a sea area target object based on the dynamic simulation data of the automatic ship identification system according to the scanning angle range and the scanning frequency of the radar equipment;
the first photoelectric synthesis image processing unit is used for carrying out distance measurement marking on the flight path image data based on an image ranging technology; acquiring target track longitude and latitude point data according to track image data marked by distance measurement and the shooting frequency of photoelectric shooting equipment, and acquiring the shooting distance between a sea area target object and the photoelectric shooting equipment according to the target track longitude and latitude point data;
the photoelectric synthetic image second processing unit is used for acquiring target track image data according to the shooting distance and adding a timestamp to the target track image data to obtain photoelectric image synthetic data of the sea area target object;
and the abnormal event simulation unit is used for performing abnormal event simulation based on the type and the motion process of the sea area target object, calculating the occurrence probability of the abnormal event, and modifying the simulation data of the automatic ship identification system, the sonar simulation data, the radar simulation data and the photoelectric image synthetic data if the occurrence probability of the abnormal event meets a preset threshold value so as to obtain the abnormal event simulation data of the sea area target object.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the steps of the sea area target object multidimensional data simulation method based on generation of countermeasure network according to any one of claims 1 to 5 when executing the computer program.
10. A non-transitory computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method for multi-dimensional data simulation of a sea area target based on generation of an antagonistic network according to any one of claims 1 to 5.
CN202110511422.5A 2021-05-11 2021-05-11 Sea area target object multi-dimensional data simulation method and system based on generation countermeasure network Active CN113177264B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110511422.5A CN113177264B (en) 2021-05-11 2021-05-11 Sea area target object multi-dimensional data simulation method and system based on generation countermeasure network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110511422.5A CN113177264B (en) 2021-05-11 2021-05-11 Sea area target object multi-dimensional data simulation method and system based on generation countermeasure network

Publications (2)

Publication Number Publication Date
CN113177264A true CN113177264A (en) 2021-07-27
CN113177264B CN113177264B (en) 2022-06-24

Family

ID=76928781

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110511422.5A Active CN113177264B (en) 2021-05-11 2021-05-11 Sea area target object multi-dimensional data simulation method and system based on generation countermeasure network

Country Status (1)

Country Link
CN (1) CN113177264B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115660027A (en) * 2022-10-10 2023-01-31 北京邮电大学 Multi-device sea area target data generation method and system supporting small samples
CN116186907A (en) * 2023-05-04 2023-05-30 中国人民解放军海军工程大学 Method, system and medium for analyzing navigable state based on state of marine subsystem

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105205226A (en) * 2015-08-28 2015-12-30 武汉理工大学 Two-dimensional maritime simulation system
CN109670411A (en) * 2018-11-30 2019-04-23 武汉理工大学 Based on the inland navigation craft point cloud data depth image processing method and system for generating confrontation network
CN109948722A (en) * 2019-03-27 2019-06-28 中国人民解放军战略支援部队航天工程大学 A method of identification extraterrestrial target
KR102016977B1 (en) * 2019-04-01 2019-09-02 (주)지오시스템리서치 Sea surface temperature images regenerating system and method usign gan deep learning
CN111860640A (en) * 2020-07-17 2020-10-30 大连海事大学 Specific sea area data set augmentation method based on GAN
CN112699717A (en) * 2019-10-23 2021-04-23 中国科学院遥感与数字地球研究所 SAR image generation method and generation device based on GAN network

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105205226A (en) * 2015-08-28 2015-12-30 武汉理工大学 Two-dimensional maritime simulation system
CN109670411A (en) * 2018-11-30 2019-04-23 武汉理工大学 Based on the inland navigation craft point cloud data depth image processing method and system for generating confrontation network
CN109948722A (en) * 2019-03-27 2019-06-28 中国人民解放军战略支援部队航天工程大学 A method of identification extraterrestrial target
KR102016977B1 (en) * 2019-04-01 2019-09-02 (주)지오시스템리서치 Sea surface temperature images regenerating system and method usign gan deep learning
CN112699717A (en) * 2019-10-23 2021-04-23 中国科学院遥感与数字地球研究所 SAR image generation method and generation device based on GAN network
CN111860640A (en) * 2020-07-17 2020-10-30 大连海事大学 Specific sea area data set augmentation method based on GAN

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
宋珂慧等: "基于生成式对抗网络的结构化数据表生成模型", 《计算机研究与发展》, vol. 56, no. 9, 30 September 2019 (2019-09-30) *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115660027A (en) * 2022-10-10 2023-01-31 北京邮电大学 Multi-device sea area target data generation method and system supporting small samples
CN116186907A (en) * 2023-05-04 2023-05-30 中国人民解放军海军工程大学 Method, system and medium for analyzing navigable state based on state of marine subsystem
CN116186907B (en) * 2023-05-04 2023-09-15 中国人民解放军海军工程大学 Method, system and medium for analyzing navigable state based on state of marine subsystem

Also Published As

Publication number Publication date
CN113177264B (en) 2022-06-24

Similar Documents

Publication Publication Date Title
Nii et al. Signal-to-symbol transformation: HASP/SIAP case study
CN113177264B (en) Sea area target object multi-dimensional data simulation method and system based on generation countermeasure network
US20220230387A1 (en) Generating realistic point clouds
Borchers et al. Using hidden Markov models to deal with availability bias on line transect surveys
Rosa et al. Challenges in producing policy-relevant global scenarios of biodiversity and ecosystem services
CN113420009B (en) Electromagnetic data analysis device, system and method based on big data
CN106030612B (en) Photo thermal map is provided
CA2951533A1 (en) Automated generation of digital elevation models
Wang et al. BANet: Small and multi-object detection with a bidirectional attention network for traffic scenes
Souza et al. Efficient runtime capture of multiworkflow data using provenance
CN112819171A (en) Data searching method and system based on table function and computer storage medium
CN114841228A (en) Generating a data structure for specifying a visual data set
Latifi et al. Modelling stratified forest attributes using optical/LiDAR features in a central European landscape
CN114676204B (en) Multi-attribute-based marine monitoring data three-dimensional management method and system
Kalinicheva et al. Predicting vegetation stratum occupancy from airborne LiDAR data with deep learning
Basnet et al. Analysis of multifactorial social unrest events with spatio-temporal k-dimensional tree-based dbscan
Crawford et al. Learning 3D object-oriented world models from unlabeled videos
Martin-Schwarze et al. Joint modeling of distances and times in point-count surveys
CN110413662A (en) A kind of multi-channel type economic data input system, acquisition system and method
CN116303839B (en) Index calculation method for geospatial data
CN115660027B (en) Multi-equipment sea area target data generation method and system supporting small samples
EP4283255A1 (en) Controlling an aquatic vessel
Kermarrec et al. Optimal Surface Fitting of Point Clouds Using Local Refinement: Application to GIS Data
Bormann et al. Real-time indexing of point cloud data during LiDAR capture
CN117115366B (en) Environmental model reconstruction method, system and equipment based on unmanned system three-dimensional perception

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant