CN113642252A - Target positioning method and device based on single satellite, electronic equipment and medium - Google Patents

Target positioning method and device based on single satellite, electronic equipment and medium Download PDF

Info

Publication number
CN113642252A
CN113642252A CN202111017113.9A CN202111017113A CN113642252A CN 113642252 A CN113642252 A CN 113642252A CN 202111017113 A CN202111017113 A CN 202111017113A CN 113642252 A CN113642252 A CN 113642252A
Authority
CN
China
Prior art keywords
data
positioning
target
satellite
feature
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
CN202111017113.9A
Other languages
Chinese (zh)
Other versions
CN113642252B (en
Inventor
李修建
董洛兵
衣文军
王平
朱梦均
何新
王献青
强仕
吉元昊
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.)
Xidian University
National University of Defense Technology
Original Assignee
Xidian University
National University of Defense Technology
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 Xidian University, National University of Defense Technology filed Critical Xidian University
Priority to CN202111017113.9A priority Critical patent/CN113642252B/en
Publication of CN113642252A publication Critical patent/CN113642252A/en
Application granted granted Critical
Publication of CN113642252B publication Critical patent/CN113642252B/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]
    • G06F30/20—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
    • 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
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Engineering & Computer Science (AREA)
  • Medical Informatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Computer Hardware Design (AREA)
  • Geometry (AREA)
  • Position Fixing By Use Of Radio Waves (AREA)

Abstract

The disclosure relates to a target positioning method, a target positioning device, electronic equipment and a computer readable medium based on a single satellite. The method comprises the following steps: acquiring direction-finding positioning data of a single satellite for a plurality of periods of a target object according to a positioning request of the target object; acquiring sample data of the single satellite; training a machine learning model according to the sample data and the direction-finding positioning data of the plurality of periods to generate a single-satellite positioning model; inputting the directional location data of the plurality of cycles into the single-star location model to obtain the accurate position of the target object. The target positioning method, the target positioning device, the electronic equipment and the computer readable medium based on the single satellite fully mine time and space characteristics among long-period mass data generated in satellite measurement. The target is positioned based on the excavated features, so that the problem of target label data loss is solved, the self-adaptive capacity of target positioning is improved, and high-precision positioning of the target is realized.

Description

Target positioning method and device based on single satellite, electronic equipment and medium
Technical Field
The present disclosure relates to the field of computer information processing, and in particular, to a target positioning method and apparatus based on a single satellite, an electronic device, and a computer readable medium.
Background
With the development of modern science and technology and aerospace engineering, the satellite-borne positioning system is widely concerned due to the advantages of strong analysis capability, wide coverage range, small influence of environment and the like. The single satellite positioning is greatly applied due to the reasons of low cost, simple algorithm, high implementability and the like. However, with the generation of mass data and the improvement of positioning requirements, the traditional single-satellite positioning method is difficult to break through the self-limitation, and cannot obtain a positioning result with higher precision. The main problems can be divided into the following points:
1. in the single-satellite passive positioning process, the influence of the measurement accuracy of the arrival angle on the final positioning result is large, and how to carry out iteration or correction calculation on the target coordinate to reduce the measurement error of the arrival angle is an important problem for improving the positioning accuracy.
2. Under a long-period observation scene, mass observation data with long time span can be generated for the same radiation source target. How to dig out the time and space association relation implied among the mass data is a key problem for improving the positioning precision.
3. Due to the influence of environmental factors such as satellite elevation, orbit characteristics, detector system errors, satellite motion speed, atmosphere/marine environment and the like, under the actual condition, the data characteristics of the training data and the data characteristics of the target data have differences, and the target data cannot obtain the optimal prediction result on the final model. How to extract the common characteristics of the training data and the target data is an urgent problem to be solved for improving the positioning accuracy.
4. In satellite positioning practical applications, the satellite scout data usually contains direction finding data, but does not contain the label data of the irradiation target (the real coordinates of the irradiation target), which brings huge challenges to the training of the positioning model. How to train the positioning model under the condition of lacking the marking data becomes an irrevocable problem.
Therefore, a new single satellite-based target positioning method, apparatus, electronic device and computer readable medium are needed.
The above information disclosed in this background section is only for enhancement of understanding of the background of the disclosure and therefore it may contain information that does not constitute prior art that is already known to a person of ordinary skill in the art.
Disclosure of Invention
In view of the above, the present disclosure provides a target positioning method, device, electronic device and computer readable medium based on a single satellite, which fully mine temporal and spatial characteristics among long-period mass data generated in satellite measurement. The target is positioned based on the excavated features, so that the problem of target label data loss is solved, the self-adaptive capacity of target positioning is improved, and high-precision positioning of the target is realized.
Additional features and advantages of the disclosure will be set forth in the detailed description which follows, or in part will be obvious from the description, or may be learned by practice of the disclosure.
According to an aspect of the present disclosure, a method for positioning a target based on a single satellite is provided, the method including: acquiring direction-finding positioning data of a single satellite for a plurality of periods of a target object according to a positioning request of the target object; acquiring sample data of the single satellite; training a machine learning model according to the sample data and the direction-finding positioning data of the plurality of periods to generate a single-satellite positioning model; inputting the directional location data of the plurality of cycles into the single-star location model to obtain the accurate position of the target object.
In an exemplary embodiment of the present disclosure, before obtaining multiple periods of direction-finding positioning data of a single satellite on a target object according to a positioning request of the target object, the method further includes: and measuring the target object by a single satellite for multiple periods to generate direction-finding positioning data of the multiple periods, wherein the direction-finding positioning data comprises the coordinates of the single satellite, the estimated position of the target object, and the azimuth angle between the single satellite and the target object.
In an exemplary embodiment of the present disclosure, before acquiring the sample data of the single satellite, the method further includes: and generating the sample data based on the measurement of a single satellite on the preset target object, wherein the sample data comprises the accurate position of the preset target object and a plurality of corresponding periods of direction-finding positioning data.
In an exemplary embodiment of the present disclosure, training a machine learning model according to the sample data and the direction finding positioning data of the plurality of cycles, and generating a single star positioning model includes: taking the sample data as source domain data; taking the direction-finding positioning data of the plurality of periods as target field data; and training a base learning device, a feature regression device, a coordinate regression device and a field discriminator in a machine learning model according to the source field data and the target field data to generate the single star positioning model.
In an exemplary embodiment of the present disclosure, training a base learner, a feature regressor, a coordinate regressor, a domain discriminator in a machine learning model according to the source domain data and the target domain data to generate the single star location model includes: inputting the source domain data and the target domain data into a base learner to generate spliced data; inputting the spliced data into a feature regression device to generate data regression features; updating confrontation learning parameters of a feature regressor and a field discriminator based on the data regression feature; the coordinate regressor carries out coordinate prediction and parameter updating according to the shared characteristics of the source field data and the target field data; and when the calculation result of the coordinate regressor is converged, generating the single-satellite positioning model according to the current parameters.
In an exemplary embodiment of the present disclosure, inputting the source domain data and the target domain data into a base learner to generate stitched data includes: and inputting the source domain data and the target domain data into a basis learner for slice splicing and feature normalization processing to generate spliced data.
In an exemplary embodiment of the present disclosure, inputting the stitched data into a feature regressor to generate a data regression feature includes: generating a training set and a prediction set according to the source domain data; generating a test set according to the target field data; and performing poor training on various regression models based on the training set, the prediction set and the test set to obtain the data regression features.
In an exemplary embodiment of the present disclosure, updating the confrontation learning parameters of the feature regressor and the domain discriminator based on the data regression feature includes: inputting the data regression feature into a feature regressor and a field discriminator; the feature regressor and the field discriminator are used for counterstudy; updating the countermeasure learning parameters during countermeasure learning based on the value of the loss function.
In an exemplary embodiment of the present disclosure, the coordinate regressor performs coordinate prediction and parameter update according to the shared features of the source domain data and the target domain data, including: the coordinate regressor acquires the shared characteristics of the source field data and the target field data according to the updated confrontation learning parameters; and performing coordinate prediction through a minimum square error or a minimum absolute error based on the shared features.
In an exemplary embodiment of the present disclosure, inputting the plurality of periods of direction finding positioning data into the single star positioning model to obtain the precise position of the target object comprises: performing data fusion on the sample data and the direction-finding positioning data of the plurality of periods based on a plurality of regression models to generate feature fusion data; extracting shared features in the sample data and the direction-finding positioning data of the plurality of periods based on the feature fusion data; determining an accurate location of the target object based on the shared characteristic.
According to an aspect of the present disclosure, a single satellite-based target positioning apparatus is provided, the apparatus including: the direction-finding model is used for acquiring direction-finding positioning data of a single satellite for a plurality of periods of the target object according to a positioning request of the target object; the sample module is used for acquiring sample data of the single satellite; the training module is used for training a machine learning model according to the sample data and the direction-finding positioning data of the plurality of periods to generate a single-satellite positioning model; and the position module is used for inputting the direction-finding positioning data of the plurality of periods into the single-star positioning model so as to obtain the accurate position of the target object.
According to an aspect of the present disclosure, an electronic device is provided, the electronic device including: one or more processors; storage means for storing one or more programs; when executed by one or more processors, cause the one or more processors to implement a method as above.
According to an aspect of the disclosure, a computer-readable medium is proposed, on which a computer program is stored, which program, when being executed by a processor, carries out the method as above.
According to the target positioning method, the target positioning device, the electronic equipment and the computer readable medium based on the single satellite, the direction-finding positioning data of the single satellite for a plurality of periods of the target object is obtained according to the positioning request of the target object; acquiring sample data of the single satellite; training a machine learning model according to the sample data and the direction-finding positioning data of the plurality of periods to generate a single-satellite positioning model; and inputting the direction-finding positioning data of the plurality of periods into the single-satellite positioning model to acquire the accurate position of the target object, and fully mining the time and space characteristics among long-period mass data generated in satellite measurement. The target is positioned based on the excavated features, so that the problem of target label data loss is solved, the self-adaptive capacity of target positioning is improved, and high-precision positioning of the target is realized.
The single satellite-based target positioning method, the single satellite-based target positioning device, the single satellite-based electronic equipment and the computer readable medium can extract data features similar to target field data from source field data for radiating target coordinate prediction and combine learned knowledge into local representation. The method creatively introduces the ideas of data fusion, counterstudy and transfer study into a single-satellite positioning method, and fully excavates the time and space characteristics among long-period mass data. Meanwhile, the problem of target label data loss is solved, the self-adaptive capacity of the method is improved, and the high-precision positioning of the radiation source target is finally realized.
The method has strong practicability in the aspect of target positioning in the civil field and the military field. In the civil field, the invention can play an important role in vehicle navigation, geophysical resource exploration and other aspects; in the military field, the invention occupies an indispensable position in electronic reconnaissance, can effectively search the military strength of enemies, and provides important technical support for missile interception, weapon guidance, target reconnaissance and other applications.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.
Drawings
The above and other objects, features and advantages of the present disclosure will become more apparent by describing in detail exemplary embodiments thereof with reference to the attached drawings. The drawings described below are merely some embodiments of the present disclosure, and other drawings may be derived from those drawings by those of ordinary skill in the art without inventive effort.
Fig. 1 is a system block diagram illustrating a single satellite based target positioning method and apparatus according to an exemplary embodiment.
Fig. 2 is a flow chart illustrating a method for single satellite based target location according to an exemplary embodiment.
Fig. 3 is a flow chart illustrating a method for single satellite based target location according to another exemplary embodiment.
Fig. 4 is a diagram illustrating a model training architecture for a single satellite based target location method according to another exemplary embodiment.
Fig. 5 is a diagram illustrating a mid-model training phase of a single satellite based target positioning method according to another exemplary embodiment.
Fig. 6 is a graph of model training data for a single satellite based target positioning method, according to another exemplary embodiment.
Fig. 7 is a graph illustrating dimensional changes in model training for a single satellite based target localization method according to another exemplary embodiment.
Fig. 8 is a graph of the result of comparison experiments of the target positioning method and the DF algorithm based on a single satellite.
FIG. 9 is a block diagram illustrating a single satellite based target positioning device in accordance with an exemplary embodiment.
FIG. 10 is a block diagram illustrating an electronic device in accordance with an example embodiment.
FIG. 11 is a block diagram illustrating a computer-readable medium in accordance with an example embodiment.
Detailed Description
Example embodiments will now be described more fully with reference to the accompanying drawings. Example embodiments may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of example embodiments to those skilled in the art. The same reference numerals denote the same or similar parts in the drawings, and thus, a repetitive description thereof will be omitted.
Furthermore, the described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a thorough understanding of embodiments of the disclosure. One skilled in the relevant art will recognize, however, that the subject matter of the present disclosure can be practiced without one or more of the specific details, or with other methods, components, devices, steps, and so forth. In other instances, well-known methods, devices, implementations, or operations have not been shown or described in detail to avoid obscuring aspects of the disclosure.
The block diagrams shown in the figures are functional entities only and do not necessarily correspond to physically separate entities. I.e. these functional entities may be implemented in the form of software, or in one or more hardware modules or integrated circuits, or in different networks and/or processor means and/or microcontroller means.
The flow charts shown in the drawings are merely illustrative and do not necessarily include all of the contents and operations/steps, nor do they necessarily have to be performed in the order described. For example, some operations/steps may be decomposed, and some operations/steps may be combined or partially combined, so that the actual execution sequence may be changed according to the actual situation.
It will be understood that, although the terms first, second, third, etc. may be used herein to describe various components, these components should not be limited by these terms. These terms are used to distinguish one element from another. Thus, a first component discussed below may be termed a second component without departing from the teachings of the disclosed concept. As used herein, the term "and/or" includes any and all combinations of one or more of the associated listed items.
It is to be understood by those skilled in the art that the drawings are merely schematic representations of exemplary embodiments, and that the blocks or processes shown in the drawings are not necessarily required to practice the present disclosure and are, therefore, not intended to limit the scope of the present disclosure.
Fig. 1 is a system block diagram illustrating a single satellite based target positioning method and apparatus according to an exemplary embodiment. As shown in fig. 1, the system architecture 10 may include a single satellite, a target object (radiation source), an information integration center, a data center, and terminal equipment.
The satellite reconnaissance part is responsible for reconnaissance of the target object and is a source of system data. The reconnaissance data includes the true coordinates of the satellite, the position of the target, and the true azimuth of the satellite relative to the target object during the measurement. After the reconnaissance data is acquired by the satellite, the data may be transmitted to a data center for storage.
The information comprehensive processing center is responsible for realizing the single-satellite fusion method, namely training and calling the single-satellite positioning model. Data required by the model can be acquired from a data center and comprises source domain data and target domain data.
The data center is responsible for storing data and models, the data comprises source field data and direction finding positioning (target field) data, and the models comprise successfully trained single-star positioning models. The source domain data may be regarded as fixed sample data, including a coordinate tag. The direction-finding positioning data is data of a certain target detected by a satellite, and the direction-finding positioning data does not contain a coordinate tag.
The terminal is responsible for initiating requests for model training and model invocation. The request is transmitted to the information comprehensive processing center through the network, and the information comprehensive processing center realizes the training and calling of the model by calling the data of the data center.
The network part is responsible for communication between the remaining four parts. The satellite reconnaissance part and the network part are single arrows, the satellite can reconnaissance the radiation source target and transmit reconnaissance data to the data center through the network, but cannot transmit data to the satellite reconnaissance part through the network. The remaining three parts are the process of bidirectional interaction with the network part.
More specifically, the detailed operation flow of the system architecture 10 is as follows:
firstly, direction finding positioning data is generated by satellite measurement and is transmitted to a data center in a unified mode. The data includes the true coordinates of the satellite, the target position, and the true azimuth of the satellite relative to the radiation source during the measurement.
And the terminal is responsible for sending a positioning model training request and a positioning task request to the information comprehensive processing center. And c, transmitting a positioning model training request to the step c, and transmitting a positioning task request to the step c.
And initiating a request for obtaining sample data and long-period direction-finding data of the target object to a data center after the information comprehensive processing center obtains the positioning model training request, wherein more specifically, the long-period direction-finding data can be 3000 periods of measurement data, and each period can measure 1500 points of data.
And fourthly, after receiving the request of the information comprehensive processing center, the data center sends the sample data and the long-period direction-finding data of the related target to the information comprehensive processing center, and the data are used for training the model.
And fifthly, after receiving the data required by training, the information comprehensive processing center performs training on the single star positioning model of the target, and after the model training is finished, the information comprehensive processing center sends the trained model to the data center for storage.
After the information comprehensive processing center obtains the positioning task request, sending a calling request to the data center, wherein the calling request comprises a called positioning model and appointed direction finding data.
And seventhly, the data center returns the model and the data to the information comprehensive processing center after receiving the request.
And after acquiring the model and the direction finding data, the information comprehensive processing center calls the model and the data to complete the detection of the target, and then transmits the prediction result to the data center for storage. And meanwhile, transmitting the result to the terminal to provide decision support for the terminal.
Fig. 2 is a flow chart illustrating a method for single satellite based target location according to an exemplary embodiment. The single satellite based target positioning method 20 comprises at least steps S202 to S208.
As shown in fig. 2, in S202, direction-finding positioning data of a single satellite for a plurality of periods of a target object is obtained according to a positioning request of the target object.
Before obtaining the direction-finding positioning data of a single satellite for a plurality of periods of the target object according to the positioning request of the target object, the method further comprises: and measuring the target object by a single satellite for multiple periods to generate direction-finding positioning data of the multiple periods, wherein the direction-finding positioning data comprises the coordinates of the single satellite, the estimated position of the target object, and the azimuth angle between the single satellite and the target object.
In S204, sample data of the single satellite is acquired. Wherein, before obtaining the sample data of the single satellite, the method further comprises: and generating the sample data based on the measurement of a single satellite on the preset target object, wherein the sample data comprises the accurate position of the preset target object and a plurality of corresponding periods of direction-finding positioning data.
In S206, training a machine learning model according to the sample data and the direction finding positioning data of the plurality of periods, and generating a single star positioning model. The sample data may be, for example, source domain data; taking the direction-finding positioning data of the plurality of periods as target field data; and training a base learning device, a feature regression device, a coordinate regression device and a field discriminator in a machine learning model according to the source field data and the target field data to generate the single star positioning model.
Training a base learner, a feature regressor, a coordinate regressor and a field discriminator in a machine learning model according to the source field data and the target field data to generate the single-satellite positioning model, comprising: inputting the source domain data and the target domain data into a base learner to generate spliced data; inputting the spliced data into a feature regression device to generate data regression features; updating confrontation learning parameters of a feature regressor and a field discriminator based on the data regression feature; the coordinate regressor carries out coordinate prediction and parameter updating according to the shared characteristics of the source field data and the target field data; and when the calculation result of the coordinate regressor is converged, generating the single-satellite positioning model according to the current parameters. The specific contents of the model training will be described in detail in the embodiment corresponding to fig. 5.
In S208, the directional positioning data of the plurality of cycles is input into the single star positioning model to obtain the precise position of the target object. The sample data and the direction-finding positioning data of the plurality of periods can be subjected to data fusion based on a plurality of regression models to generate feature fusion data; extracting shared features in the sample data and the direction-finding positioning data of the plurality of periods based on the feature fusion data; determining an accurate location of the target object based on the shared characteristic.
According to the target positioning method based on the single satellite, direction-finding positioning data of the single satellite for multiple periods of a target object are obtained according to a positioning request of the target object; acquiring sample data of the single satellite; training a machine learning model according to the sample data and the direction-finding positioning data of the plurality of periods to generate a single-satellite positioning model; and inputting the direction-finding positioning data of the plurality of periods into the single-satellite positioning model to acquire the accurate position of the target object, and fully mining the time and space characteristics among long-period mass data generated in satellite measurement. The target is positioned based on the excavated features, so that the problem of target label data loss is solved, the self-adaptive capacity of target positioning is improved, and high-precision positioning of the target is realized.
It should be clearly understood that this disclosure describes how to make and use particular examples, but the principles of this disclosure are not limited to any details of these examples. Rather, these principles can be applied to many other embodiments based on the teachings of the present disclosure.
As can be seen from fig. 3, the model training process in the present application includes a basis learner, a feature regressor, a domain discriminator, and a coordinate regressor. And the base learner performs slice splicing and feature normalization processing on the input data according to a single-star direction finding positioning mechanism, and outputs an analysis result obtained by splicing to the feature processor. The feature regressor is responsible for taking the regression result of the base learner as the input to extract the features of the source field data and the target field data samples, the feature regressor consists of a convolutional neural network and a field discriminator together to form a confrontation training network, and the coordinate prediction loss of the samples obtained through the prediction of the coordinate regressor and the domain discrimination loss obtained through the field discriminator are transmitted to the feature selector. The field discriminator is responsible for inputting sample characteristics obtained by the characteristic regressor, and the field discriminator is responsible for judging whether the sample data is from a source field or a target field and forming a confrontation training network together with the alternative training of the characteristic regressor, so that the characteristics extract the similar data characteristics of the data in different fields. And the coordinate regressor is responsible for predicting the coordinates of the target field sample by using the trained model.
A feed-forward architecture is defined in model training, and for each input sample data, the model predicts a coordinate label y and a domain label d E {0,1 }. In the training phase, the direction-finding data input x is firstly passed through the base learning device b ═ Gb(·；θb) N is subjected to primary regression prediction and converted into an N-dimensional feature vector, namely b belongs to RNThe number of multiple linear regression models used for N-based learning, then b is passed through a feature extraction network f-Gf(·；θf) Is converted into a D-dimensional feature vector, and f ∈ RDAll the feature vectors extracted by the feature regressor are input into the domain discrimination network Gd(·；θd) Performing domain discrimination, wherein the feature vector corresponding to the source domain data enters a coordinate regression network Gy(·；θy) And acquiring corresponding radiation target coordinates.
Suppose the input samples X e X, where X represents the input space, and the output data is Y e Y. In the disclosure, x represents the satellite coordinates and the radiation source direction angle of each observation point of a single satellite in the process of reconnaissance, and the final target of the model is to predict Y ∈ Y.
Combined with the actual situation, influenced by the environmental factors such as satellite elevation, orbit characteristics, detector system error, satellite motion speed, atmosphere/marine environment and the like,
Figure BDA0003240280520000111
are not the same. Although we do not know the specific form of the two distributions, it is inevitable that the two mechanisms are the sameThere are implicit similar features. Based on the idea of transfer learning, the source domain distribution S (x, y) and the target domain distribution D (x, y) can be distributed. The final goal of the method is to give an input x, and to obtain a high accuracy predicted coordinate y. During training, the training sample is assumed to be x1,x2,...,xNDistribution of edge distributions S (x) and T (x) from source and target domains, while defining diIs a domain label of the ith training sample, wherein diE 0, 1. If the sample belongs to the source domain, then xi～S(x),d i0. X if the sample belongs to the target domaini～T(x),di＝1。
The input of the network in the positioning model is a direction-finding positioning source field data set with a coordinate label and a direction-finding positioning target field data set without the coordinate label, and the output is the coordinate label of the target field.
As shown in fig. 4, the single-star direction finding positioning model of the present disclosure is divided into three learning stages of regression feature learning, countermeasure learning, and coordinate regression learning, in the figure, the solid line represents forward propagation, the dotted line represents backward propagation, and the weight parameters of the feature extraction network are updated by alternately learning the feature regressor and the domain discriminator.
Fig. 5 is a flow chart illustrating a method for single satellite based target location according to another exemplary embodiment. The process 50 shown in fig. 5 is a detailed description of "training the machine learning model according to the sample data and the direction-finding positioning data of the plurality of cycles to generate the single-star positioning model".
As shown in fig. 5, in S502, the source domain data and the target domain data are input into a base learner to generate concatenated data. More specifically, the source domain data and the target domain data may be input into a basis learner for slice stitching and feature normalization processing to generate the stitched data.
In S504, the stitching data is input into a feature regressor to generate a data regression feature. A training set and a prediction set may be generated from the source domain data; generating a test set according to the target field data; and performing poor training on various regression models based on the training set, the prediction set and the test set to obtain the data regression features.
The base learner adopts various model integration strategies, firstly processes sample data, uses source field data as a training set, uses target domain data as a test set, divides the source field data set into 5 parts, four parts are used as the training set, and one part is used as a prediction set. Next, a basis regression model is selected. Six multiple linear regression models, Lasso regression (Lasso), Ridge regression (Ridge), support vector machine regression (SVR), Kernel Ridge Regression (KRR), bayesian linear regression (bayesian Ridge), elastic network (ElasticNet), were selected as base models in this disclosure. In the training stage, the six models are respectively subjected to cross training, and the source domain data set S is assumed to be divided into S ═ S i1,2, 5, the target domain data set is T, the training steps are as follows,
1. the regression model is trained for five times, and S-S is respectively trainediAs a training set siObtaining prediction results as prediction set
Figure BDA0003240280520000121
Using T as a test set
Figure BDA0003240280520000122
2. Stitching prediction set results
Figure BDA0003240280520000123
The test set takes the average of five test results:
Figure BDA0003240280520000124
3. repeating the steps, training each model to obtain the source-collar return characteristics
Figure BDA0003240280520000126
Target return feature
Figure BDA0003240280520000125
After the data characteristics of the source field sample and the target field sample are obtained, the regression effects of the six models can be judged by judging the Root Mean Square Error (RMSE) of the regression evaluation indexes, and the regression models with larger errors of prediction results are removed. In the single-satellite direction-finding positioning system described above, the observation satellite measures the radiation target at n (in this example, n is 1500) different positions of the orbit, and in practice, 30 sample data are used to converge the direction-finding positioning model, as shown in fig. 6 and 7. Each sample point can therefore be divided into 50 groups for input to the basis learner for separately obtaining regression features. After the regression feature learning stage is completed, 50 groups of regression features obtained from the same radiation target observation value are spliced into a matrix and input into the feature regressor.
In S506, the countermeasure learning parameters of the feature regressor and the domain discriminator are updated based on the data regression feature. The data regression features may be input to a feature regressor and a domain discriminator; the feature regressor and the field discriminator are used for counterstudy; updating the countermeasure learning parameters during countermeasure learning based on the value of the loss function.
The source field data and the target field data are input into a feature regressor for counterstudy after regression feature learning, the counterstudy is composed of the feature regressor and a field discriminator, the feature regressor is used for extracting common features of the source field data and the target field data to confuse the field discriminator, the field discriminator is responsible for discriminating the source of the input feature field, the loss of the counterstudy stage is composed of two parts, and the loss function can be defined as:
Figure BDA0003240280520000131
wherein L isyPredicting the loss for the coordinates, LdIn order to discriminate the loss in the domain,
Figure BDA0003240280520000137
for the coordinate loss in the ith training,
Figure BDA0003240280520000136
for the ith training midrange discrimination loss, the parameter λ represents the weight between the two learners under control, the target expectation coordinate prediction result loss against learning is low, and the domain discrimination result loss is high. In training, in order to obtain domain-invariant features, a feature mapping parameter θ is soughtfMaximizing discriminator loss, minimizing coordinate prediction loss, and seeking thetadMinimize discriminator loss, seek θyThe coordinate prediction penalty is minimized. The optimum value of the relevant parameter can be expressed as follows
Figure BDA0003240280520000132
The model uses a random gradient descent method to perform gradient update of parameters, as shown in the formula:
Figure BDA0003240280520000133
Figure BDA0003240280520000134
Figure BDA0003240280520000135
wherein mu represents learning rate, can change with time, and is different from a general random gradient descent method in that-lambda is introduced in the formula to distinguish the influence of a coordinate predictor and a domain classifier on the parameter updating of the feature extraction model, specifically, the input of the domain classifier and the coordinate regressor is from the feature regressor, but the input of the domain classifier is required to maximize the domain classification loss, and the input of the coordinate predictor is required to minimize the coordinate predictor loss, so that the feature extraction layer theta is extracted from the feature regressorfAttempting to increase the domain discriminator penalty while reducing the penalty of coordinate prediction in a stochastic gradient descent process, assumingIf-lambda is not introduced, the loss of the domain discriminator is reduced, and the feature regressor cannot obtain the common features of the source domain and the target domain, so that the transfer learning of the subsequent coordinate regressor cannot be completed.
The existing framework-integrated Stochastic Gradient Descent (SGD) method cannot directly finish Gradient updating operation of the formula, antagonistic learning is usually finished in stages in the GAN, and a Gradient Reversal Layer (GRL) is introduced into the model to avoid respectively training a feature regressor and a field discriminator. GRL does not introduce other additional parameters besides λ, and during forward propagation GRL returns the input signature as an identity transform. In the back propagation process, GRL increases lambda parameter and automatically transmits the gradient negation to the upper layer network, if R is usedλ(x) Representing the forward and backward propagation behavior of the GRL layer, this process is expressed as the formula:
Figure BDA0003240280520000141
wherein I is an identity matrix, GRL is added between the feature regressor and the field discriminator, and the discrimination loss of the field discriminator in the back propagation process is automatically negated before being propagated to the feature regressor, thereby realizing counterstudy, and also being written into a formula:
Figure BDA0003240280520000142
the SGD can be used to update the above formula after GRL is added to the model, and meanwhile, in the training process, the gradient propagation influence of the domain classifier can be gradually increased by using dynamic lambda in order to reduce the influence of the early noise signal of the domain discriminator.
Figure BDA0003240280520000143
In the formula, p represents the relative value of the iteration process, namely the ratio of the current iteration number to the total iteration number, and gamma is an adjustable hyperparameter. Through the processing of the counterstudy stage, the source field characteristics extracted by the characteristic regressor are more similar to the target field characteristics, so that the coordinate regressor trained through source field data is also suitable for the target field, and the radiation source target in the target field is predicted.
In S508, the coordinate regressor performs coordinate prediction and parameter update according to the shared features of the source domain data and the target domain data. The coordinate regressor acquires the shared characteristics of the source field data and the target field data according to the updated confrontation learning parameters; and performing coordinate prediction through a minimum square error or a minimum absolute error based on the shared features.
After the counterstudy stage, the features extracted by the feature regressor can be used for coordinate prediction:
Figure BDA0003240280520000151
wherein L isyA smoothed L1 norm Loss function (SmoothL1 Loss) is used, whose functional structure is as follows:
Figure BDA0003240280520000152
and when the difference between the predicted value and the true value is small, the error estimation is carried out by using the minimum mean square error, when the difference is large, the estimation is carried out by using the minimum absolute value error, the whole transfer learning process is that the countermeasure learning and the coordinate prediction learning are carried out alternately, and after the learning in the period is finished, the target coordinate prediction can be carried out on a new sample.
In S510, when the calculation result of the coordinate regressor converges, the single-star positioning model is generated according to the current parameters.
According to the method and the device, under the condition that the target field data lack coordinate tags, a prediction model capable of performing coordinate prediction on the target field samples can be obtained through learning based on the shared data characteristics of the counterlearning extraction source field sample data and the target field sample data.
The method is compared with a least square pure orientation passive localization algorithm (DF) and an elastic network (ElasticNet) regression algorithm, and the comparison result is shown in FIG. 8. In fig. 8, an ElasticNet algorithm, the disclosed algorithm, and a DF algorithm respectively perform position prediction on 100 different position radiation targets under 1000 monte carlo experiments, and take an average value, wherein the DF algorithm performs coordinate prediction on the same radiation target by using 50 different measurement data sets, and a final result takes an average value. As can be seen from the figure, compared with the traditional DF positioning algorithm, the accuracy of the method is improved by about 66.67%, and the radiation source target position characteristics implied in the long-period observation data can be effectively extracted, so that the target positioning accuracy is improved. In addition, the model positioning result based on data fusion is more stable, which is probably because the algorithm adopts a decision-level data fusion idea, and the basis learner performs regression feature extraction and averaging on the radiation source target measurement data, so that the fault tolerance of the model on the measurement data is higher.
The target positioning method based on the single satellite selects observation point data with different periods and different distance intervals to carry out single-satellite direction finding positioning, selects the period with the minimum error and the distance intervals as the optimal measurement parameters, and solves the problem of error correction of the angle of arrival measurement.
According to the target positioning method based on the single satellite, multiple groups of direction finding data are regressed by utilizing multiple linear regression models, data characteristics are obtained and spliced, and feature level fusion of single-satellite long period data is achieved.
The target positioning method based on the single satellite integrates multiple regression models, fully excavates the time and position relation implied by the same radiation source target in a long-period scene, and realizes decision-level data fusion of single-satellite long-period data.
According to the target positioning method based on the single satellite, the knowledge transfer between the source field and the target field is realized by using the idea of transfer learning, and the problem of missing of radiated target label data in the model training process is solved.
The target positioning method based on the single satellite applies the anti-migration learning idea to the single satellite positioning method, and the sharing characteristics between the data with the tags in the source field and the data without the tags in the target field are mined, so that the difference between the training data and the target data caused by factors such as satellite elevation, orbit characteristics, detector system errors, satellite movement speed, atmosphere/ocean environment and the like is eliminated.
Those skilled in the art will appreciate that all or part of the steps implementing the above embodiments are implemented as computer programs executed by a CPU. When executed by the CPU, performs the functions defined by the above-described methods provided by the present disclosure. The program may be stored in a computer readable storage medium, which may be a read-only memory, a magnetic or optical disk, or the like.
Furthermore, it should be noted that the above-mentioned figures are only schematic illustrations of the processes involved in the methods according to exemplary embodiments of the present disclosure, and are not intended to be limiting. It will be readily understood that the processes shown in the above figures are not intended to indicate or limit the chronological order of the processes. In addition, it is also readily understood that these processes may be performed synchronously or asynchronously, e.g., in multiple modules.
The following are embodiments of the disclosed apparatus that may be used to perform embodiments of the disclosed methods. For details not disclosed in the embodiments of the apparatus of the present disclosure, refer to the embodiments of the method of the present disclosure.
FIG. 9 is a block diagram illustrating a single satellite based target positioning device in accordance with an exemplary embodiment. As shown in fig. 9, the single satellite based target positioning device 90 includes: a direction-finding model 902, a sample module 904, a training module 906, and a location module 908.
The direction-finding model 902 is configured to obtain, according to a positioning request of a target object, direction-finding positioning data of a single satellite for multiple periods of the target object;
a sample module 904 for obtaining sample data of the single satellite;
the training module 906 is configured to train a machine learning model according to the sample data and the direction finding positioning data of the multiple periods, and generate a single-satellite positioning model;
the location module 908 is configured to input the plurality of periods of direction finding location data into the single star location model to obtain an accurate location of the target object.
FIG. 10 is a block diagram illustrating an electronic device in accordance with an example embodiment.
An electronic device 1000 according to this embodiment of the disclosure is described below with reference to fig. 10. The electronic device 1000 shown in fig. 10 is only an example and should not bring any limitation to the functions and the scope of use of the embodiments of the present disclosure.
As shown in fig. 10, the electronic device 1000 is embodied in the form of a general purpose computing device. The components of the electronic device 1000 may include, but are not limited to: at least one processing unit 1010, at least one memory unit 1020, a bus 1030 that couples various system components including the memory unit 1020 and the processing unit 1010, a display unit 1040, and the like.
Wherein the storage unit stores program code executable by the processing unit 1010 to cause the processing unit 1010 to perform steps according to various exemplary embodiments of the present disclosure described in this specification. For example, the processing unit 1010 may perform the steps shown in fig. 2 and 5.
The memory unit 1020 may include readable media in the form of volatile memory units, such as a random access memory unit (RAM)10201 and/or a cache memory unit 10202, and may further include a read only memory unit (ROM) 10203.
The memory unit 1020 may also include a program/utility 10204 having a set (at least one) of program modules 10205, such program modules 10205 including, but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which, or some combination thereof, may comprise an implementation of a network environment.
Bus 1030 may be any one or more of several types of bus structures including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, and a local bus using any of a variety of bus architectures.
The electronic device 1000 may also communicate with one or more external devices 1000' (e.g., keyboard, pointing device, bluetooth device, etc.) such that a user can communicate with devices with which the electronic device 1000 interacts, and/or any devices (e.g., router, modem, etc.) with which the electronic device 1000 can communicate with one or more other computing devices. Such communication may occur through input/output (I/O) interfaces 1050. Also, the electronic device 1000 may communicate with one or more networks (e.g., a Local Area Network (LAN), a Wide Area Network (WAN), and/or a public network such as the internet) via the network adapter 1060. A network adapter 1060 may communicate with other modules of the electronic device 1000 via the bus 1030. It should be appreciated that although not shown, other hardware and/or software modules may be used in conjunction with the electronic device 1000, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, among others.
Through the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein may be implemented by software, or by software in combination with necessary hardware. Therefore, as shown in fig. 11, the technical solution according to the embodiment of the present disclosure may be embodied in the form of a software product, which may be stored in a non-volatile storage medium (which may be a CD-ROM, a usb disk, a removable hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which may be a personal computer, a server, or a network device, etc.) to execute the above method according to the embodiment of the present disclosure.
The software product may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
The computer readable storage medium may include a propagated data signal with readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A readable storage medium may also be any readable medium that is not a readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a readable storage medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.
Program code for carrying out operations for the present disclosure may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computing device (e.g., through the internet using an internet service provider).
The computer readable medium carries one or more programs which, when executed by a device, cause the computer readable medium to perform the functions of: acquiring direction-finding positioning data of a single satellite for a plurality of periods of a target object according to a positioning request of the target object; acquiring sample data of the single satellite; training a machine learning model according to the sample data and the direction-finding positioning data of the plurality of periods to generate a single-satellite positioning model; inputting the directional location data of the plurality of cycles into the single-star location model to obtain the accurate position of the target object.
Those skilled in the art will appreciate that the modules described above may be distributed in the apparatus according to the description of the embodiments, or may be modified accordingly in one or more apparatuses unique from the embodiments. The modules of the above embodiments may be combined into one module, or further split into multiple sub-modules.
Through the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein may be implemented by software, or by software in combination with necessary hardware. Therefore, the technical solution according to the embodiments of the present disclosure may be embodied in the form of a software product, which may be stored in a non-volatile storage medium (which may be a CD-ROM, a usb disk, a removable hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which may be a personal computer, a server, a mobile terminal, or a network device, etc.) to execute the method according to the embodiments of the present disclosure.
Exemplary embodiments of the present disclosure are specifically illustrated and described above. It is to be understood that the present disclosure is not limited to the precise arrangements, instrumentalities, or instrumentalities described herein; on the contrary, the disclosure is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims.

Claims (13)

1. A target positioning method based on a single satellite is characterized by comprising the following steps:
acquiring direction-finding positioning data of a single satellite for a plurality of periods of a target object according to a positioning request of the target object;
acquiring sample data of the single satellite;
training a machine learning model according to the sample data and the direction-finding positioning data of the plurality of periods to generate a single-satellite positioning model;
inputting the directional location data of the plurality of cycles into the single-star location model to obtain the accurate position of the target object.
2. The method of claim 1, wherein before obtaining the multi-period direction-finding positioning data of the target object from the single satellite according to the positioning request of the target object, further comprising:
and measuring the target object by a single satellite for multiple periods to generate direction-finding positioning data of the multiple periods, wherein the direction-finding positioning data comprises the coordinates of the single satellite, the estimated position of the target object, and the azimuth angle between the single satellite and the target object.
3. The method of claim 1, wherein prior to acquiring sample data for the single satellite, further comprising:
and generating the sample data based on the measurement of a single satellite on the preset target object, wherein the sample data comprises the accurate position of the preset target object and a plurality of corresponding periods of direction-finding positioning data.
4. The method of claim 1, wherein training a machine learning model from the sample data and the plurality of cycles of direction-finding positioning data generates a single-star positioning model comprising:
taking the sample data as source domain data;
taking the direction-finding positioning data of the plurality of periods as target field data;
and training a base learning device, a feature regression device, a coordinate regression device and a field discriminator in a machine learning model according to the source field data and the target field data to generate the single star positioning model.
5. The method of claim 4, wherein training a basis learner, a feature regressor, a coordinate regressor, a domain discriminator in a machine learning model to generate the single-star location model based on the source domain data and the target domain data comprises:
inputting the source domain data and the target domain data into a base learner to generate spliced data;
inputting the spliced data into a feature regression device to generate data regression features;
updating confrontation learning parameters of a feature regressor and a field discriminator based on the data regression feature;
the coordinate regressor carries out coordinate prediction and parameter updating according to the shared characteristics of the source field data and the target field data;
and when the calculation result of the coordinate regressor is converged, generating the single-satellite positioning model according to the current parameters.
6. The method of claim 5, wherein inputting the source domain data and the target domain data into a base learner to generate stitched data comprises:
and inputting the source domain data and the target domain data into a basis learner for slice splicing and feature normalization processing to generate spliced data.
7. The method of claim 5, wherein inputting the stitched data into a feature regressor generates data regression features, comprising:
generating a training set and a prediction set according to the source domain data;
generating a test set according to the target field data;
and performing poor training on various regression models based on the training set, the prediction set and the test set to obtain the data regression features.
8. The method of claim 5, wherein updating the antagonistic learning parameters of the feature regressor and the domain discriminant based on the data regression features comprises:
inputting the data regression feature into a feature regressor and a field discriminator;
the feature regressor and the field discriminator are used for counterstudy;
updating the countermeasure learning parameters during countermeasure learning based on the value of the loss function.
9. The method of claim 5, wherein coordinate regressor performs coordinate prediction and parameter update based on shared features of the source domain data and the target domain data, comprising:
the coordinate regressor acquires the shared characteristics of the source field data and the target field data according to the updated confrontation learning parameters;
and performing coordinate prediction and parameter updating through a minimum square error or a minimum absolute error based on the shared characteristic.
10. The method of claim 1, wherein inputting the plurality of cycles of direction finding positioning data into the single star positioning model to obtain the precise position of the target object comprises:
performing data fusion on the sample data and the direction-finding positioning data of the plurality of periods based on a plurality of regression models to generate feature fusion data;
extracting shared features in the sample data and the direction-finding positioning data of the plurality of periods based on the feature fusion data;
determining an accurate location of the target object based on the shared characteristic.
11. An apparatus for single satellite based target positioning, comprising:
the direction-finding model is used for acquiring direction-finding positioning data of a single satellite for a plurality of periods of the target object according to a positioning request of the target object;
the sample module is used for acquiring sample data of the single satellite;
the training module is used for training a machine learning model according to the sample data and the direction-finding positioning data of the plurality of periods to generate a single-satellite positioning model;
and the position module is used for inputting the direction-finding positioning data of the plurality of periods into the single-star positioning model so as to obtain the accurate position of the target object.
12. An electronic device, comprising:
one or more processors;
storage means for storing one or more programs;
when executed by the one or more processors, cause the one or more processors to implement the method of any one of claims 1-10.
13. A computer-readable medium, on which a computer program is stored, which, when being executed by a processor, carries out the method according to any one of claims 1-10.
CN202111017113.9A 2021-08-31 2021-08-31 Target positioning method and device based on single satellite, electronic equipment and medium Active CN113642252B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111017113.9A CN113642252B (en) 2021-08-31 2021-08-31 Target positioning method and device based on single satellite, electronic equipment and medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111017113.9A CN113642252B (en) 2021-08-31 2021-08-31 Target positioning method and device based on single satellite, electronic equipment and medium

Publications (2)

Publication Number Publication Date
CN113642252A true CN113642252A (en) 2021-11-12
CN113642252B CN113642252B (en) 2022-05-17

Family

ID=78424810

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111017113.9A Active CN113642252B (en) 2021-08-31 2021-08-31 Target positioning method and device based on single satellite, electronic equipment and medium

Country Status (1)

Country Link
CN (1) CN113642252B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114518585A (en) * 2022-01-28 2022-05-20 深圳市北斗天宇科技有限公司 Target positioning method, device, terminal and storage medium based on single satellite
CN114580615A (en) * 2022-03-04 2022-06-03 哈尔滨工程大学 Distributed small platform underwater pure orientation positioning method based on neural network

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20190080261A1 (en) * 2017-09-14 2019-03-14 Beihang University Data correlation processing method based on observation data from beidou satellite
CN109613583A (en) * 2019-01-02 2019-04-12 电子科技大学 The passive object localization method of the time difference is surveyed based on the direction finding of Dan Xingyu earth station and joint
US20200049837A1 (en) * 2018-08-09 2020-02-13 Apple Inc. Machine learning assisted satellite based positioning
CN112817020A (en) * 2019-11-18 2021-05-18 千寻位置网络有限公司 GNSS observation data quality control and positioning method based on SVM model
CN112902989A (en) * 2021-05-07 2021-06-04 腾讯科技（深圳）有限公司 Data error calibration method and device, electronic equipment and storage medium

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20190080261A1 (en) * 2017-09-14 2019-03-14 Beihang University Data correlation processing method based on observation data from beidou satellite
US20200049837A1 (en) * 2018-08-09 2020-02-13 Apple Inc. Machine learning assisted satellite based positioning
CN109613583A (en) * 2019-01-02 2019-04-12 电子科技大学 The passive object localization method of the time difference is surveyed based on the direction finding of Dan Xingyu earth station and joint
CN112817020A (en) * 2019-11-18 2021-05-18 千寻位置网络有限公司 GNSS observation data quality control and positioning method based on SVM model
CN112902989A (en) * 2021-05-07 2021-06-04 腾讯科技（深圳）有限公司 Data error calibration method and device, electronic equipment and storage medium

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
PATRICK ELLIS 等: ""A SINGLE SATELLITE GEOLOCATION SOLUTION OF AN RF EMITTER USING A CONSTRAINED UNSCENTED KALMAN FILTER"", 《IEEE》 *
曾广裕 等: ""基于扩展卡尔曼滤波的全球导航卫星系统矢量跟踪建模与仿真"", 《兵工学报》 *
黄静 等: ""单星测频静态目标无源定位研究"", 《中国空间科学技术》 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114518585A (en) * 2022-01-28 2022-05-20 深圳市北斗天宇科技有限公司 Target positioning method, device, terminal and storage medium based on single satellite
CN114518585B (en) * 2022-01-28 2022-10-21 深圳市北斗天宇科技有限公司 Target positioning method, device, terminal and storage medium based on single satellite
CN114580615A (en) * 2022-03-04 2022-06-03 哈尔滨工程大学 Distributed small platform underwater pure orientation positioning method based on neural network

Also Published As

Publication number Publication date
CN113642252B (en) 2022-05-17

Similar Documents

Publication Publication Date Title
Gao et al. Long short-term memory-based deep recurrent neural networks for target tracking
Wang et al. Hybrid inference network for few-shot SAR automatic target recognition
Gauvrit et al. A formulation of multitarget tracking as an incomplete data problem
CN113642252B (en) Target positioning method and device based on single satellite, electronic equipment and medium
Beard et al. Sensor control for multi-target tracking using Cauchy-Schwarz divergence
CN113721276B (en) Target positioning method and device based on multiple satellites, electronic equipment and medium
Kandathil et al. Accuracy and tractability of a kriging model of intramolecular polarizable multipolar electrostatics and its application to histidine
CN109752710B (en) Rapid target angle estimation method based on sparse Bayesian learning
Beard et al. A generalised labelled multi-Bernoulli filter for extended multi-target tracking
Bhattarai et al. Geomagnetic field based indoor landmark classification using deep learning
Ash et al. On the relative and absolute positioning errors in self-localization systems
Song et al. A novel method for optimum global positioning system satellite selection based on a modified genetic algorithm
US20160282444A1 (en) Methods and Apparatus for Localizing a Source of a Set of Radio Signals
Tang et al. Seamless tracking of apparent point and extended targets using Gaussian process PMHT
Chang et al. Applied state estimation and association
Vatutin et al. Comparison of sequential methods for getting separations of parallel logic control algorithms using volunteer computing
Zeng et al. Offline performance prediction of PDAF with Bayesian detection for tracking in clutter
Yao et al. Wideband DOA estimation based on deep residual learning with Lyapunov stability analysis
Gostar et al. Control of sensor with unknown clutter and detection profile using multi-Bernoulli filter
Al-Sadoon et al. New and less complex approach to estimate angles of arrival
Liu et al. Multi‐sensor fusion for multi‐target tracking using measurement division
Son et al. SIMM method based on acceleration extraction for nonlinear maneuvering target tracking
Stanković et al. Efficient DoA tracking of variable number of moving stochastic EM sources in far-field using PNN-MLP model
US11280899B2 (en) Target recognition from SAR data using range profiles and a long short-term memory (LSTM) network
CN111382830A (en) Signal arrival angle estimation method and device based on hybrid particle swarm optimization

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