CN112014801B - SPWVD and improved AlexNet based composite interference identification method - Google Patents

SPWVD and improved AlexNet based composite interference identification method Download PDF

Info

Publication number
CN112014801B
CN112014801B CN202010937218.5A CN202010937218A CN112014801B CN 112014801 B CN112014801 B CN 112014801B CN 202010937218 A CN202010937218 A CN 202010937218A CN 112014801 B CN112014801 B CN 112014801B
Authority
CN
China
Prior art keywords
convolution
interference
composite
model
frequency
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.)
Active
Application number
CN202010937218.5A
Other languages
Chinese (zh)
Other versions
CN112014801A (en
Inventor
张劲东
尚东东
尹明月
杜盈
蒋宜林
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.)
Nanjing University of Aeronautics and Astronautics
Original Assignee
Nanjing University of Aeronautics and Astronautics
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 Nanjing University of Aeronautics and Astronautics filed Critical Nanjing University of Aeronautics and Astronautics
Priority to CN202010937218.5A priority Critical patent/CN112014801B/en
Publication of CN112014801A publication Critical patent/CN112014801A/en
Application granted granted Critical
Publication of CN112014801B publication Critical patent/CN112014801B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
    • G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
    • G01S7/021—Auxiliary means for detecting or identifying radar signals or the like, e.g. radar jamming signals
    • 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
    • G06N20/20—Ensemble learning
    • 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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • Software Systems (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Mathematical Physics (AREA)
  • Computing Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Remote Sensing (AREA)
  • Health & Medical Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Medical Informatics (AREA)
  • Radar Systems Or Details Thereof (AREA)

Abstract

The invention discloses a composite interference identification method based on SPWVD and improved AlexNet, wherein the identification method comprises the following steps: performing time-frequency analysis on a radar receiving signal containing a composite interference signal through SPWVD to obtain a time-frequency characteristic of the composite interference; performing dimension reduction on the time-frequency characteristics of the composite interference to obtain a time-frequency image; an improved AlexNet model is established as an interference recognition model, and the improved AlexNet model is trained through the obtained time-frequency image; and then the composite interference signal is identified by the interference identification model which is trained. The identification method provided by the invention has high-efficiency and accurate identification rate on the composite interference with larger identification difficulty.

Description

SPWVD and improved AlexNet based composite interference identification method
Technical Field
The invention relates to the technical field of interference identification methods.
Background
With the continued maturation and development of radar anti-interference technology, it is difficult for a single form of active interference to produce the desired effect on radar. The composite active interference can realize multiple characteristic interference to the radar through the combination of various interference signals, such as suppression interference and deception interference, so that the radar cannot judge the interference environment, and the anti-interference means lose the effectiveness. Composite interference has now become a major challenge for radar. As an early and necessary link for the application of the anti-interference means, the research of the composite interference identification has important theoretical significance and practical value.
At present, more research work is carried out on active interference feature extraction and identification at home and abroad, but the data of composite interference identification is relatively limited. From the existing work, the deception jamming and composite jamming identification is mostly established in a signal transformation domain (time-frequency domain, bispectrum domain, wavelet domain, fractal dimension domain and the like) to perform feature extraction, so that features with obvious distinction can be obtained. And (3) comparing the differences of the deception jamming time-frequency graphs by performing time-frequency analysis on the radar received signals, and extracting time-frequency image features based on a decision tree to perform active deception jamming identification. But the method is based on the identification of decision trees, and needs to determine a threshold value; and (3) extracting the double-spectrum slice of the deception jamming as a characteristic parameter, and inputting the characteristic parameter into an SVM vector machine for recognition. However, the kernel function and some constants of the SVM vector machine are not theoretically proven; extracting characteristic parameters of the composite interference signals in time domain, frequency domain, fractal dimension domain and the like, inputting the characteristic parameters into a BP neural network classifier for recognition, wherein the BP neural network has too many weights and too large calculated amount; aiming at the composite interference signal of radio frequency noise same-distance deception interference, the method for extracting the box dimension and the L-Z complexity distribution characteristics is adopted to realize classification recognition on the composite interference signal, but the composite interference signal is sensitive to the selected parameters and cannot distinguish more types of composite interference. Therefore, the complex characteristic extraction and recognition algorithm has great parameter difficulty, and the conventional recognition for various complex interferences has no effective algorithm support.
Disclosure of Invention
The invention aims to provide a method capable of efficiently identifying and distinguishing multiple types of composite interference and efficiently and accurately distinguishing an interference signal and a target signal.
The invention also aims to provide an application of the method.
The invention firstly discloses the following technical scheme:
a method of composite interference identification based on SPWVD and improved alexent, comprising:
s1: performing time-frequency analysis on a radar receiving signal containing a composite interference signal through pseudo-smooth Wigner distribution (SPWVD) to obtain time-frequency characteristics of the radar receiving signal;
s2: performing feature dimension reduction on the obtained time-frequency features to obtain a time-frequency image;
s3: establishing an interference identification model, and training the interference identification model through the time-frequency image;
s4: carrying out signal recognition through the interference recognition model after training;
the interference recognition model is an improved AlexNet model.
The improved AlexNet model is a model obtained by adjusting an original AlexNet model to a certain extent.
In some embodiments, the modified AlexNet model is a reduced local response normalization layer and a fully connected layer in the AlexNet model, and is obtained by replacing large convolution kernels in the convolution layers in the AlexNet model with multiple small convolution kernels in the first and second convolution layers.
In the above embodiment, the large and small values represent a relative relationship, that is, the convolution kernels in the two models are compared, where the convolution kernel with a relatively smaller size is a small convolution kernel, and the convolution kernel with a relatively larger size is a large convolution kernel.
In some embodiments, the sizes between the plurality of small convolution kernels are not exactly equal.
In some embodiments, the first convolution layer in the modified AlexNet model includes three convolution kernels, 7 x 7, 3 x 3, and 3 x 3.
In this embodiment, the convolution kernel is significantly reduced relative to the 11×11 convolution kernel at the first convolution layer of the AlexNet model.
In some embodiments, the number of the three convolution kernels is 96.
In some embodiments, the convolution steps of the three convolution kernels are 2, and 1, respectively.
In some embodiments, the second convolution layer in the modified AlexNet model comprises two convolution kernels of 3 x 3 and 3 x 3.
In this embodiment, the convolution kernel of the present scheme is significantly reduced relative to the 5×5 convolution kernel of the AlexNet model at the second convolution layer.
In some embodiments, the number of both convolution kernels is 256.
In some embodiments, the convolution step size of both convolution kernels is 1.
In some embodiments, the third, fourth, and fifth convolution layer convolution kernel sizes in the modified AlexNet model are each 3 x 3.
In some embodiments, the number of third, fourth, and fifth convolution layer convolution kernels is 384, and 256, respectively.
In some embodiments, the convolution steps of the third, fourth, and fifth convolution layer convolution kernels are each 1.
In some embodiments, the pooling layer kernel sizes after the first, second, and fifth convolution layers in the modified AlexNet model are each 3 x 3.
In some embodiments, the step size of the pooling layer kernel after the first, second, and fifth convolution layers is 2.
In some embodiments, the fifth convolution layer in the modified AlexNet model is followed by two fully connected layers, the number of neurons in the two fully connected layers being 4096 and 7, respectively.
In some embodiments, the feature dimension reduction includes discretized pseudo-smoothed wigner wiley distribution, image cropping, and image normalization.
In some embodiments, the image normalization uses a maximum minimisation method.
In some embodiments, the composite interference includes jammers and spoofing jammers.
In some embodiments, the complex form of the complex interference is selected from one or more of additive complex, product complex, and convolutional complex.
Preferably the complex form is an additive complex.
In some embodiments, the radar received signal includes: the method comprises the steps of a range gate trailing composite noise amplitude modulation signal, a range gate trailing composite noise frequency modulation signal, a speed gate trailing composite noise amplitude modulation signal, a speed gate trailing composite noise frequency modulation signal, a range-speed combined trailing composite noise amplitude modulation signal, a range-speed combined trailing composite noise frequency modulation signal and a target echo.
The invention has the following beneficial effects: compared with a common AlexNet model, the improved model provided by the invention has the advantages that a full connection layer is deleted, the parameters of a network are reduced, the convergence speed of the network is accelerated, the efficiency is greatly improved, and meanwhile, the deeper and finer image characteristics can be extracted. The identification method has high identification rate on complex interference with high difficulty, and in some specific embodiments, when the signal-to-noise ratio is 0db, the identification rate of 7 radar receiving signals in the complex interference is over 94 percent.
Drawings
Fig. 1 is a time-frequency diagram of a superimposed signal after pseudo-wigner-wili distribution transformation in the embodiment.
Fig. 2 is a pseudo-wigner-wili distribution time-frequency diagram of 7 radar reception signals in the embodiment.
Fig. 3 is a flow chart of the composite interference feature dimension reduction in the specific embodiment.
Fig. 4 shows the structure of the modified alexent model in the embodiment.
Fig. 5 is a flowchart of the recognition algorithm described in the embodiment.
Fig. 6 is a graph of loss and accuracy curves for the training set of the AlexNet model and the modified AlexNet model described in the detailed description.
Fig. 7 is a graph of test set accuracy variation of the modified alexent model described in the detailed description.
Fig. 8 is a statistical result of the change of the recognition rate of the improved AlexNet model test set with the signal to noise ratio in the specific embodiment.
Fig. 9 is a comparison of the AlexNet model and the improved AlexNet model overall recognition rate described in the detailed description.
Fig. 10 is a confusion matrix with a signal-to-noise ratio of 3db in the embodiment.
Detailed Description
The present invention will be described in detail with reference to the following examples and drawings, but it should be understood that the examples and drawings are only for illustrative purposes and are not intended to limit the scope of the present invention in any way. All reasonable variations and combinations that are included within the scope of the inventive concept fall within the scope of the present invention.
The invention can recognize the composite interference through the process shown in fig. 5, and specifically comprises the following steps:
s1: performing time-frequency analysis on a radar receiving signal containing an interference signal through SPWVD to obtain a time-frequency characteristic of composite interference;
s2: performing feature dimension reduction on the obtained time-frequency features to obtain a time-frequency image;
s2: establishing an improved AlexNet recognition model, and training the model through the obtained time-frequency image;
s3: and (5) performing recognition through the recognition model after training.
The identification method of the present invention has an advantageous effect on the identification of complex type disturbances over the prior art, so that the disturbances may preferably be complex type disturbances, as they include both suppressing disturbances and spoofing disturbances.
Among these, the complex forms of the interference may be various, such as one or more of additive complex, product complex, and convolution complex, preferably additive complex.
Specifically, the method comprises the steps of distance wave gate dragging composite noise amplitude modulation, distance wave gate dragging composite noise frequency modulation, speed wave gate dragging composite noise amplitude modulation, speed wave gate dragging composite noise frequency modulation, distance-speed combined dragging composite noise amplitude modulation and distance-speed combined dragging composite noise frequency modulation.
In a more specific implementation, step S1 may include:
the received signal is time-frequency analyzed by pseudo-smoothed wigner-willi distribution. Wherein the pseudo-smoothed Wigner-wili distribution (Smoothing Pesudo-Wigner-Ville Distribution, SPWVD) is as follows:
where s denotes the radar received signal, h (τ) and g (u) are window functions in both frequency and time domain directions.
The inventor carries out SPWVD conversion on the composite interference of the wave gate trailing interference signal and the noise modulation signal through the embodiment, and can obtain a time-frequency diagram after conversion as shown in the attached figure 1, and can see that the two signals are obviously separated, and the cross terms among different signals are obviously restrained.
The inventor also performs SPWVD conversion on a radar receiving signal containing a target signal and different composite interference through the above embodiment, as shown in fig. 2, where graph (a) represents an SPWVD time-frequency graph of the target signal, graph (b) represents an SPWVD time-frequency graph of range gate trailing composite noise modulation, graph (c) represents an SPWVD time-frequency graph of range gate trailing composite noise modulation, graph (d) represents an SPWVD time-frequency graph of range gate trailing composite noise modulation, graph (e) represents an SPWVD time-frequency graph of range-speed joint trailing composite noise modulation, and graph (f) represents an SPWVD time-frequency graph of range-speed joint trailing composite noise modulation. It can be seen that after SPWVD transformation, the feature distinction is obvious for different received signals, which indicates that the SPWVD distribution diagram can be used for distinguishing and identifying different types of composite interference signals.
After the SPWVD time-frequency diagram is obtained, the SPWVD time-frequency diagram can be used as a training set, a testing set or a target to be identified to be input into an interference identification model for training, testing or identification.
However, in order to further improve the recognition efficiency and accuracy, the feature dimension reduction can be performed on the obtained SPWVD time-frequency features before the model is input, as follows:
in particular, the feature dimension reduction may include, for example, discretized SPWVD, image cropping, and image normalization.
The time-frequency image of the composite interference signal obtained by the discretized SPWVD can be obtained according to the following method: let t=nt s (T s For a sampling period), f=m/N (N is the number of sampling points in one period) then (1) can be changed to:
wherein L is 1 The window length (assumed to be odd) is g (n), L 2 The window length 0N, 0 m N, and- (L) for h (m) (assuming an odd number) 1 -1)/2≤l≤(L 1 -1)/2、-(L 2 -1)/2≤i≤(L 2 -1)/2 corresponds to the discrete variables t, f, u, τ, respectively.
The image clipping method comprises the following steps:
A r ＝A(k(j-n 0 )+m 0 :k(j-n 1 )+m 1 ,j) (3)
wherein the slope of the chirp signal is k, points (n 0 ,m 0 )、(n 1 ,m 1 ) A and A r The image matrixes before and after cutting are respectively equal to or more than 1 and equal to or less than N.
The image normalization uses a maximum minima method.
Through the above embodiment, the inventors perform dimension reduction on the SPWVD time-frequency characteristic, and fig. 3 is a composite interference characteristic dimension reduction flowchart.
And then, taking the time-frequency diagram with the multiple groups of characteristics subjected to dimension reduction as a training set to be input into an interference recognition model for training.
The interference recognition model is improved on the basis of a deep convolutional neural network AlexNet model.
Specifically, a model structure shown in fig. 4 can be selected, and compared with an AlexNet model, the following improvements are performed:
(1) The three small convolution kernels of 7×7 and 3×3 are used for the convolution layer 1 to replace the large convolution kernel of 11×11, and the two convolution kernels of 3×3 are used for the convolution layer 2 to replace the convolution kernel of 5×5, with specific parameters shown in table 1. A smaller convolution kernel may be used to extract more deeply imaged features. The two 3×3 stacked convolutional layers are 5×5 in receptive field and have more nonlinearities than one 5×5 convolutional layer, so that the decision function is more decision and plays a role of implicit regularization.
Table 1 specific parameters of convolution layer 1 and convolution layer 2
(2) The Local Response Normalization (LRN) module is deleted. The parameters of the LRN layer often need to be cross-validated when setting them, and the LRN layer cannot promote the network's normalization capability if it is initialized and normalized by appropriate parameters. Deleting the LRN layer in the present invention makes the model easier to parallelize.
(3) The full link layer 7 of the master model is deleted. The parameters of the full connection layer account for 96.2% of the total parameters of the AlexNet network. The invention deletes one full connection layer in the model, which can reduce the parameters of the network and accelerate the calculation speed of the network.
Compared with the AlexNet model, the model of the invention is deeper and thinner, and the comparison situation is shown in Table 2.
Table 2 comparison of network and original network parameters herein
The following simulation experiment is performed through the identification process and the identification model described in the above specific embodiments:
simulation conditions:
the simulation parameter settings are shown in table 3. According to the composite interference model in the table, 7 radar receiving signals are generated between signal to noise ratio 0 and 15db respectively, the total training set of each db is 7000, 250 samples are generated at random for each radar receiving signal, and the total test set is 1750. Fig. 6 and fig. 7 are graphs of loss and accuracy for training and testing sets of AlexNet networks and modified AlexNet networks, respectively, at a signal to noise ratio of 3 db. As can be seen from the graph, the loss value of the network is lower than that of the AlexNet network in many times, the dropping speed is faster, and the recognition rate is higher. The method extracts deeper and finer image features by adopting a small convolution kernel instead of a large convolution kernel, deletes a full connection layer, reduces parameters of a network, and accordingly increases calculation speed.
TABLE 3 experimental parameters
Fig. 8 shows the statistical result of the change of the recognition rate along with the signal-to-noise ratio by adopting the network test set of the invention, and it can be seen from the figure that the recognition performance of 7 signals is improved along with the improvement of the dry-to-noise ratio, and the recognition rate of each signal is more than 94% when the signal-to-noise ratio is 0 db.
Fig. 9 is a comparison of AlexNet networks and improved overall recognition rates of AlexNet networks. From the graph, when the signal-to-noise ratio is 0db, the AlexNet network is 95.8%, and the overall recognition rate of the network is 97.1%. When the signal-to-noise ratio is 5db, the identification rate of the AlexNet network is 99.6%, and the identification rate of the network is 100%. The identification rate of the network is obviously higher than that of an AlexNet network.
FIG. 10 is a graph of an identification confusion matrix generated using the network of the present invention at a signal-to-noise ratio of 3db, 3 range-trailing composite noise amplitude modulated signals were incorrectly identified as range-speed joint trailing composite noise amplitude modulated signals, 1 range-trailing composite noise frequency modulated signal was incorrectly identified as range-speed joint trailing composite noise frequency modulated signals, the 4 distance-speed combined trailing composite noise amplitude modulation signals are erroneously identified as distance trailing composite noise amplitude modulation signals, the 2 distance-speed combined trailing composite noise frequency modulation signals are erroneously identified as distance trailing composite noise frequency modulation signals, and the overall identification rate is 99.2%.
The above examples are only preferred embodiments of the present invention, and the scope of the present invention is not limited to the above examples. All technical schemes belonging to the concept of the invention belong to the protection scope of the invention. It should be noted that modifications and adaptations to the present invention may occur to one skilled in the art without departing from the principles of the present invention and are intended to be within the scope of the present invention.

Claims (8)

1. A composite interference identification method based on SPWVD and improved AlexNet is characterized in that: comprising the following steps:
s1: carrying out time-frequency analysis on a radar receiving signal containing a composite interference signal through SPWVD to obtain a time-frequency characteristic of the radar receiving signal;
s2: performing feature dimension reduction on the obtained time-frequency features to obtain a time-frequency image;
s3: establishing an interference identification model, and training the model through the obtained time-frequency image;
s4: carrying out signal recognition through the interference recognition model after training;
the interference recognition model is an improved AlexNet model; the improved AlexNet model is used for reducing a local response normalization layer and a full connection layer in the AlexNet model, and is obtained by replacing a large convolution kernel in a convolution layer in the AlexNet model with a plurality of small convolution kernels in the first convolution layer and the second convolution layer, wherein the sizes of the small convolution kernels are not completely equal.
2. The method for identifying composite interference according to claim 1, wherein: the feature dimension reduction comprises one or more of discretized pseudo-smooth Wiggner Wiggy distribution, image clipping and image normalization.
3. The method for identifying composite interference according to claim 1, wherein: the first convolution layer in the modified AlexNet model comprises three convolution kernels of 7×7, 3×3 and 3×3, wherein the number of each convolution kernel is 96, and the convolution step length of each convolution kernel is 2, 2 and 1 respectively.
4. The method for identifying composite interference according to claim 1, wherein: the second convolution layer in the improved AlexNet model comprises two convolution kernels of 3×3 and 3×3, wherein the number of each convolution kernel is 256, and the convolution step length of each convolution kernel is 1.
5. The method for identifying composite interference according to claim 1, wherein: the convolution kernel sizes of the third, fourth and fifth convolution layers in the modified AlexNet model are all 3×3, wherein the number of convolution kernels of each layer is 384, 384 and 256 respectively, and the convolution step sizes are all 1.
6. The method for identifying composite interference according to claim 1, wherein: the core sizes of the pooling layers after the first convolution layer, the second convolution layer and the fifth convolution layer in the improved AlexNet model are 3 multiplied by 3, the convolution step sizes are 2, the fifth convolution layer is connected with two full-connection layers, and the neuron numbers of the full-connection layers are 4096 and 7 respectively.
7. The method for identifying composite interference according to claim 1, wherein: the complex form of the complex interference is selected from one or more of additive complex, product complex and convolution complex.
8. The method for identifying composite interference according to claim 7, wherein: the radar receiving signals comprise a range gate trailing composite noise amplitude modulation signal, a range gate trailing composite noise frequency modulation signal, a speed gate trailing composite noise amplitude modulation signal, a speed gate trailing composite noise frequency modulation signal, a range-speed combined trailing composite noise amplitude modulation signal, a range-speed combined trailing composite noise frequency modulation signal and a target echo signal.
CN202010937218.5A 2020-09-08 2020-09-08 SPWVD and improved AlexNet based composite interference identification method Active CN112014801B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010937218.5A CN112014801B (en) 2020-09-08 2020-09-08 SPWVD and improved AlexNet based composite interference identification method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010937218.5A CN112014801B (en) 2020-09-08 2020-09-08 SPWVD and improved AlexNet based composite interference identification method

Publications (2)

Publication Number Publication Date
CN112014801A CN112014801A (en) 2020-12-01
CN112014801B true CN112014801B (en) 2023-10-24

Family

ID=73521238

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010937218.5A Active CN112014801B (en) 2020-09-08 2020-09-08 SPWVD and improved AlexNet based composite interference identification method

Country Status (1)

Country Link
CN (1) CN112014801B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112859012B (en) * 2021-01-20 2023-12-01 北京理工大学 Radar spoofing interference identification method based on cascade convolution neural network
CN113052080A (en) * 2021-03-26 2021-06-29 南京航空航天大学 Illegal unmanned aerial vehicle detection method based on unsupervised generation countermeasure network
CN113887506B (en) * 2021-10-22 2024-04-30 吉林大学 MDS-CNN-based interference signal classification and identification method and system

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR101910540B1 (en) * 2018-03-14 2018-10-22 국방과학연구소 Apparatus and method for recognizing radar waveform using time-frequency analysis and neural network
CN110927706A (en) * 2019-12-10 2020-03-27 电子科技大学 Convolutional neural network-based radar interference detection and identification method

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR101910540B1 (en) * 2018-03-14 2018-10-22 국방과학연구소 Apparatus and method for recognizing radar waveform using time-frequency analysis and neural network
CN110927706A (en) * 2019-12-10 2020-03-27 电子科技大学 Convolutional neural network-based radar interference detection and identification method

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
Automatic Modulation Recognition for Radar Signals via Multi-Branch ACSE Networks;Shunjun Wei等;《IEEE Access》;全文 *
基于AlexNet模型的雷达信号调制类型识别;郭立民;陈鑫;陈涛;;吉林大学学报(工学版)(03);全文 *
基于时频图像Zernike矩特征的欺骗干扰识别;杨兴宇等;《现代雷达》;摘要，第91页右栏倒数第1-2段，第92-93页第1-2节及图1-2 *
基于轻量化ALEXNET模型的雷达信号识别算法研究;陈鑫;《中国优秀博硕士（硕士）学位论文全文库 信息科技辑》;摘要，正文第22-30页第3章，第31-34页第4章，第43-48页第5章及图3.1、3.3、4.1 *

Also Published As

Publication number Publication date
CN112014801A (en) 2020-12-01

Similar Documents

Publication Publication Date Title
CN112014801B (en) SPWVD and improved AlexNet based composite interference identification method
CN109379311B (en) Ultra-short wave specific signal identification method based on convolutional neural network
Zhang et al. An efficient deep learning model for automatic modulation recognition based on parameter estimation and transformation
CN107301432B (en) Self-adaptive radiation source modulation identification method based on time-frequency analysis
CN107124381B (en) Automatic identification method for digital communication signal modulation mode
CN114201988B (en) Satellite navigation composite interference signal identification method and system
CN108509911B (en) Interference signal identification method based on convolutional neural network
Wu et al. Jamming signals classification using convolutional neural network
CN110996343B (en) Intelligent recognition system and recognition method of interference recognition model based on deep convolutional neural network
CN109890043B (en) Wireless signal noise reduction method based on generative countermeasure network
Ma et al. Automatic modulation classification under non-Gaussian noise: A deep residual learning approach
CN110166387B (en) Method and system for identifying signal modulation mode based on convolutional neural network
CN112364729A (en) Modulation identification method based on characteristic parameters and BP neural network
CN111680737B (en) Radar radiation source individual identification method under differential signal-to-noise ratio condition
CN109766791B (en) Communication signal modulation identification method based on self-encoder
CN112949387A (en) Intelligent anti-interference target detection method based on transfer learning
CN111796242B (en) Improved block sparse Bayesian anti-interference method based on power characteristic value extraction
CN106911603A (en) A kind of broadband monitoring pattern Imitating signal modulation style real-time identification method
CN114896887A (en) Frequency-using equipment radio frequency fingerprint identification method based on deep learning
CN113205140A (en) Semi-supervised specific radiation source individual identification method based on generative countermeasure network
Dong et al. Radar signal modulation recognition based on bispectrum features and deep learning
Li et al. LPI Radar signal modulation recognition with feature fusion based on time-frequency transforms
Yin et al. Co-channel multi-signal modulation classification based on convolution neural network
CN112213697A (en) Feature fusion method for radar deception jamming recognition based on Bayesian decision theory
CN116383719A (en) MGF radio frequency fingerprint identification method for LFM radar

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