CN110515096B - Convolutional neural network-based satellite navigation interference signal identification device and method - Google Patents
Convolutional neural network-based satellite navigation interference signal identification device and method Download PDFInfo
- Publication number
- CN110515096B CN110515096B CN201910772412.XA CN201910772412A CN110515096B CN 110515096 B CN110515096 B CN 110515096B CN 201910772412 A CN201910772412 A CN 201910772412A CN 110515096 B CN110515096 B CN 110515096B
- Authority
- CN
- China
- Prior art keywords
- neural network
- interference signal
- layer
- interference
- convolutional
- 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
Links
Images
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
- G01S19/00—Satellite radio beacon positioning systems; Determining position, velocity or attitude using signals transmitted by such systems
- G01S19/01—Satellite radio beacon positioning systems transmitting time-stamped messages, e.g. GPS [Global Positioning System], GLONASS [Global Orbiting Navigation Satellite System] or GALILEO
- G01S19/015—Arrangements for jamming, spoofing or other methods of denial of service of such systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Radar, Positioning & Navigation (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Remote Sensing (AREA)
- General Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Networks & Wireless Communication (AREA)
- Image Analysis (AREA)
- Noise Elimination (AREA)
Abstract
The invention discloses a satellite navigation interference signal identification device and method based on a convolutional neural network, based on the advantages of a CNN deep learning network in two-dimensional data processing, an interference signal innovatively converts a one-dimensional interference signal after down sampling into a two-dimensional image form in a multi-channel repeated mode, and a convolutional neural network for interference signal identification is constructed on the basis of the one-dimensional interference signal and the two-dimensional image form, wherein the deep learning network comprises three convolutional layers which respectively comprise 32 and 3 convolutional layers
Convolution layer of 3, 16 3
3 convolutional layers and 16 of 3
3, introduce the normalization layer behind each convolution layer, introduce the pooling layer behind each normalization layer, add the full connection layer at last and regard as the output layer, the convolution layer all selects the ReLU as the activation function, the full connection layer adopts softmax as the activation function, finishes training the back at CNN classification model, effectively discerns satellite interference signal and accomplishes sorting, more high-efficient, quick.
Description
Technical Field
The invention belongs to the technical field of satellite navigation anti-interference, and particularly relates to a satellite navigation interference signal identification device and method based on a convolutional neural network.
Background
The detection and identification of the traditional satellite navigation signal are generally considered to have four stages, namely interference source detection, interference source spectrum analysis, interference source feature extraction and interference source identification.
In the aspect of interference Detection algorithms, energy Detection algorithms (ED), namely a time domain short-time Energy Detection algorithm and a frequency domain short-time Energy Detection algorithm are mainly used in China, so that rapid interference source Detection can be realized, and a good Detection effect can be obtained; in addition, the requirements can be fulfilled by Short-time Fourier Transform (STFT) and Wigner-Ville distribution.
The interference source spectrum analysis is a process of performing short-time Fourier transform on a signal after the existence of interference is determined, and then confirming the position of an interference frequency point and the interference bandwidth of the interference frequency point through a frequency point detection algorithm. The interference source feature extraction generally selects interference feature factors such as interference source existence factors, interference signal 3dB bandwidth, interference signal frequency domain kurtosis, interference signal time domain kurtosis, interference signal Wigner-Hough kurtosis and interference signal wavelet coefficient peak-to-average power ratio to distinguish signals.
After the processing of the interference signal is completed, the interference recognition can be implemented by using two traditional interference recognition classification algorithms, support Vector Machine (SVM) and decision tree, which are based on feature value extraction. SVMs are supervised learning models that analyze data and recognize patterns, and can be used for classification and regression tasks. This module is particularly useful in situations where there is a large amount of correct data and the number of abnormal situations that one attempts to detect is not significant. The decision tree is a statistical method based on facts, in actual problems, aiming at different states of the same problem, an optimal corresponding scheme is selected according to the problem states under a certain probability criterion to form a decision, and all the problems and the corresponding schemes thereof form the decision tree together. However, the two interference identification and classification algorithms based on feature value extraction have many difficulties and disadvantages in practical application: aiming at the technical field of satellite navigation with complex and various interference signals, the method based on the decision tree needs manual feature extraction, is time-consuming and labor-consuming, has larger error and general recognition efficiency; the algorithm adopting the SVM needs to adopt a large amount of training data, has poor self-adaptive processing capacity, is difficult to acquire interference data in the fields of satellite navigation interference and the like, and cannot be applied to an actual system. Therefore, it becomes important to design an efficient and fast identification device suitable for the field of satellite navigation interference.
Disclosure of Invention
The invention provides a satellite navigation interference signal recognition device based on a convolutional neural network and a method thereof aiming at the existing problems, wherein the device comprises a preprocessing module, a convolutional neural network module and an output module, wherein the preprocessing module is used for converting a one-dimensional satellite navigation typical interference signal into a two-dimensional image; the convolutional neural network module comprises three convolutional layers, wherein the convolutional layers respectively comprise 32 convolutional layers of 3 multiplied by 3, 16 convolutional layers of 3 multiplied by 3 and 16 convolutional layers of 3 multiplied by 3, a normalization layer is introduced after each convolutional layer, a pooling layer is introduced after each normalization layer, and finally a full-connection layer is added to serve as an output layer and used for performing simulation training on two-dimensional image signals, effectively extracting signal characteristics and classifying the signal characteristics, and outputting identification and classification results through an output module.
In order to achieve the purpose, the invention adopts the technical scheme that: the satellite navigation interference signal identification method based on the convolutional neural network mainly comprises the following steps:
s1, signal preprocessing: converting the one-dimensional satellite navigation typical interference signal into a two-dimensional image;
s2, constructing a convolutional neural network: the deep learning network comprises three convolutional layers, wherein the convolutional layers respectively comprise 32 convolutional layers with the number of 3 x 3, 16 convolutional layers with the number of 3 x 3 and 16 convolutional layers with the number of 3 x 3, a normalization layer is introduced after each convolutional layer, a pooling layer is introduced after each normalization layer, and finally a full-connection layer is added to serve as an output layer, the convolutional layers all select a ReLU to serve as an activation function, and the full-connection layer adopts softmax to serve as the activation function;
s3, training and classifying: and (3) inputting the two-dimensional image into the convolutional neural network in the step (S2), automatically extracting characteristics of the signal, finishing classification recognition through training, and finally outputting a recognition and sorting result of the interference signal.
As an improvement of the present invention, in step S1, the signal is down-converted to baseband processing, and length truncation is performed on the zero intermediate frequency signal.
As an improvement of the present invention, in the step S1, the one-dimensional signal is converted into a two-dimensional image after being copied.
As another improvement of the present invention, in the step S3, in the network training process, the training database is generated by adjusting different waveform parameters of the interference signal, and is trained for different interference signal types, where the training algorithm is an sgdm algorithm.
In order to achieve the purpose, the invention also adopts the technical scheme that: the satellite navigation interference signal identification device based on the convolutional neural network comprises a preprocessing module, a convolutional neural network module and an output module, wherein the convolutional neural network module is respectively connected with the preprocessing module and the output module,
the preprocessing module is used for converting the one-dimensional satellite navigation typical interference signal into a two-dimensional image;
the convolutional neural network module comprises three convolutional layers, wherein the convolutional layers respectively comprise 32 convolutional layers of 3 multiplied by 3, 16 convolutional layers of 3 multiplied by 3 and 16 convolutional layers of 3 multiplied by 3, a normalization layer is introduced after each convolutional layer, a pooling layer is introduced after each normalization layer, and finally a full-connection layer is added to serve as an output layer;
the output module is used for the training of the convolutional neural network module and the output of the classification result, and can display two-dimensional image input information, convolutional neural network simulation training information and identification sorting result information.
Compared with the prior art, the invention discloses a satellite navigation interference signal identification device based on a convolutional neural network and a method thereof, which have the following advantages:
(1) The advantages of the CNN deep learning network in two-dimensional data processing are utilized to convert various interference signals into image forms, so that the interference identification problem is converted into an image processing problem, and the advantages and the characteristics of the CNN network are effectively highlighted;
(2) By constructing the CNN network and utilizing the image identification method, the interference signal can be directly identified, the process of manual feature extraction is avoided, and the problem of reduced identification performance caused by insufficient feature extraction is solved; (3) Under the condition that satellite navigation interference data are insufficient, the deep learning network based on the CNN is utilized, and the sorting and the recognition of interference signals can be realized, so that the recognition performance superior to that of traditional methods such as an SVM and the like is obtained, and the method is more efficient and faster.
Drawings
FIG. 1 is a schematic diagram illustrating a signal preprocessing flow of step S1 in a convolutional neural network-based satellite navigation jamming signal identification method according to the present invention;
FIG. 2 is a schematic diagram of a convolutional neural network structure in step S2 of the method for identifying satellite navigation interference signals based on a convolutional neural network according to the present invention;
3.1 and 3.2 are software design pages of the output module of the satellite navigation jamming signal recognition device based on the convolutional neural network according to the present invention;
fig. 4.1 is a gray scale diagram after preprocessing of the mono interference signal in embodiment 2 of the present invention;
fig. 4.2 is a gray scale diagram after processing of multi-tone interference signals in embodiment 2 of the present invention;
fig. 4.3 is a gray scale diagram after processing of the narrowband BPSK interference signal in embodiment 2 of the present invention;
fig. 4.4 is a gray scale diagram after processing of the wideband BPSK interference signal in embodiment 2 of the present invention;
fig. 4.5 is a gray scale diagram after processing of a narrow-band gaussian interference signal in embodiment 2 of the present invention;
fig. 4.6 is a gray scale diagram of the processing of the wide-band gaussian interference signal in embodiment 2 of the present invention;
FIG. 4.7 is a gray scale diagram of the interference signal processing of the pulse signal in embodiment 2 of the present invention;
fig. 4.8 is a gray scale diagram of linear frequency sweep interference signal processing in embodiment 2 of the present invention;
fig. 5 is a diagram illustrating the recognition effect of eight interference signals in embodiment 2 of the present invention, wherein,
fig. 5.1 is a single tone interference recognition effect diagram, fig. 5.2 is a multi-tone interference recognition effect diagram, fig. 5.3 is a narrowband BPSK interference recognition effect diagram, fig. 5.4 is a wideband BPSK interference recognition effect diagram, fig. 5.5 is a narrowband gaussian interference recognition effect diagram, fig. 5.6 is a wideband gaussian interference recognition effect diagram, fig. 5.7 is a pulse signal interference recognition effect diagram, and fig. 5.8 is a linear frequency sweep interference recognition effect diagram;
FIG. 6 is a schematic diagram of a training process of a convolutional neural network according to embodiment 2 of the present invention;
fig. 7 is a schematic diagram of a simulation result of the convolutional neural network in embodiment 2 of the present invention.
Detailed Description
The invention will be explained in more detail below with reference to the drawings and examples.
Example 1
The satellite navigation interference signal identification method based on the convolutional neural network mainly comprises the following steps:
s1, signal preprocessing: converting a one-dimensional satellite navigation typical interference signal into a two-dimensional image, as shown in fig. 1, where fig. 1 is a schematic diagram of a signal preprocessing flow, considering that if a satellite signal is not down-converted to an intermediate frequency or a baseband, the signal is too long, and therefore down-conversion is selected to the baseband processing; in order to reduce the calculation complexity of subsequent processing and the memory use in the training and recognition process, the zero intermediate frequency signal is subjected to length truncation. After the code of the one-dimensional interference signal is realized primarily, the one-dimensional signal is copied for multiple times into a two-dimensional image form, image representations of the interference signal under all parameters are obtained by adjusting different signal waveform parameters, and the image representations are stored in an interference signal image database;
s2, constructing a convolutional neural network: the CNN-based deep learning network can effectively reduce data volume and improve network learning speed by introducing convolutional layers, as shown in fig. 2, fig. 2 is a schematic diagram of a convolutional neural network structure applied to satellite interference signal identification and sorting in the present invention, and it can be seen from the diagram that the deep learning network includes three convolutional layers, respectively including 32 convolutional layers of 3 × 3, 16 convolutional layers of 3 × 3 and 16 convolutional layers of 3 × 3, and the use of the convolutional layers effectively reduces computational complexity, and in order to solve gradient dispersion, a normalization layer is introduced after the convolutional layers. In order to further reduce the scale of the extracted feature mapping graph, a pooling layer is introduced behind each normalization layer, the pooling layer is used for improving the robustness of the system, the nonlinear layer is used for fitting a nonlinear function, and finally, a full-connection layer is added as an output layer, the convolution layers select a ReLU as an activation function, the problem of gradient disappearance possibly existing in back propagation is effectively solved, and the full-connection layer adopts softmax as the activation function;
s3, training and classifying: inputting the two-dimensional interference image into the convolutional neural network in the step S2, automatically extracting characteristics of the signal, finishing classification recognition through training, and finally outputting a recognition and sorting result of the interference signal.
The invention preprocesses the signal, constructs the two-dimensional image data of the interference signal, then utilizes the convolution neural network to automatically extract the characteristics of the signal, completes the classification identification through training, and inputs the two-dimensional interference image after down sampling into the convolution neural network, thereby efficiently and rapidly realizing the detection and identification of multiple interference sources.
Example 2
The satellite navigation interference signal identification device based on the convolutional neural network comprises a preprocessing module, a convolutional neural network module and an output module, wherein the convolutional neural network module is respectively connected with the preprocessing module and the output module,
the preprocessing module is used for converting the one-dimensional satellite navigation typical interference signal into a two-dimensional image;
the convolutional neural network module comprises three convolutional layers, wherein the convolutional layers respectively comprise 32 convolutional layers with the number of 3 multiplied by 3, 16 convolutional layers with the number of 3 multiplied by 3 and 16 convolutional layers with the number of 3 multiplied by 3, a normalization layer is introduced after each convolutional layer, a pooling layer is introduced after each normalization layer, and finally a full-connection layer is added to serve as an output layer;
the output module is used for outputting the training and classification results of the convolutional neural network module, and can display two-dimensional image input information, convolutional neural network simulation training information and identification and sorting result information, as shown in fig. 3.1, and both fig. 3.1 and fig. 3.2 are software design pages of the invention.
The display interface of fig. 3.1 is divided into two main parts, the upper part displays the corresponding input signal and its prediction result, wherein the left graph can display the time domain waveform of the input signal, the right side displays its power spectrum, and the lower part displays the prediction result. When the prediction result is consistent with the signal mode selected by the user, the prediction is correct; otherwise, the prediction is wrong.
The CNN network training part is arranged below, and when training for the first time and retraining are needed, a user can select whether to regenerate a new data set or use an original data set. In the simulation process, a plurality of signals can be respectively and randomly generated according to the characteristics of different interference signals, one part of the signals is randomly selected as a training set, the rest of the signals are used as a test set, and the richness of a data set is met through the random process.
After the data set generation is finished, the button is grayed again by blue, and then the button of the retraining CNN network can be clicked to train. The training process can be shown in real time in fig. 3.2. The whole training process is divided into four stages, each stage comprises 56 iterations, and 224 iterations are carried out. And the test is carried out every thirty times, in order to clearly see the accuracy and the change process of the loss function, the curve is further smoothed on the basis of real-time display, and the results between two times of training are connected in a straight line.
Eight typical interference signals are selected and respectively are single-tone interference, multi-tone interference, gaussian narrow-band interference, BPSK narrow-band interference, pulse interference, gaussian wide-band interference, BPSK wide-band interference, and linear frequency-sweep interference, wherein key characteristic parameters corresponding to each interference signal are shown in table 1 below.
TABLE 1 random Key characteristics parameter for an interfering Signal
| Interference signal class | Key features | Interference signal class | Key feature |
| Single tone interference | Interference frequency | Multi-tone interference | Interference frequency and frequency point number |
| Narrowband BPSK interference | Symbol length and bit stream | Wideband BPSK interference | Symbol length and bit stream |
| Narrow band gaussian interference | Center frequency and bandwidth | Wide band gaussian interference | Center frequency and bandwidth |
| Pulse signal interference | Symbol length and duty cycle | Linear swept frequency interference | Starting frequency and sweep rate |
When single-tone interference is generated, in order to generate an enough database, the interference frequency points can be adjusted to generate, and the rest seven typical satellite interference signal databases are also generated by adjusting key characteristic parameters; when multi-tone interference is generated, selecting proper interference frequency and the number of interference frequency points; when generating narrowband BPSK interference, on the premise that the frequency domain is less than 10% of the baseband, randomly selecting the length of a code element, and randomly selecting the code element to send 1 or-1 with a certain probability; the wideband BPSK interference and the narrowband BPSK interference occupy different frequency domains, and on the basis that the frequency domain is larger than 10% of the baseband, the length of the selected code element is adjusted to generate a plurality of different bit streams, but the length of the code element is shorter than that of the narrowband BPSK interference because the frequency domain is more; both narrow-band gaussian interference and wide-band gaussian interference are generated by selecting specific center frequency and bandwidth, but the difference between the frequency domains is set to be relatively large as the size of the frequency domain like narrow-band BPSK and wide-band BPSK, so as to avoid the situation of misidentification; selecting proper code element length and duty ratio for the pulse signal; the linear sweep selects the sweep start frequency and sweep rate such that the frequency of its interfering signal varies linearly with time. Through the key parameter adjustment and selection process, a satisfactory interference signal database can be generated, and subsequent training and testing of the CNN network are facilitated.
There are 1000 two-dimensional gray-scale maps of 500 length and 10 width in each sample, and the ratio of training set to test set is 9:1, a two-dimensional signal database is generated, the gray-scale maps of the eight interference signals of which are shown in fig. 4.1-4.8. As can be seen from the figure, the monophonic interference exhibits regular black and white intervals, and the higher the frequency, the denser the black and white stripes; the multi-tone interference is a signal generated by overlapping a plurality of single-tone interferences, so that the variation is more hierarchical, the gray level is more, and the multi-tone interference has certain regularity because of the possibility of having a common period; narrowband BPSK interference can only exhibit a very few symbols within a limited sampling time because of the long symbol width; wideband BPSK interference may exhibit more symbols in a limited sampling time than narrowband BPSK interference; the line regularity of the narrow-band Gaussian interference is weaker than that of other interference signals; the line regularity of the broadband Gaussian interference signal is weaker than that of the narrowband Gaussian interference signal; the pulse interference is also obvious black and white, but due to different duty ratios, the width proportion of black and white is different in one period; the linear sweep clearly fades from white to black, as the frequency is higher and higher, and therefore appears clearly faded.
Simulation training was then performed through four stages, with 56 iterations per stage. The training process is shown in fig. 6, the simulation result is shown in fig. 7, and the simulation accuracy reaches 92.25%. In the figure, it can be seen that in the initial training stage, the accuracy rate rises faster, and the loss function falls faster, which indicates that the designed network can effectively extract features and classify the features, after 90 iterations, the accuracy rate and the loss function both tend to be gentle, or even slightly decrease, indicating that learning enters the stable stage, and at this time, increasing the learning frequency cannot significantly improve the accuracy rate, so the designed training frequency is reasonable.
The interference signal category is selected in the upper module shown in fig. 3.1 to observe the identification effect of the interference signal. At present, eight kinds of interference signals can be selected in the selection of the interference signal drop-down frame, which are respectively "single tone", "multitone", "narrow BPSK", "wide BPSK", "narrow guass", "pulse" and "sweet pilot". Wherein, selecting "singletone" will randomly generate tone interference; selecting multitone interference can be generated randomly by multitone; selecting "narrow BPSK" will randomly generate narrowband BPSK interference; selecting "wide BPSK" will randomly generate wideband BPSK interference; selecting "narrow guide" will generate narrow-band Gaussian interference randomly; the wideband Gaussian interference is randomly generated by selecting the 'wideband Guassion'; pulse interference is randomly generated by selecting 'pulse'; the choice of "sweepFre" will randomly generate linear swept-frequency interference.
Every time the selection is carried out in the interference signal drop-down frame, the generation program of the corresponding interference signal is operated again, key characteristic parameters such as interference frequency, code element width and the like corresponding to the generation program are randomly generated again, so that the interference signals generated after the selection are different every time, the richness of signal generation can be expanded, and the identification effect of the interference signals can be observed more scientifically and reasonably. The recognition effect is shown in fig. 5.1-5.8, and it can be seen that the interference signals can be recognized. Therefore, the system designed by the invention is effective and reasonable.
The foregoing shows and describes the general principles, principal features, and advantages of the invention. It will be understood by those skilled in the art that the present invention is not limited by the foregoing examples, and that the foregoing examples and descriptions are merely illustrative of the principles of the invention, and that various changes and modifications can be made without departing from the spirit and scope of the invention, which is intended to be within the scope of the invention as claimed. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (3)
1. The satellite navigation interference signal identification method based on the convolutional neural network is characterized by mainly comprising the following steps of:
s1, signal preprocessing: down-converting a one-dimensional satellite navigation typical interference signal to a baseband for processing, performing length truncation on a zero intermediate frequency signal, and repeatedly copying the one-dimensional satellite navigation typical interference signal and then converting the one-dimensional satellite navigation typical interference signal into a two-dimensional image form;
s2, constructing a convolutional neural network: the convolutional neural network comprises three convolutional layers, wherein the convolutional layers respectively comprise 32 convolutional layers of 3 x 3, 16 convolutional layers of 3 x 3 and 16 convolutional layers of 3 x 3, a normalization layer is introduced after each convolutional layer, a pooling layer is introduced after each normalization layer, and finally a full-connection layer is added to serve as an output layer, the convolutional layers select a ReLU to serve as an activation function, and the full-connection layer adopts softmax to serve as the activation function;
s3, training and classifying: and (3) inputting the two-dimensional image into the convolutional neural network in the step (S2), automatically extracting characteristics of the signal, finishing classification recognition through training, and finally outputting a recognition and sorting result of the interference signal.
2. The method according to claim 1, wherein in the step S3 of network training, the training database is generated by adjusting different waveform parameters of the interference signal, and is trained for different types of interference signals, and the training algorithm is sgdm algorithm.
3. The convolutional neural network-based satellite navigation jamming signal identifying apparatus using the method of claim 1, wherein: comprises a preprocessing module, a convolution neural network module and an output module, wherein the convolution neural network module is respectively connected with the preprocessing module and the output module,
the preprocessing module is used for converting the one-dimensional satellite navigation typical interference signal into a two-dimensional image;
the convolutional neural network module comprises three convolutional layers, wherein the convolutional layers respectively comprise 32 convolutional layers of 3 multiplied by 3, 16 convolutional layers of 3 multiplied by 3 and 16 convolutional layers of 3 multiplied by 3, a normalization layer is introduced after each convolutional layer, a pooling layer is introduced after each normalization layer, and finally a full-connection layer is added to serve as an output layer;
the output module is used for the training of the convolutional neural network module and the output of the classification result, and can display two-dimensional image input information, convolutional neural network simulation training information and recognition and sorting result information.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910772412.XA CN110515096B (en) | 2019-08-20 | 2019-08-20 | Convolutional neural network-based satellite navigation interference signal identification device and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910772412.XA CN110515096B (en) | 2019-08-20 | 2019-08-20 | Convolutional neural network-based satellite navigation interference signal identification device and method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110515096A CN110515096A (en) | 2019-11-29 |
| CN110515096B true CN110515096B (en) | 2023-03-31 |
Family
ID=68625926
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910772412.XA Active CN110515096B (en) | 2019-08-20 | 2019-08-20 | Convolutional neural network-based satellite navigation interference signal identification device and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110515096B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20210334626A1 (en) | 2020-04-28 | 2021-10-28 | Novatel Inc. | Gnss-receiver interference detection using deep learning |
| CN111562597A (en) * | 2020-06-02 | 2020-08-21 | 南京敏智达科技有限公司 | Beidou satellite navigation interference source identification method based on BP neural network |
| CN111783558A (en) * | 2020-06-11 | 2020-10-16 | 上海交通大学 | Satellite navigation interference signal type intelligent identification method and system |
| CN113970762B (en) * | 2021-09-26 | 2024-05-24 | 西安交通大学 | Multistage interference source positioning method and system |
| CN114254713B (en) * | 2022-02-28 | 2022-05-27 | 北京理工大学 | Classification system and method based on time-frequency transformation and dynamic mode decomposition |
| CN114675307B (en) * | 2022-03-14 | 2024-04-30 | 合肥工业大学 | Satellite navigation receiver interference detection method based on fractional order pseudo Wigner-Ville distribution |
| CN115563485A (en) * | 2022-09-14 | 2023-01-03 | 电子科技大学 | Low-complexity interference identification method based on deep learning |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108234370B (en) * | 2017-12-22 | 2020-12-15 | 西安电子科技大学 | Communication signal modulation mode identification method based on convolutional neural network |
| CN108229404B (en) * | 2018-01-09 | 2022-03-08 | 东南大学 | Radar echo signal target identification method based on deep learning |
| CN108616470A (en) * | 2018-03-26 | 2018-10-02 | 天津大学 | Modulation Signals Recognition method based on convolutional neural networks |
| CN108509911B (en) * | 2018-04-03 | 2020-06-12 | 电子科技大学 | Interference signal identification method based on convolutional neural network |
| CN109120563B (en) * | 2018-08-06 | 2020-12-29 | 电子科技大学 | Modulation recognition method based on neural network integration |
| CN109657604B (en) * | 2018-12-17 | 2021-02-12 | 中国人民解放军战略支援部队信息工程大学 | Satellite amplitude-phase signal identification and demodulation method and device based on cyclic neural network |
-
2019
- 2019-08-20 CN CN201910772412.XA patent/CN110515096B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN110515096A (en) | 2019-11-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110515096B (en) | Convolutional neural network-based satellite navigation interference signal identification device and method | |
| CN111964908B (en) | Bearing fault diagnosis method under variable working condition based on multi-scale weight distribution convolutional neural network model | |
| CN108573225B (en) | Partial discharge signal pattern recognition method and system | |
| CN110108992B (en) | Cable partial discharge fault identification method and system based on improved random forest algorithm | |
| CN110298085A (en) | Analog-circuit fault diagnosis method based on XGBoost and random forests algorithm | |
| CN105701470A (en) | Analog circuit fault characteristic extraction method based on optimal wavelet packet decomposition | |
| CN106874950A (en) | A kind of method for identifying and classifying of transient power quality recorder data | |
| CN110044623A (en) | The rolling bearing fault intelligent identification Method of empirical mode decomposition residual signal feature | |
| CN102760444A (en) | Support vector machine based classification method of base-band time-domain voice-frequency signal | |
| CN109918417B (en) | Time sequence data self-adaptive segmentation, dimension reduction and characterization method based on wavelet transformation and application | |
| CN110726898A (en) | Power distribution network fault type identification method | |
| CN109241849B (en) | Feature decomposition selection and fault diagnosis method for main engine of intelligent power plant steam turbine | |
| CN106570183A (en) | Color picture retrieval and classification method | |
| CN112101487B (en) | Compression method and device for fine-grained recognition model | |
| CN110020637A (en) | A kind of analog circuit intermittent fault diagnostic method based on more granularities cascade forest | |
| CN104156628A (en) | Ship radiation signal recognition method based on multi-kernel learning and discriminant analysis | |
| CN112817954A (en) | Missing value interpolation method based on multi-method ensemble learning | |
| CN116776075A (en) | Fan blade health monitoring system and monitoring method thereof | |
| CN116935892A (en) | Industrial valve anomaly detection method based on audio key feature dynamic aggregation | |
| CN103528820B (en) | A kind of Fault Diagnosis of Roller Bearings based on distance evaluation factor potential-energy function | |
| CN112836788A (en) | Low-power-consumption deep learning network method for interference type identification | |
| CN105023239A (en) | Hyperspectral data dimensionality reduction method based on ultra-pixel and maximum boundary distribution | |
| CN115276857B (en) | Full-blind spectrum detection method based on combination of Cholesky decomposition and convolutional neural network | |
| Cope et al. | Classifying plant leaves from their margins using dynamic time warping | |
| CN111539275B (en) | Electrical load classification method and system based on load characteristic visualization |
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 |