CN111639595A - Unmanned aerial vehicle micro-motion characteristic signal detection method based on weight-agnostic neural network - Google Patents
Unmanned aerial vehicle micro-motion characteristic signal detection method based on weight-agnostic neural network Download PDFInfo
- Publication number
- CN111639595A CN111639595A CN202010473731.3A CN202010473731A CN111639595A CN 111639595 A CN111639595 A CN 111639595A CN 202010473731 A CN202010473731 A CN 202010473731A CN 111639595 A CN111639595 A CN 111639595A
- Authority
- CN
- China
- Prior art keywords
- weight
- neural network
- agnostic
- signal
- cycle
- 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
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/42—Global feature extraction by analysis of the whole pattern, e.g. using frequency domain transformations or autocorrelation
- G06V10/431—Frequency domain transformation; Autocorrelation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Abstract
The invention discloses an unmanned aerial vehicle micro-motion characteristic signal detection method based on a weight-agnostic neural network, which is characterized by comprising the following steps of: 1) calculating a cyclic spectrum of the signal; 2) obtaining a cyclic spectrum contour map through MATLAB processing, and selecting an observation region; 3) training a weight-agnostic neural network; 4) and performing micro-motion feature recognition by using the trained weight unknown neural network. The method has good anti-interference performance, the neural network is simpler in structure, the calculated amount is smaller, and the accuracy of identifying the micro characteristic signals of the unmanned aerial vehicle is higher.
Description
Technical Field
The invention relates to the field of unmanned aerial vehicle signal detection, in particular to an unmanned aerial vehicle micro-motion characteristic signal detection method based on a weight unknown neural network.
Background
The wide popularization of unmanned aerial vehicles has driven the development and the transformation of relevant trades, but because legal system, market system are not sound enough, and the influence of other factors, unmanned aerial vehicle disturbs the navigation, threatens public safety's incident also frequently to take place, has caused comparatively abominable influence, so to the effectual supervision and the standard of unmanned aerial vehicle delay. The rotation of unmanned aerial vehicle rotor can carry out the micro-motion modulation to the electromagnetic wave signal of scattering, but unmanned aerial vehicle belongs to "low little slowly" target, and the little pulier frequency that its motion produced is less, and the micro-motion characteristic signal is comparatively faint, is difficult to be observed.
In recent years, a neural network has been developed in a breakthrough in the field of feature recognition and the like, and some scholars have applied the neural network to modulation recognition of communication signals. After the neural network is trained, the features which can not be observed by human eyes can be found, so that the aim of signal identification is fulfilled.
The idea that a weight-agnostic neural network can perform tasks without any explicit weight training is to find an innate neural network architecture and then only need a single shared weight that is randomly initialized to perform the task.
Disclosure of Invention
The invention aims to provide an unmanned aerial vehicle micro-motion signal detection method based on a weight agnostic neural network, aiming at the problem that micro-motion characteristic signals are difficult to observe. The method has the advantages of good anti-interference performance, simpler structure of the neural network, smaller calculated amount and higher accuracy of identifying the micro-motion signals of the unmanned aerial vehicle.
The technical scheme for realizing the purpose of the invention is as follows:
the unmanned aerial vehicle micro-motion signal detection method based on the weight unknown neural network is different from the prior art and comprises the following steps:
1) calculating a cyclic spectrum of the signal: signal x (t) is a generalized cyclostationary process, and x (t) is the autocorrelation function:
Rx(t,τ)＝E[x(t)x*(t-τ)](1)，
due to Rx(T, τ) is a periodic function of the period of time T, thus for Rx(t, τ) performing Fourier series expansion to obtain the following formula:
in the formula: m/T, called the cycle frequency of the signal x (T), 1/T being the basis of the cycle frequency, the coefficients of its Fourier series being:
in the formula:
is the cyclic autocorrelation function of the signal x (t), which is a function of both the cyclic frequency α and the time interval tau, and of the signal x (t)
Obtaining a cyclic spectrum density function of the signal, namely a cyclic spectrum, by calculating Fourier transform, wherein the expression is as follows:
2) obtaining an isocontour of the cyclic spectrum: the cycle spectrogram is the representation of the cycle stationarity and the energy distribution condition of the signal, so the cycle spectrograms of different signals have difference, the contour map of the cycle spectrogram is formed by connecting points with the same value, is a two-dimensional plane map and is a plane representation map representing the distribution of the signal energy, the frequency and the cycle frequency, and the process of obtaining the contour map of the cycle spectrogram is as follows:
2-1) drawing a cyclic spectrum contour map of the FM signals by using a countourf function in an MATLAB tool;
2-2) in order to reduce the influence of noise, selecting the uppermost area in the cyclic spectrum contour map as an observation target;
2-3) preprocessing the contour map in the step 2-2): graying is carried out firstly, and then the size of the picture is modified to 28 multiplied by 28;
3) training the weight-agnostic neural network: training the weight-agnostic neural network includes:
3-1) initialization: determining the number of input neurons of the weight-agnostic neural network topological structure according to the pixel points of the picture preprocessed in the step 2-3), determining the number of output neurons according to the number of recognition targets, and finally randomly generating a series of initial weight-agnostic neural networks only with simple links;
3-2) evaluation: setting a single shared weight for each weight-agnostic neural network, then evaluating each weight-agnostic neural network, wherein the evaluation is divided into two parts of performance and structural complexity, and evaluating the performance, wherein a series of fixed weight values are firstly required to be set for the weight-agnostic neural network: ([ -2, -1, -0.5, +0.5, +1, +2]), then the neural network discerns the classification to training set picture under different weighted values, count the accuracy rate of discerning and take the mean value of all results finally, the mean value is the performance and grade, the structural complexity is judged according to the unknown neural network topological structure of the weight, including the number of hidden layers of the topological structure, the number of nodes and the number of connections between the nodes, the more complicated of their quantity, the lower the complexity the grade is higher;
3-3) sorting: sequencing the weight-agnostic neural network structures according to the evaluation scores, and reserving the weight-agnostic neural network with the evaluation score ranked at the top;
3-4) variations: changing the reserved network topology structure to create a new group by adopting a mode of inserting nodes into the weight unknown neural network, adding new connections among the nodes and changing an activation function of the nodes, and then repeating the steps 3-2) -3-4) to determine the finally trained weight unknown neural network;
4) and (3) performing micro-motion feature recognition by adopting the trained weight-agnostic neural network, namely inputting the picture preprocessed in the step 2-3) into the weight-agnostic neural network trained in the step 3), recording a corresponding signal recognition rate, and then improving the recognition rate by finely adjusting the weight value of the neural network.
Compared with the existing unmanned aerial vehicle micro-motion characteristic signal detection method, the technical scheme has the following characteristics:
the cyclic spectrum has strong anti-noise performance and also contains abundant signal characteristics, thereby being beneficial to extracting characteristic parameters; secondly, the cyclic spectrum contour map of the signal is a two-dimensional contour plane of the cyclic spectrum, the contour maps of different modulation signals have certain difference in energy distribution and shape, and the remaining 80% of the regions of the contour map except the signal characteristic information are blank, so that good sparse characteristics are brought to the subsequent training of the neural network; the weight-agnostic neural network does not need to carry out complicated parameter adjustment work, and the neural network model with excellent performance can be obtained only by searching the neural network framework which is the most suitable for the current task.
The method has the advantages of good anti-interference performance, simpler structure of the neural network, smaller calculated amount and higher accuracy of identifying the micro characteristic signal of the unmanned aerial vehicle.
Drawings
FIG. 1 is a schematic flow chart of the method in the example;
FIG. 2 is a schematic diagram of a cyclic spectrum of an FM signal in an embodiment;
FIG. 3 (a) is a schematic diagram showing an isometric view of a cyclic spectrum of an FM signal in an embodiment;
FIG. 3 (b) is a schematic view of an observation region of an isocontour in the example;
FIG. 4 is a schematic flow chart of an embodiment of training a weight-agnostic neural network;
fig. 5 is a schematic diagram of a trained neural network topology in the embodiment.
Detailed Description
The invention will be further illustrated, but not limited, by the following description of the embodiments with reference to the accompanying drawings.
Example (b):
referring to fig. 1, the unmanned aerial vehicle jiggle feature signal detection method based on the weight-agnostic neural network comprises the following steps:
1) calculating a cyclic spectrum of the signal: as shown in fig. 2, taking FM signal x (t) as an example, the signal is a generalized cyclostationary process, and its autocorrelation function is:
Rx(t,τ)＝E[x(t)x*(t-τ)](1)，
due to Rx(T, τ) is a periodic function of the period of time T, thus for Rx(t, τ) performing Fourier series expansion to obtain the following formula:
in the formula: m/T, called the cycle frequency of the signal x (T), 1/T being the basis of the cycle frequency, the coefficients of its Fourier series being:
in the formula:
is the cyclic autocorrelation function of the signal x (t), which is a function of both the cyclic frequency α and the time interval tau, and of the signal x (t)
Fourry findingAnd (3) obtaining a cyclic spectrum density function, namely a cyclic spectrum, of the signal through leaf standing transformation, wherein the expression is as follows:
2) obtaining an isocontour of the cyclic spectrum: the cycle spectrogram is the representation of the cycle stationarity and the energy distribution condition of the signal, so the cycle spectrograms of different signals have difference, the contour map of the cycle spectrogram is formed by connecting points with the same value, is a two-dimensional plane map and is a plane representation map representing the distribution of the signal energy, the frequency and the cycle frequency, and the process of obtaining the contour map of the cycle spectrogram is as follows:
2-1) as shown in (a) of fig. 3, a cyclic spectrum contour map of the FM frequency modulated signal is plotted using a countourf function in the MATLAB tool;
2-2) as shown in (b) of fig. 3, in order to reduce the influence of noise, the uppermost region in the cyclic spectrum contour map is selected as an observation target;
2-3) preprocessing the contour map in the step 2-2): graying is carried out firstly, and then the size of the picture is modified to 28 multiplied by 28;
3) training the weight-agnostic neural network: as shown in fig. 4, training the weight-agnostic neural network includes:
3-1) initialization: determining the number of input neurons of the weight-agnostic neural network topological structure according to the pixel points of the picture preprocessed in the step 2-3), determining the number of output neurons according to the number of recognition targets, and finally randomly generating a series of initial weight-agnostic neural networks only with simple links;
3-2) evaluation: setting a single shared weight for each weight-agnostic neural network, then evaluating each weight-agnostic neural network, wherein the evaluation is divided into two parts, namely performance and structural complexity, evaluating the performance, firstly setting a series of fixed weight values ([ -2, -1, -0.5, +0.5, +1, +2]) for the weight-agnostic neural network, then identifying and classifying the training set pictures by the neural network under different weight values, finally counting the identification accuracy and taking the average value of all results, wherein the average value is the performance score, the structural complexity is judged according to the topological structure of the weight-agnostic neural network, and comprises the number of hidden layers of the topological structure, the number of nodes and the number of connections among the nodes, the more the number of the hidden layers, the more the complexity and the lower the complexity, the higher the score;
3-3) sorting: sequencing the weight-agnostic neural network structures according to the evaluation scores, and reserving the weight-agnostic neural network with the evaluation score ranked at the top;
3-4) variations: changing the reserved network topology structure to create a new group by adopting modes of inserting nodes into the weight unknown neural network, adding new connections among the nodes, changing an activation function of the nodes and the like, and then repeating the steps 3-2) -3-4) to determine the finally trained weight unknown neural network;
4) and (3) performing micro-motion feature recognition by adopting the trained weight-agnostic neural network, namely inputting the picture preprocessed in the step 2-3) into the weight-agnostic neural network trained in the step 3), recording a corresponding signal recognition rate, and then improving the recognition rate by finely adjusting the weight value of the neural network.
Claims (1)
1. The unmanned aerial vehicle micro-motion characteristic signal detection method based on the weight-agnostic neural network is characterized by comprising the following steps of:
1) calculating a cyclic spectrum of the signal: signal x (t) is a generalized cyclostationary process, and x (t) is the autocorrelation function:
Rx(t,τ)＝E[x(t)x*(t-τ)](1)，
due to Rx(T, τ) is a periodic function of the period of time T, thus for Rx(t, τ) performing Fourier series expansion to obtain the following formula:
in the formula: m/T, called the cycle frequency of the signal x (T), 1/T being the basis of the cycle frequency, the coefficients of its Fourier series being:
in the formula:
is the cyclic autocorrelation function of the signal x (t), which is a function of both the cyclic frequency α and the time interval tau, and of the signal x (t)
Obtaining a cyclic spectrum density function of the signal, namely a cyclic spectrum, by calculating Fourier transform, wherein the expression is as follows:
2) obtaining an isocontour of the cyclic spectrum: the cycle spectrogram is the representation of the cycle stationarity and the energy distribution condition of the signal, so the cycle spectrograms of different signals have difference, the contour map of the cycle spectrogram is formed by connecting points with the same value, is a two-dimensional plane map and is a plane representation map representing the distribution of the signal energy, the frequency and the cycle frequency, and the process of obtaining the contour map of the cycle spectrogram is as follows:
2-1) drawing a cyclic spectrum contour map of the FM signals by using a countourf function in an MATLAB tool;
2-2) selecting the uppermost area in the cyclic spectrum contour map as an observation target;
2-3) preprocessing the contour map in the step 2-2): graying is carried out firstly, and then the size of the picture is modified to 28 multiplied by 28;
3) training the weight-agnostic neural network: training the weight-agnostic neural network includes:
3-1) initialization: determining the number of input neurons of the weight-agnostic neural network topological structure according to the pixel points of the picture preprocessed in the step 2-3), determining the number of output neurons according to the number of recognition targets, and finally randomly generating a series of initial weight-agnostic neural networks only with simple links;
3-2) evaluation: setting a single shared weight for each weight-agnostic neural network, then evaluating each weight-agnostic neural network, wherein the evaluation is divided into two parts of performance and structural complexity, and evaluating the performance, wherein a series of fixed weight values are firstly required to be set for the weight-agnostic neural network: ([ -2, -1, -0.5, +0.5, +1, +2]), then the neural network discerns the classification to training set picture under different weighted values, count the accuracy rate of discerning and take the mean value of all results finally, the mean value is the performance and grade, the structural complexity is judged according to the unknown neural network topological structure of the weight, including the number of hidden layers of the topological structure, the number of nodes and the number of connections between the nodes, the more complicated of their quantity, the lower the complexity the grade is higher; 3-3) sorting: sequencing the weight-agnostic neural network structures according to the evaluation scores, and reserving the weight-agnostic neural network with the evaluation score ranked at the top;
3-4) variations: changing the reserved network topology structure to create a new group by inserting nodes into the weight-agnostic neural network, adding new connections among the nodes, changing an activation function of the nodes and the like, and then repeating the steps 3-2) -3-4) to determine the finally trained weight-agnostic neural network;
4) and (3) performing micro-motion feature recognition by adopting the trained weight-agnostic neural network, namely inputting the picture preprocessed in the step 2-3) into the weight-agnostic neural network trained in the step 3), recording a corresponding signal recognition rate, and then improving the recognition rate by finely adjusting the weight value of the neural network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010473731.3A CN111639595B (en) | 2020-05-29 | 2020-05-29 | Unmanned aerial vehicle micro-motion characteristic signal detection method based on weight-agnostic neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010473731.3A CN111639595B (en) | 2020-05-29 | 2020-05-29 | Unmanned aerial vehicle micro-motion characteristic signal detection method based on weight-agnostic neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111639595A true CN111639595A (en) | 2020-09-08 |
| CN111639595B CN111639595B (en) | 2022-03-18 |
Family
ID=72331253
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010473731.3A Active CN111639595B (en) | 2020-05-29 | 2020-05-29 | Unmanned aerial vehicle micro-motion characteristic signal detection method based on weight-agnostic neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111639595B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112689288A (en) * | 2020-12-18 | 2021-04-20 | 桂林电子科技大学 | Radio frequency fingerprint extraction and identification method based on WANN |
Citations (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105488466A (en) * | 2015-11-26 | 2016-04-13 | 中国船舶工业系统工程研究院 | Deep neural network and underwater sound target vocal print feature extraction method |
| US20160171896A1 (en) * | 2014-12-12 | 2016-06-16 | Amazon Technologies, Inc. | Commercial and General Aircraft Avoidance using Light Pattern Detection |
| CN105720515A (en) * | 2016-02-03 | 2016-06-29 | 国网福建省电力有限公司 | Flight barrier-clearing unmanned aerial vehicle |
| US20160337027A1 (en) * | 2015-05-13 | 2016-11-17 | Ubiqomm Llc | Ground terminal and gateway beam pointing toward an unmanned aerial vehicle (uav) for network access |
| CN106530281A (en) * | 2016-10-18 | 2017-03-22 | 国网山东省电力公司电力科学研究院 | Edge feature-based unmanned aerial vehicle image blur judgment method and system |
| CN106707258A (en) * | 2017-03-03 | 2017-05-24 | 电子科技大学 | Multi-parameter estimation method for micro-motion target under non-Gaussian background |
| CN107192341A (en) * | 2017-07-27 | 2017-09-22 | 广州吉越中蜂创新科技有限公司 | A kind of height measuring device |
| CN107480395A (en) * | 2017-08-29 | 2017-12-15 | 燕山大学 | A kind of construction method and system of vehicle steering knuckle loading spectrum forecast model |
| CN107576387A (en) * | 2017-08-18 | 2018-01-12 | 浙江大学 | A kind of unmanned plane detection method based on the identification of vocal print multiple-harmonic |
| CN107977987A (en) * | 2017-11-20 | 2018-05-01 | 北京理工大学 | A kind of UAV system multiple target detection tracking, instruction system and method |
| CN108255198A (en) * | 2017-12-28 | 2018-07-06 | 广州亿航智能技术有限公司 | Camera cradle head control system and control method under a kind of unmanned plane during flying state |
| US20180211179A1 (en) * | 2017-01-23 | 2018-07-26 | DGS Global Systems, Inc. | Systems, methods, and devices for automatic signal detection with temporal feature extraction within a spectrum |
| CN108871409A (en) * | 2018-05-02 | 2018-11-23 | 北京畅东科技有限公司 | A kind of fault detection method and system |
| CN109614930A (en) * | 2018-12-11 | 2019-04-12 | 湖南华诺星空电子技术有限公司 | A kind of unmanned plane spectrum detection method based on deep learning |
| CN109802905A (en) * | 2018-12-27 | 2019-05-24 | 西安电子科技大学 | Digital signal Automatic Modulation Recognition method based on CNN convolutional neural networks |
| CN110321953A (en) * | 2019-07-03 | 2019-10-11 | 中山大学 | Deep learning intelligent modulation recognition methods based on circulation Power estimation |
| CN111050315A (en) * | 2019-11-27 | 2020-04-21 | 北京邮电大学 | Wireless transmitter identification method based on multi-core two-way network |
| CN111079556A (en) * | 2019-11-25 | 2020-04-28 | 航天时代飞鸿技术有限公司 | Multi-temporal unmanned aerial vehicle video image change area detection and classification method |
-
2020
- 2020-05-29 CN CN202010473731.3A patent/CN111639595B/en active Active
Patent Citations (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160171896A1 (en) * | 2014-12-12 | 2016-06-16 | Amazon Technologies, Inc. | Commercial and General Aircraft Avoidance using Light Pattern Detection |
| US20160337027A1 (en) * | 2015-05-13 | 2016-11-17 | Ubiqomm Llc | Ground terminal and gateway beam pointing toward an unmanned aerial vehicle (uav) for network access |
| CN105488466A (en) * | 2015-11-26 | 2016-04-13 | 中国船舶工业系统工程研究院 | Deep neural network and underwater sound target vocal print feature extraction method |
| CN105720515A (en) * | 2016-02-03 | 2016-06-29 | 国网福建省电力有限公司 | Flight barrier-clearing unmanned aerial vehicle |
| CN106530281A (en) * | 2016-10-18 | 2017-03-22 | 国网山东省电力公司电力科学研究院 | Edge feature-based unmanned aerial vehicle image blur judgment method and system |
| US20180211179A1 (en) * | 2017-01-23 | 2018-07-26 | DGS Global Systems, Inc. | Systems, methods, and devices for automatic signal detection with temporal feature extraction within a spectrum |
| CN106707258A (en) * | 2017-03-03 | 2017-05-24 | 电子科技大学 | Multi-parameter estimation method for micro-motion target under non-Gaussian background |
| CN107192341A (en) * | 2017-07-27 | 2017-09-22 | 广州吉越中蜂创新科技有限公司 | A kind of height measuring device |
| CN107576387A (en) * | 2017-08-18 | 2018-01-12 | 浙江大学 | A kind of unmanned plane detection method based on the identification of vocal print multiple-harmonic |
| CN107480395A (en) * | 2017-08-29 | 2017-12-15 | 燕山大学 | A kind of construction method and system of vehicle steering knuckle loading spectrum forecast model |
| CN107977987A (en) * | 2017-11-20 | 2018-05-01 | 北京理工大学 | A kind of UAV system multiple target detection tracking, instruction system and method |
| CN108255198A (en) * | 2017-12-28 | 2018-07-06 | 广州亿航智能技术有限公司 | Camera cradle head control system and control method under a kind of unmanned plane during flying state |
| CN108871409A (en) * | 2018-05-02 | 2018-11-23 | 北京畅东科技有限公司 | A kind of fault detection method and system |
| CN109614930A (en) * | 2018-12-11 | 2019-04-12 | 湖南华诺星空电子技术有限公司 | A kind of unmanned plane spectrum detection method based on deep learning |
| CN109802905A (en) * | 2018-12-27 | 2019-05-24 | 西安电子科技大学 | Digital signal Automatic Modulation Recognition method based on CNN convolutional neural networks |
| CN110321953A (en) * | 2019-07-03 | 2019-10-11 | 中山大学 | Deep learning intelligent modulation recognition methods based on circulation Power estimation |
| CN111079556A (en) * | 2019-11-25 | 2020-04-28 | 航天时代飞鸿技术有限公司 | Multi-temporal unmanned aerial vehicle video image change area detection and classification method |
| CN111050315A (en) * | 2019-11-27 | 2020-04-21 | 北京邮电大学 | Wireless transmitter identification method based on multi-core two-way network |
Non-Patent Citations (1)
| Title |
|---|
| 刘宜成 等: ""基于轨迹和形态识别的无人机检测方法"", 《计算机工程》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112689288A (en) * | 2020-12-18 | 2021-04-20 | 桂林电子科技大学 | Radio frequency fingerprint extraction and identification method based on WANN |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111639595B (en) | 2022-03-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101587186B (en) | Characteristic extraction method of radar in-pulse modulation signals | |
| CN105550636A (en) | Method and device for identifying target types | |
| CN111220958A (en) | Radar target Doppler image classification and identification method based on one-dimensional convolutional neural network | |
| CN112257521B (en) | CNN underwater acoustic signal target identification method based on data enhancement and time-frequency separation | |
| CN107301381A (en) | Recognition Method of Radar Emitters based on deep learning and multi-task learning strategy | |
| CN105679313A (en) | Audio recognition alarm system and method | |
| CN103994820B (en) | A kind of moving target recognition methods based on micropore diameter microphone array | |
| CN108478216A (en) | A kind of epileptic seizure intelligent Forecasting early period based on convolutional neural networks | |
| CN107064894A (en) | A kind of clutter suppression method based on deep learning | |
| CN113488060B (en) | Voiceprint recognition method and system based on variation information bottleneck | |
| CN109884591A (en) | A kind of multi-rotor unmanned aerial vehicle acoustical signal Enhancement Method based on microphone array | |
| CN111639595B (en) | Unmanned aerial vehicle micro-motion characteristic signal detection method based on weight-agnostic neural network | |
| CN109519340A (en) | A kind of wind turbine generator drive system method for diagnosing faults | |
| CN113640768B (en) | Low-resolution radar target identification method based on wavelet transformation | |
| CN111368930B (en) | Radar human body posture identification method and system based on multi-class spectrogram fusion and hierarchical learning | |
| CN112859012A (en) | Radar deception jamming identification method based on cascade convolution neural network | |
| CN110390949A (en) | Acoustic Object intelligent identification Method based on big data | |
| CN115712867A (en) | Multi-component radar signal modulation identification method | |
| CN114626412B (en) | Multi-class target identification method and system for unattended sensor system | |
| CN115293214A (en) | Underwater sound target recognition model optimization method based on sample expansion network | |
| CN111968671B (en) | Low-altitude sound target comprehensive identification method and device based on multidimensional feature space | |
| CN114724245A (en) | CSI-based incremental learning human body action identification method | |
| CN115064182A (en) | Fan fault feature identification method of self-adaptive Mel filter in strong noise environment | |
| CN110458219B (en) | phi-OTDR vibration signal identification algorithm based on STFT-CNN-RVFL | |
| CN111274989A (en) | Deep learning-based field vehicle identification method |
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 |