CN113095137B - Signal characteristic recognition device and method based on machine learning - Google Patents

Signal characteristic recognition device and method based on machine learning Download PDF

Info

Publication number
CN113095137B
CN113095137B CN202110259492.6A CN202110259492A CN113095137B CN 113095137 B CN113095137 B CN 113095137B CN 202110259492 A CN202110259492 A CN 202110259492A CN 113095137 B CN113095137 B CN 113095137B
Authority
CN
China
Prior art keywords
signal
module
digital
machine learning
identifying
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
CN202110259492.6A
Other languages
Chinese (zh)
Other versions
CN113095137A (en
Inventor
俞春华
陈毅豪
张周
王琼
李瑞婧
张兴敢
柏业超
唐岚
吉志海
高春永
徐鹏
赵润森
蒋品
付军峰
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
Original Assignee
Nanjing University
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 filed Critical Nanjing University
Priority to CN202110259492.6A priority Critical patent/CN113095137B/en
Publication of CN113095137A publication Critical patent/CN113095137A/en
Application granted granted Critical
Publication of CN113095137B publication Critical patent/CN113095137B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

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
    • G06F2218/04—Denoising
    • 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
    • G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine 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
    • G06N3/044—Recurrent networks, e.g. Hopfield 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
    • G06N3/04—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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/047—Probabilistic or stochastic 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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/048—Activation functions
    • 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
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Software Systems (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Probability & Statistics with Applications (AREA)
  • Medical Informatics (AREA)
  • Signal Processing (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Digital Transmission Methods That Use Modulated Carrier Waves (AREA)

Abstract

The invention belongs to the technical field of signal feature recognition, in particular to a signal feature recognition device and a signal feature recognition method based on machine learning.

Description

Signal characteristic recognition device and method based on machine learning
Technical Field
The invention belongs to the technical field of signal feature recognition, and particularly relates to a signal feature recognition device and method based on machine learning.
Background
Aiming at the existing high-order cumulant method, the characteristic parameters are required to be designed manually, the characteristic parameters are difficult to construct for signals such as 16QAM, 64QAM and the like, the existing deep learning method has higher time and space complexity, and the implementation in hardware is difficult.
Disclosure of Invention
The invention aims to solve the technical problems that a machine learning method is adopted to automatically extract the characteristics of signals and then identify the digital modulation mode of the signals, and meanwhile, the operation complexity is reduced, so that the method is convenient to realize in hardware.
A signal characteristic recognition device based on machine learning comprises a radio frequency signal processing module, a digital signal processing and control module and a network interface module which are connected in sequence;
the radio frequency signal processing module is used for receiving, filtering and amplifying signals and outputting the signals to the digital signal processing and controlling module, and the digital signal processing and controlling module is used for identifying signal characteristics and identifying signal digital modulation modes of the received signals and outputting the identified signal characteristics and the identified signal digital modulation modes to the outside through the network interface module.
The radio frequency signal processing module comprises a filter circuit and a low noise amplifying circuit, wherein the filter circuit filters signals firstly and then sends the signals to the low noise amplifying circuit for amplifying, and the amplified signals are sent to the digital signal processing and controlling module.
Preferably, the digital signal processing and controlling module comprises an ADC module and a signal characteristic identifying module; the signal characteristic recognition module comprises a GPU processing module for recognizing the signal type and an FPGA processing module for recognizing the digital modulation mode, wherein the GPU processing module comprises a down-conversion module and a GPU chip, and the FPGA processing module comprises ZYNQ series chips.
A signal characteristic recognition method based on machine learning comprises a signal type recognition method and a signal digital modulation mode recognition method. The signal type identification method comprises the following steps:
step 1, performing down-conversion treatment on a baseband signal subjected to analog-to-digital conversion treatment in a down-conversion module;
step 2, performing median filtering on the signals subjected to the down-conversion treatment, and performing short-time Fourier transform on the signals subjected to the median filtering to obtain a time-frequency characteristic diagram; converting the time-frequency characteristic diagram into a gray scale diagram;
and 3, identifying the gray level map by adopting a trained CNN model, and outputting a result through a softmax classifier.
The GPU chip comprises a computer program which is used for executing the steps 2-3.
The identification method of the signal digital modulation mode comprises the following steps:
step 1, performing median filtering on a baseband signal subjected to analog-to-digital conversion;
step 2, mapping the signals subjected to median filtering processing to a constellation plane, normalizing the constellation in a single-channel gray scale pattern form, and inputting the normalized signal into a VGG-9 convolutional neural network for extracting features;
the VGG-9 convolutional neural network comprises 6 convolutional layers, a pooling layer is added after every two convolutional layers, the spatial features extracted by the convolutional layer are converted into one-dimensional feature vectors through a full-connection layer after the last convolutional layer, and the one-dimensional feature vectors identify the digital modulation mode of signals through two full-connection layers and a softmax function.
And 3, identifying a signal modulation mode through a softmax classifier.
The ZYNQ series chip comprises a computer program, wherein the computer program is used for executing the steps 2-3.
The beneficial effects are that: the invention adopts a machine learning method, can identify the type and the digital modulation mode of signals, realizes the identification of various signal characteristics, aims at the problems that the traditional method in the signal identification is difficult to manually extract characteristic parameters, the existing deep learning method is high in calculation complexity and the like, carries out median filtering treatment on the signals, eliminates isolated noise points, reduces the interference of Gaussian white noise on the signals, and simultaneously reduces the calculation complexity. And a device for signal characteristic recognition is designed, and the requirements of processing performance diversification are met by adopting a hardware combination form.
Drawings
FIG. 1 is a block diagram of the apparatus of the present invention;
FIG. 2 is a schematic diagram of a signal type identification flow;
FIG. 3 is a schematic diagram of a digital modulation scheme identification process;
FIG. 4 is a CNN model training flow;
FIG. 5 is a block diagram of a VGG-9 convolutional neural network.
Detailed Description
The invention provides a signal characteristic recognition device and a signal characteristic recognition method based on machine learning, as shown in fig. 1, the signal characteristic recognition device based on machine learning comprises a radio frequency signal processing module, a digital signal processing and control module and a network interface module, wherein the radio frequency signal processing module is used for receiving, filtering and amplifying a wide-range baseband signal, the digital signal processing and control module is used for recognizing signal characteristics of the received signal, and the recognized signal characteristics are transmitted to an external communication module through the network interface module.
The radio frequency signal processing module comprises a filter circuit and a low noise amplifying circuit, wherein the filter circuit filters signals firstly and then sends the signals to the low noise amplifying circuit for amplifying, and the amplified signals are sent to the digital signal processing and controlling module.
The digital signal processing and controlling module comprises an ADC module and a signal characteristic identifying module;
the signal input to the digital signal processing and controlling module is converted into a digital signal through the ADC module, the signal type and the digital modulation mode of the signal are identified in the signal characteristic identifying module, and the identified signal type and the digital modulation mode are transmitted to an external communication module through the network interface module.
The signal characteristic recognition module comprises a GPU processing module and an FPGA processing module.
The GPU processing module comprises a down-conversion module and a GPU chip, wherein in the digital signal processing and control module taking a GPU processor as a core, the down-conversion module performs down-conversion processing to generate I, Q signals, performs signal characteristic identification in the GPU chip and can display signal characteristics through a display interface.
The ADC module adopts a wide-band and high-integration ADC chip, the down-conversion module adopts an FPGA chip for processing, and the GPU chip can be an RTX series GPU chip.
In the GPU processing module, various deep learning frames and neural networks with different layers can be flexibly selected, the requirements of mass storage equipment, parallel computing and low-precision computing are met, meanwhile, brand-new architecture and extremely fast computing nodes are adopted, fewer fast nodes are utilized to obtain higher performance, and the throughput of a data center is greatly improved.
The FPGA processing module comprises a ZYNQ series chip, the ZYNQ series chip of Xilinx company consists of a dual ARM core Processing System (PS) and Programmable Logic (PL), and the dual ARM cores of the PS system are respectively provided with Linux and RTOS operating systems. The Linux system runs a network layer and application programs; the RTOS operating system runs a signal characteristic identification and frequency control strategy, and the signal characteristic identification and frequency control strategy is written in the C/C++ language; the PL layer deploys a signal characteristic recognition program and is written by using verilog HDL language. A radio frequency chip AD9361 is selected, and the adjustable bandwidth range of the chip receiving signals is 200 kHz-56 MHz.
The signal characteristic recognition method based on machine learning comprises a signal type recognition method and a signal digital modulation mode recognition method.
As shown in fig. 2, the signal type identification method includes the following steps:
step 1, performing down-conversion treatment on a baseband signal subjected to analog-to-digital conversion treatment in a down-conversion module;
step 2, performing median filtering on the signals subjected to the down-conversion treatment, and performing short-time Fourier transform on the signals subjected to the median filtering to obtain a time-frequency characteristic diagram; converting the time-frequency characteristic diagram into a gray scale diagram;
and 3, identifying the gray level map by adopting a trained CNN model, and outputting a signal characteristic identification result by a softmax classifier. The CNN model comprises a convolution layer, a pooling layer and a full connection layer.
As shown in fig. 4, the training process of the CNN model includes two processes of forward propagation and backward propagation. The forward propagation process is to input the preprocessed data into a convolutional neural network, extract the data features of each category to form feature vectors, and output classification results through a softmax classifier. The back propagation process is to calculate the difference between the output result of the network and the label value through the loss function, and update the parameters such as convolution kernel, weight, deviation and the like in the network by using a gradient descent algorithm according to a chained derivative rule. And continuously iterating until the difference between the output result and the label value is converged. And after the trained CNN model parameters are stored, calling the CNN model to identify the signal characteristics.
As shown in fig. 3, the method for identifying the digital modulation scheme of the signal includes the following steps:
step 1, performing median filtering on a baseband signal subjected to analog-to-digital conversion processing to eliminate isolated noise points so as to reduce interference of Gaussian white noise on the signal;
step 2, mapping the signals subjected to median filtering processing to a constellation plane, normalizing the constellation in a single-channel gray scale pattern form, and inputting the normalized signal into a VGG-9 convolutional neural network for extracting features;
and 3, identifying a signal modulation mode through a softmax classifier.
The VGG-9 convolutional neural network structure is shown in figure 5. The network is built based on the idea that a VGG-16 convolutional neural network structure adopts a plurality of small convolutional kernels to replace a plurality of layers of large convolutional kernels. And 6 layers of convolution layers are adopted, and a pooling layer is added after every two layers of convolution layers. The feature image output by the last convolution layer is converted into a one-dimensional feature vector, and the one-dimensional feature vector identifies the digital modulation mode of the signal through the two full-connection layers and the softmax classifier. The VGG-9 convolutional neural network further comprises a Relu function, and the Relu function with the advantage of gradient disappearance prevention is adopted as an activation function of each convolutional layer. Because the parameters of the full-connection layer are more, a Dropout is added after each full-connection layer, and some neurons are randomly ignored in each round of training process, so that the model is prevented from being overfitted.
The invention mainly adopts a machine learning method and combines a hardware device to realize the characteristic recognition function of the wireless signal. According to different processing capacities, a hardware combination form of two kernel row modules, namely an FPGA and a GPU is adopted. Aiming at the type identification of the wireless signals, after median filtering is carried out on the baseband signals after down-conversion, short-time Fourier transform is carried out on the signals, a time-frequency characteristic diagram is obtained, the time-frequency characteristic diagram is converted into a gray diagram, and the gray diagram is sent to CNN for identification. Aiming at the problems that the traditional method for manually extracting characteristic parameters is difficult, the existing deep learning method is high in computational complexity and the like in the digital modulation mode identification, a digital modulation mode identification algorithm combining constellation diagram optimization and low-operand VGG-9 is provided. Firstly, carrying out median filtering processing on signals to eliminate isolated noise points so as to reduce interference of Gaussian white noise on the signals; then mapping the signals to a constellation plane, normalizing the constellation in the form of a single-channel gray scale map, and inputting the normalized constellation into a VGG-9 convolutional neural network to extract features; finally, the signal modulation mode is identified through a softmax classifier.

Claims (6)

1. The signal characteristic recognition method based on machine learning is characterized by comprising a signal type recognition method and a signal digital modulation mode recognition method;
the signal type identification method comprises the following steps:
step 1, performing down-conversion treatment on a baseband signal subjected to analog-to-digital conversion treatment in a down-conversion module;
step 2, performing median filtering on the signals subjected to the down-conversion treatment, and performing short-time Fourier transform on the signals subjected to the median filtering to obtain a time-frequency characteristic diagram; converting the time-frequency characteristic diagram into a gray scale diagram;
and 3, identifying the gray level map by adopting a trained CNN model, and outputting a signal type identification result through a softmax classifier.
2. The method for identifying signal features based on machine learning according to claim 1, wherein the method for identifying the digital modulation scheme of the signal comprises the steps of:
step 1, performing median filtering on a baseband signal subjected to analog-to-digital conversion;
step 2, mapping the signals subjected to median filtering processing to a constellation plane, normalizing the constellation in a single-channel gray scale pattern form, and inputting the normalized signal into a VGG-9 convolutional neural network for extracting features;
and 3, identifying a signal modulation mode through a softmax classifier.
3. The machine learning-based signal feature recognition method of claim 2, wherein the VGG-9 convolutional neural network comprises 6 convolutional layers, a pooling layer is added after each two convolutional layers, the spatial features extracted by the convolutional layer are converted into one-dimensional feature vectors through a full connection layer after the last convolutional layer, and the one-dimensional feature vectors recognize the digital modulation mode of the signal through two full connection layers and a softmax function.
4. The signal characteristic recognition device based on machine learning is characterized by comprising a radio frequency signal processing module, a digital signal processing and control module and a network interface module which are connected in sequence;
the radio frequency signal processing module is used for receiving, filtering and amplifying signals and outputting the signals to the digital signal processing and controlling module, and the digital signal processing and controlling module is used for identifying signal characteristics and identifying signal digital modulation modes of the received signals and outputting the identified signal characteristics and the identified signal digital modulation modes to the outside through the network interface module;
the digital signal processing and controlling module comprises an ADC module and a signal characteristic identifying module; the signal characteristic recognition module comprises a GPU processing module for recognizing the signal type and an FPGA processing module for recognizing the digital modulation mode, wherein the GPU processing module comprises a down-conversion module and a GPU chip, and the FPGA processing module comprises a ZYNQ series chip;
the GPU chip contains a computer program for performing steps 2-3 of claim 1.
5. The machine learning based signal feature recognition device of claim 4 wherein the ZYNQ series chip comprises a computer program for performing steps 2-3 of claim 2.
6. The machine learning based signal feature recognition device of claim 4 wherein the radio frequency signal processing module comprises a filter circuit and a low noise amplification circuit, the filter circuit filters the signal before the signal is amplified by the low noise amplification circuit, and the amplified signal is sent to the digital signal processing and control module.
CN202110259492.6A 2021-03-10 2021-03-10 Signal characteristic recognition device and method based on machine learning Active CN113095137B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110259492.6A CN113095137B (en) 2021-03-10 2021-03-10 Signal characteristic recognition device and method based on machine learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110259492.6A CN113095137B (en) 2021-03-10 2021-03-10 Signal characteristic recognition device and method based on machine learning

Publications (2)

Publication Number Publication Date
CN113095137A CN113095137A (en) 2021-07-09
CN113095137B true CN113095137B (en) 2023-12-29

Family

ID=76666848

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110259492.6A Active CN113095137B (en) 2021-03-10 2021-03-10 Signal characteristic recognition device and method based on machine learning

Country Status (1)

Country Link
CN (1) CN113095137B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116610388B (en) * 2023-07-19 2023-09-19 成都泰格微电子研究所有限责任公司 Method and device for quickly starting ADC and linux system

Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104819846A (en) * 2015-04-10 2015-08-05 北京航空航天大学 Rolling bearing sound signal fault diagnosis method based on short-time Fourier transform and sparse laminated automatic encoder
CN104994045A (en) * 2015-06-26 2015-10-21 北京航空航天大学 Platform and method for automatically identifying digital modulation mode based on USRP platform
CN105656826A (en) * 2016-03-18 2016-06-08 清华大学 Modulation recognizing method and system based on order statistics and machine learning
CN107979554A (en) * 2017-11-17 2018-05-01 西安电子科技大学 Radio signal Modulation Identification method based on multiple dimensioned convolutional neural networks
CN108805112A (en) * 2018-09-18 2018-11-13 深圳大学 A kind of motion recognition system combined based on machine learning and radar
CN109886075A (en) * 2018-12-27 2019-06-14 成都数之联科技有限公司 A kind of signal modulation pattern recognition methods based on planisphere
CN111092836A (en) * 2019-12-13 2020-05-01 中国人民解放军空军工程大学 Signal modulation mode identification method and device
CN111709329A (en) * 2020-05-31 2020-09-25 中国人民解放军63892部队 Unmanned aerial vehicle measurement and control signal high-speed identification method based on deep learning
CN111884962A (en) * 2020-06-01 2020-11-03 山东师范大学 Signal modulation type classification method and system based on convolutional neural network
WO2020228141A1 (en) * 2019-05-13 2020-11-19 清华大学 Electromagnetic signal identification method and device for constructing graph convolutional network on basis of implicit knowledge

Patent Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104819846A (en) * 2015-04-10 2015-08-05 北京航空航天大学 Rolling bearing sound signal fault diagnosis method based on short-time Fourier transform and sparse laminated automatic encoder
CN104994045A (en) * 2015-06-26 2015-10-21 北京航空航天大学 Platform and method for automatically identifying digital modulation mode based on USRP platform
CN105656826A (en) * 2016-03-18 2016-06-08 清华大学 Modulation recognizing method and system based on order statistics and machine learning
CN107979554A (en) * 2017-11-17 2018-05-01 西安电子科技大学 Radio signal Modulation Identification method based on multiple dimensioned convolutional neural networks
CN108805112A (en) * 2018-09-18 2018-11-13 深圳大学 A kind of motion recognition system combined based on machine learning and radar
CN109886075A (en) * 2018-12-27 2019-06-14 成都数之联科技有限公司 A kind of signal modulation pattern recognition methods based on planisphere
WO2020228141A1 (en) * 2019-05-13 2020-11-19 清华大学 Electromagnetic signal identification method and device for constructing graph convolutional network on basis of implicit knowledge
CN111092836A (en) * 2019-12-13 2020-05-01 中国人民解放军空军工程大学 Signal modulation mode identification method and device
CN111709329A (en) * 2020-05-31 2020-09-25 中国人民解放军63892部队 Unmanned aerial vehicle measurement and control signal high-speed identification method based on deep learning
CN111884962A (en) * 2020-06-01 2020-11-03 山东师范大学 Signal modulation type classification method and system based on convolutional neural network

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
基于信噪比分级的信号调制类型识别;陈晋音;蒋焘;郑海斌;;计算机科学(第S1期);320-327 *
基于极端学习机的数字通信调制识别;宋丽辉;张靓;;激光杂志(第03期);123-126 *

Also Published As

Publication number Publication date
CN113095137A (en) 2021-07-09

Similar Documents

Publication Publication Date Title
Zhang et al. An efficient deep learning model for automatic modulation recognition based on parameter estimation and transformation
CN110824450B (en) Radar target HRRP robust identification method in noise environment
CN109410974B (en) Voice enhancement method, device, equipment and storage medium
CN113326930B (en) Data processing method, neural network training method, related device and equipment
CN113095137B (en) Signal characteristic recognition device and method based on machine learning
CN110926782A (en) Circuit breaker fault type judgment method and device, electronic equipment and storage medium
US11700156B1 (en) Intelligent data and knowledge-driven method for modulation recognition
CN113065526B (en) Electroencephalogram signal classification method based on improved depth residual error grouping convolution network
CN112867010B (en) Radio frequency fingerprint embedded real-time identification method and system based on convolutional neural network
CN108478216A (en) A kind of epileptic seizure intelligent Forecasting early period based on convolutional neural networks
Yang et al. One-dimensional deep attention convolution network (ODACN) for signals classification
CN112014801B (en) SPWVD and improved AlexNet based composite interference identification method
CN112418175A (en) Rolling bearing fault diagnosis method and system based on domain migration and storage medium
CN113343801B (en) Automatic wireless signal modulation and identification method based on lightweight convolutional neural network
Liu et al. Radar emitter signal detection with convolutional neural network
Essai et al. Automatic Modulation Classification: Convolutional Deep Learning Neural Networks Approaches
CN116580290B (en) Unmanned aerial vehicle identification method, unmanned aerial vehicle identification device and storage medium
CN116761223B (en) Method for realizing 4G radio frequency communication by using 5G baseband chip and vehicle-mounted radio frequency system
CN113259289A (en) Single-channel aliasing signal modulation mode identification method based on residual error neural network
CN107769801A (en) A kind of method of the lifting radio-frequency transmitter intermediate frequency signal to noise ratio based on neutral net
CN110874828A (en) Neural network model and ultrasonic beam forming method based on neural network model
CN114077819B (en) Fault arc detection model construction method, device, equipment and storage medium
CN115589349A (en) QAM signal modulation identification method based on deep learning channel self-attention mechanism
CN112347921B (en) PDW sequence preprocessing method, system, computer equipment and storage medium
CN112036461B (en) Handwriting digital image recognition method, device, equipment and computer storage medium

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