CN113516027A - Data enhancement method and system based on radio signal fusion - Google Patents
Data enhancement method and system based on radio signal fusion Download PDFInfo
- Publication number
- CN113516027A CN113516027A CN202110467591.3A CN202110467591A CN113516027A CN 113516027 A CN113516027 A CN 113516027A CN 202110467591 A CN202110467591 A CN 202110467591A CN 113516027 A CN113516027 A CN 113516027A
- Authority
- CN
- China
- Prior art keywords
- model
- data
- data set
- samples
- module
- 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 OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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 OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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 OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Signal Processing (AREA)
- Monitoring And Testing Of Transmission In General (AREA)
Abstract
本发明公开了一种基于无线电信号融合的数据增强方法，包括：(S1)：数据集预处理及深度学习模型预训练；(S2)：基于无线电信号融合的数据增强方法；(S3)：利用数据增强方法生成并筛选出扩增数据集；(S4)：测试增强前后的模型分类精度。本发明还包括种基于无线电信号融合的数据增强系统，由模型预训练模块、信号融合模块、数据筛选模块和模型测试模块组成；本发明可以根据无线电信号的片段互换生成大量的增强样本，然后与原数据集结合来训练深度学习模型，从而提升模型的泛化能力。
The invention discloses a data enhancement method based on radio signal fusion, comprising: (S1): data set preprocessing and deep learning model pretraining; (S2): data enhancement method based on radio signal fusion; (S3): using The data augmentation method generates and filters out the augmented data set; (S4): Test the model classification accuracy before and after augmentation. The invention also includes a data enhancement system based on radio signal fusion, which is composed of a model pre-training module, a signal fusion module, a data screening module and a model testing module; the invention can generate a large number of enhanced samples according to the fragment exchange of radio signals, and then Combine with the original data set to train the deep learning model, thereby improving the generalization ability of the model.
Description
Technical Field
The invention relates to a data enhancement method, which is mainly applied to the field of data enhancement of deep learning models, in particular to a data enhancement method and system based on radio signal fusion.
Background
With the development of neural networks, deep learning models gradually step into ultra-large scale neural networks from simple fully-connected layers, but these networks usually require a large amount of training data to avoid overfitting. However, many application scenarios cannot obtain sufficient data to support a huge network model, and therefore, a data enhancement technology is developed, which expands a data set by generating more equivalent data through limited data to improve the size and quality of a training set, thereby constructing a more accurate and robust deep learning model.
In wireless communication, modulation information is shared between a transmitter and a receiver in a standard communication scenario, but in a specific scenario, such as the military field, a receiving end can only perform blind demodulation on a modulation signal through an automatic modulation classification scheme under the condition that the receiving end does not know what modulation mode is used by a transmitting end. Therefore, in recent years, radio signal identification based on a deep learning model is also gradually studied and exhibits good classification performance. A well-performing model often requires a large number of training samples, however, collecting a large number of high quality and reliable radio samples is sometimes difficult. How to rapidly enhance data from the data of the existing tags is a main motivation for solving the problems of the invention.
At present, a technical scheme disclosed in a patent with application number 201910694936.1 is available as a data enhancement method in the signal field, and the data enhancement method is a small sample radio signal enhancement identification method based on ACGAN. The method comprises the steps of classifying radio signals through a Long Short Term Memory (LSTM) network, further generating a countermeasure network (ACGAN) by using an Auxiliary generation method to expand a small sample data set, further enhancing an identification model, and completing small sample radio signal enhanced identification. The technology is applied to radio signal data enhancement technology, however, the training of ACGAN is unstable, and a great deal of training skill is needed. The technology of the patent is efficient and feasible, and provides a data enhancement method based on radio signal fusion.
Disclosure of Invention
In order to solve the problem of limited data volume in the prior art and further improve the accuracy and robustness of the model in a classification model, the invention provides a data enhancement method based on radio signal fusion, and the effectiveness and universality of the enhanced identification method provided by the method are verified through experiments, and data sets with different sizes can be effectively expanded, so that the model accuracy is improved, and the generalization capability of the model is further improved.
The technical conception of the invention is as follows: the method is suitable for data with time sequence, effectively amplifies the data set by the mentioned data fusion method, reduces the cost of collecting training samples, and further provides more sufficient training samples for the neural network model. The experimental results prove the feasibility and effectiveness of the method provided by the invention.
The technical scheme adopted by the invention for realizing the aim is as follows:
a data enhancement method based on radio signal fusion, characterized by: the method comprises the following steps:
s1: preprocessing a data set and pre-training a deep learning model;
s2: a data enhancement method based on radio signal fusion;
s3: generating and screening an amplification data set by using a data enhancement method;
s4: testing the classification precision of the model before and after enhancement;
preferably, the step S1 specifically includes:
s1.1: using radio signals as the data set V, the data set is first subjected to a normalization pre-processing, and then the signals of different classes are divided into respective sub-data sets { V }1，v2，…，vNN is the number of data set categories;
s1.2: and pre-training a classification model by using a data set V, wherein the classification model adopts a resnet model, the resnet model comprises a residual module consisting of a 1 x 1 convolution, two residual blocks and a maximum pooling layer, and two full connection layers, and selu and softmax are adopted as activation functions of the full connection layers, so that the pre-classification of the signal data set is realized.
Preferably, the step S2 specifically includes:
for a subdata set vkAny two signals s, s' of (k-1, 2, …, N), where s-is(s)1，s2，s3，…，sL) Is a signal sequence of length L. Selecting an arbitrary sample point s in a signal sp(p ∈ {1, 2, …, L }) as a breakpoint, and the signal sequence(s) following the breakpoint is addedp+1，…，sL-1，sL) And(s)p+1’，…，sL-1’，sL') exchange positions with each other and keep the sequence before the break point constant, the resulting signals are(s) respectively1，s2，…，sp，sp+1’，…，sL') and(s)1’，s2’，…，sp’，sp+1，…，sL)。
Preferably, the step S3 specifically includes:
s3.1: for each subdata set vkTaking the similarity S as a measurement index, selecting two similar samples S, S' to perform signal fusion operation so as to avoid the samples with larger differences from generating interference terms to influence the stability of the model, wherein the similarity S is defined as follows:
wherein s isiIs the ith sample point, and L is the length of the signal sequence;
s3.2: and (4) amplifying the data set, namely selecting a proper threshold value Y to judge whether two samples are subjected to signal fusion, randomly selecting two radio signal samples in each subdata set, performing the operation and storing the samples if the similarity S calculated by the two samples is more than or equal to Y, and abandoning the exchange if the S is less than Y. Iterating the process on each sub data set for multiple times until m new samples are generated; assuming that there are M samples per class of the original dataset, the synthesized new samples are equivalent to enhancing the dataset by M/M times, and then all M × N new samples are merged into V, and the resulting enhanced dataset DV is (N × M, L).
Preferably, the step S4 specifically includes:
and putting the amplified data set DV into a neural network for retraining, using the same test data set test model, comparing the precision before and after enhancement, and verifying the effectiveness of the method.
The system for realizing the data enhancement method based on the radio signal fusion comprises the following steps: the system comprises a model pre-training module, a signal fusion module, a data screening module and a model testing module;
the model pre-training module is used for putting the data set into a deep learning model for training on the premise of normalization pre-processing of the data set;
the signal fusion module is a signal-oriented data enhancement method, and specifically comprises the following steps: selecting signal samples of the same category, setting a breakpoint, exchanging signal sequences after the breakpoint, and keeping the sequences before the breakpoint unchanged, so that two new enhanced samples can be combined from every two signal samples;
the data screening module selects a sample to be fused by utilizing a similarity function, and specifically comprises the following steps: judging whether the two samples are suitable for data enhancement according to a threshold value, fusing the two samples to amplify an original data set after the two samples meet the condition, and then amplifying a proper amount of enhanced data set through iteration to train a deep learning model;
the model test module is used for testing the generalization ability of the deep learning model before and after enhancement, and specifically comprises the following steps: after an enhanced data set is generated, putting the enhanced data set into a deep learning model for training, then testing the precision of the model, comparing the precision with the testing precision of an original model, and verifying the effectiveness of the method;
the model pre-training module, the signal fusion module, the data screening module and the model testing module are connected in sequence.
The invention has the beneficial effects that:
(1) the data enhancement method based on radio signal fusion is provided, the operation is convenient and easy, and the scale of a signal data set is greatly expanded;
(2) the invention can provide effective help for further improving the model precision after a large number of samples are generated. Meanwhile, the method has certain enhancement effect on different radio signal data sets, and has remarkable effect on small radio signal data sets.
Drawings
FIG. 1 is a flow chart of the method of the present invention.
FIG. 2 is a schematic diagram of data fusion in accordance with the present invention.
Fig. 3 is a diagram of a neural network model structure applied to the present invention.
Fig. 4 is a schematic diagram of the system architecture of the present invention.
Detailed Description
The following detailed description of embodiments of the invention is provided in connection with the accompanying drawings.
Referring to fig. 1 to 4, a data enhancement method and system based on radio signal fusion, the invention uses modulation signal data set rml2016.10a published by the university of bradley, which contains 11 modulation types, and the signal-to-noise ratio range of each modulation type signal is an even number between [ -20,18 ]. Each radio signal sample is 128x2 in size. The number of training set samples was 176000 and the number of test set samples was 44000. The method uses a portion of the data (880 training samples) in the 18db signal in the data set, while training using a modified ResNet classification model, to test the effectiveness of the enhancement method. The method comprises the following steps:
s1: the data set preprocessing and deep learning model pre-training method specifically comprises the following steps:
s1.1: using radio signals as the data set V, the data set is first subjected to a normalization pre-processing, and then the signals of different classes are divided into respective sub-data sets { V }1，v2，…，vNN is the number of data set categories.
S1.2: and pre-training a classification model by using a data set V, wherein the classification model adopts a resnet model, the resnet model comprises a residual module consisting of a 1 x 1 convolution, two residual blocks and a maximum pooling layer, and two full connection layers, and selu and softmax are adopted as activation functions of the full connection layers, so that the pre-classification of the signal data set is realized.
S2: fig. 2 shows a data enhancement method based on radio signal fusion, which specifically includes:
for a subdata set vkAny two signals s, s' of (k-1, 2, …, N), where s-is(s)1，s2，s3，…，sL) Is a signal sequence of length L. Selecting an arbitrary sample point s in a signal sp(p ∈ {1, 2, …, L }) as a breakpoint, and the signal sequence(s) following the breakpoint is addedp+1，…，sL-1，sL) And(s)p+1’，…，sL-1’，sL') exchange positions with each other and keep the sequence before the break point constant, the resulting signals are(s) respectively1，s2，…，sp，sp+1’，…，sL') and(s)1’，s2’，…，sp’，sp+1，…，sL)。
S3: generating and screening an amplification data set by using a data enhancement method, which specifically comprises the following steps:
s3.1: for each subdata set vkTaking the similarity S as a measurement index, selecting two similar samples S, S' to perform signal fusion operation so as to avoid the samples with larger differences from generating interference terms to influence the stability of the model, wherein the similarity S is defined as follows:
wherein s isiFor the ith sample point, L is the length of the signal sequence.
S3.2: and (4) amplifying the data set, namely selecting a proper threshold value Y to judge whether two samples are subjected to signal fusion, randomly selecting two radio signal samples in each subdata set, performing the operation and storing the samples if the similarity S calculated by the two samples is more than or equal to Y, and abandoning the exchange if the S is less than Y. The process is iterated multiple times over each sub data set until m new samples are generated. Assuming that there are M samples per class of the original dataset, the synthesized new samples are equivalent to enhancing the dataset by M/M times, and then all M × N new samples are merged into V, and the resulting enhanced dataset DV is (N × M, L).
S4: the method for testing the classification precision of the model before and after the enhancement specifically comprises the following steps:
and selecting Y-0.3 as a threshold value to judge whether the two samples are subjected to data fusion, and then finding out that the method has the best effect when points in the selected sequence are selected through experiments of different exchange nodes. Therefore, for a small data set, usually, the whole data set is traversed pairwise, and a midpoint is selected as a switching node, so that pairwise fusion of all data is realized, and the maximum enhancement magnification is achieved. The amplified data set DV is put into a neural network for retraining, the same test data set test model is used, the used model is shown in figure 3, the model adopts a residual block mode, the depth of the model is improved, and meanwhile the problems of gradient dispersion and performance degradation are effectively solved. And finally, comparing the precision before and after enhancement, and verifying the effectiveness of the method.
Table 1 radio signal data enhancement results
As shown in table one, the test accuracy of the method in the small sample data set of the radio signal is shown, wherein 550 training samples, 2200 test samples and the resnet model are selected, and the average value of a plurality of sets of experiments is taken as the final result. It can be seen that the method greatly improves the accuracy and generalization capability of the deep learning model.
The system for implementing a data enhancement method based on radio signal fusion of the invention, as shown in fig. 4, comprises: the device comprises a model pre-training module, a signal fusion module, a data screening module and a model testing module.
The model pre-training module is used for putting the data set into a deep learning model for training on the premise of normalization pre-processing of the data set;
the signal fusion module is a signal-oriented data enhancement method, and specifically comprises the following steps: selecting signal samples of the same category, setting a breakpoint, exchanging signal sequences after the breakpoint, and keeping the sequences before the breakpoint unchanged, so that two new enhanced samples can be combined from every two signal samples;
the data screening module selects a sample to be fused by utilizing a similarity function, and specifically comprises the following steps: judging whether the two samples are suitable for data enhancement according to a threshold value, fusing the two samples to amplify an original data set after the two samples meet the condition, and then amplifying a proper amount of enhanced data set through iteration to train a deep learning model;
the model test module is used for testing the generalization ability of the deep learning model before and after enhancement, and specifically comprises the following steps: after an enhanced data set is generated, putting the enhanced data set into a deep learning model for training, then testing the precision of the model, comparing the precision with the testing precision of an original model, and verifying the effectiveness of the method;
the model pre-training module, the signal fusion module, the data screening module and the model testing module are connected in sequence.
The method can generate a large number of enhanced samples according to the fragment interchange of the radio signals, and then train the deep learning model by combining with the original data set, so that the generalization capability of the model is improved.
The embodiments described in this specification are merely illustrative of implementations of the inventive concept and the scope of the present invention should not be considered limited to the specific forms set forth in the embodiments but rather by the equivalents thereof as may occur to those skilled in the art upon consideration of the present inventive concept.
Claims (6)
1. A data enhancement method based on radio signal fusion, characterized by: the method comprises the following steps:
s1: preprocessing a data set and pre-training a deep learning model;
s2: a data enhancement method based on radio signal fusion;
s3: generating and screening an amplification data set by using a data enhancement method;
s4: and testing the classification precision of the model before and after enhancement.
2. A data enhancement method based on radio signal fusion according to claim 1, characterized in that: the step S1 specifically includes:
s1.1: using radio signals as the data set V, the data set is first subjected to a normalization pre-processing, and then the signals of different classes are divided into respective sub-data sets { V }1，v2，…，vNN is the number of data set categories;
s1.2: and pre-training a classification model by using a data set V, wherein the classification model adopts a resnet model, the resnet model comprises a residual module consisting of a 1 x 1 convolution, two residual blocks and a maximum pooling layer, and two full connection layers, and selu and softmax are adopted as activation functions of the full connection layers, so that the pre-classification of the signal data set is realized.
3. A data enhancement method based on radio signal fusion according to claim 1, characterized in that: the step S2 specifically includes:
for a subdata set vkAny two signals s, s' of (k-1, 2, …, N), where s-is(s)1，s2，s3，…，sL) Is a signal sequence of length L. Selecting an arbitrary sample point s in a signal sp(p ∈ {1, 2, …, L }) as a breakpoint, and the signal sequence(s) following the breakpoint is addedp+1，…，sL-1，sL) And(s)p+1’，…，sL-1’，sL') exchange positions with each other and keep the sequence before the break point constant, the resulting signals are(s) respectively1，s2，…，sp，sp+1’，…，sL') and(s)1’，s2’，…，sp’，sp+1，…，sL)。
4. A data enhancement method based on radio signal fusion according to claim 1, characterized in that: the step S3 specifically includes:
s3.1: for each subdata set vkTaking the similarity S as a measurement index, selecting two similar samples S, S' to perform signal fusion operation so as to avoid the samples with larger differences from generating interference terms to influence the stability of the model, wherein the similarity S is defined as follows:
wherein s isiIs the ith sample point, and L is the length of the signal sequence;
s3.2: and (4) amplifying the data set, namely selecting a proper threshold value Y to judge whether two samples are subjected to signal fusion, randomly selecting two radio signal samples in each subdata set, performing the operation and storing the samples if the similarity S calculated by the two samples is more than or equal to Y, and abandoning the exchange if the S is less than Y. Iterating the process on each sub data set for multiple times until m new samples are generated; assuming that there are M samples per class of the original dataset, the synthesized new samples are equivalent to enhancing the dataset by M/M times, and then all M × N new samples are merged into V, and the resulting enhanced dataset DV is (N × M, L).
5. A data enhancement method based on radio signal fusion according to claim 1, characterized in that: the step S4 specifically includes:
and putting the amplified data set DV into a neural network for retraining, using the same test data set test model, comparing the precision before and after enhancement, and verifying the effectiveness of the method.
6. A system for implementing a radio signal fusion based data enhancement method as claimed in claim 1, comprising: the system comprises a model pre-training module, a signal fusion module, a data screening module and a model testing module;
the model pre-training module is used for putting the data set into a deep learning model for training on the premise of normalization pre-processing of the data set;
the signal fusion module is a signal-oriented data enhancement method, and specifically comprises the following steps: selecting signal samples of the same category, setting a breakpoint, exchanging signal sequences after the breakpoint, and keeping the sequences before the breakpoint unchanged, so that two new enhanced samples can be combined from every two signal samples;
the data screening module selects a sample to be fused by utilizing a similarity function, and specifically comprises the following steps: judging whether the two samples are suitable for data enhancement according to a threshold value, fusing the two samples to amplify an original data set after the two samples meet the condition, and then amplifying a proper amount of enhanced data set through iteration to train a deep learning model;
the model test module is used for testing the generalization ability of the deep learning model before and after enhancement, and specifically comprises the following steps: after an enhanced data set is generated, putting the enhanced data set into a deep learning model for training, then testing the precision of the model, comparing the precision with the testing precision of an original model, and verifying the effectiveness of the method;
the model pre-training module, the signal fusion module, the data screening module and the model testing module are connected in sequence.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110467591.3A CN113516027B (en) | 2021-04-28 | 2021-04-28 | Data enhancement method and system based on radio signal fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110467591.3A CN113516027B (en) | 2021-04-28 | 2021-04-28 | Data enhancement method and system based on radio signal fusion |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113516027A true CN113516027A (en) | 2021-10-19 |
| CN113516027B CN113516027B (en) | 2024-03-29 |
Family
ID=78063908
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110467591.3A Active CN113516027B (en) | 2021-04-28 | 2021-04-28 | Data enhancement method and system based on radio signal fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113516027B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114444653A (en) * | 2021-12-21 | 2022-05-06 | 武汉中海庭数据技术有限公司 | A method and system for evaluating the impact of data augmentation on the performance of deep learning models |
| CN115238763A (en) * | 2022-06-10 | 2022-10-25 | 西北工业大学 | Multi-period hydrodynamic experimental data enhancement method of simulated bat aircraft based on deep learning |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7061891B1 (en) * | 2001-02-02 | 2006-06-13 | Science Applications International Corporation | Method and system for a remote downlink transmitter for increasing the capacity and downlink capability of a multiple access interference limited spread-spectrum wireless network |
| CN109582782A (en) * | 2018-10-26 | 2019-04-05 | 杭州电子科技大学 | A kind of Text Clustering Method based on Weakly supervised deep learning |
| CN112418347A (en) * | 2020-12-09 | 2021-02-26 | 浙江工业大学 | Countermeasure enhancement method based on radio signal classification |
-
2021
- 2021-04-28 CN CN202110467591.3A patent/CN113516027B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7061891B1 (en) * | 2001-02-02 | 2006-06-13 | Science Applications International Corporation | Method and system for a remote downlink transmitter for increasing the capacity and downlink capability of a multiple access interference limited spread-spectrum wireless network |
| CN109582782A (en) * | 2018-10-26 | 2019-04-05 | 杭州电子科技大学 | A kind of Text Clustering Method based on Weakly supervised deep learning |
| CN112418347A (en) * | 2020-12-09 | 2021-02-26 | 浙江工业大学 | Countermeasure enhancement method based on radio signal classification |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114444653A (en) * | 2021-12-21 | 2022-05-06 | 武汉中海庭数据技术有限公司 | A method and system for evaluating the impact of data augmentation on the performance of deep learning models |
| CN115238763A (en) * | 2022-06-10 | 2022-10-25 | 西北工业大学 | Multi-period hydrodynamic experimental data enhancement method of simulated bat aircraft based on deep learning |
| CN115238763B (en) * | 2022-06-10 | 2024-02-13 | 西北工业大学 | Multi-period hydrodynamic experimental data enhancement method of simulated batline aircraft based on deep learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113516027B (en) | 2024-03-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107330453B (en) | A pornographic image recognition method based on step-by-step recognition and fusion of key parts detection | |
| CN112839034B (en) | A Network Intrusion Detection Method Based on CNN-GRU Hierarchical Neural Network | |
| CN107743103B (en) | Multi-node access detection and channel estimation method of MMTC (multimedia messaging and control) system based on deep learning | |
| CN111737493B (en) | Abnormal data source output method and device based on decision tree and computer equipment | |
| CN112702294B (en) | A modulation recognition method based on deep learning for multi-level feature extraction | |
| CN116912595B (en) | A cross-domain multimodal remote sensing image classification method based on contrastive learning | |
| CN114726692B (en) | SERESESESENet-LSTM-based radiation source modulation mode identification method | |
| CN113516027A (en) | Data enhancement method and system based on radio signal fusion | |
| CN111062424A (en) | Small sample food image recognition model training method and food image recognition method | |
| CN116070136A (en) | Multi-modal fusion wireless signal automatic modulation recognition method based on deep learning | |
| CN114925720B (en) | Small sample modulation signal identification method based on space-time hybrid feature extraction network | |
| CN118314424B (en) | Vehicle-road collaborative self-advancing learning multi-mode verification method based on edge scene | |
| CN111050315A (en) | Wireless transmitter identification method based on multi-core two-way network | |
| CN116614333A (en) | Modulation identification method based on Markov conversion field and deep learning | |
| CN112379859B (en) | Binary sampling processing method and device, adversarial sample generation method and device | |
| CN113343924A (en) | Modulation signal identification method based on multi-scale cyclic spectrum feature and self-attention generation countermeasure network | |
| CN111950626A (en) | Robustness evaluation method of deep neural network model for image classification based on EM | |
| Feng et al. | FCGCN: Feature correlation graph convolution network for few-shot individual identification | |
| CN116150401B (en) | Strong robustness knowledge graph triplet quality inspection network model training method and quality inspection method based on noisy data set | |
| CN119743290B (en) | Vehicle networking abnormal flow detection method based on CNN+TCN+attribute mechanism | |
| CN108805211A (en) | IN service type cognitive method based on machine learning | |
| CN115270891A (en) | A method, device, device and storage medium for generating a signal countermeasure sample | |
| CN112348060B (en) | Classification vector generation method, device, computer equipment and storage medium | |
| CN113221982A (en) | Vehicle identification method, vehicle identification model creation method and related components | |
| Qiu et al. | VAEMax: Open-set intrusion detection based on OpenMax and variational autoencoder |
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 |