CN115878982A - Underwater target identification method and system based on dual-frequency echo signal characteristics - Google Patents
Underwater target identification method and system based on dual-frequency echo signal characteristics Download PDFInfo
- Publication number
- CN115878982A CN115878982A CN202211571572.6A CN202211571572A CN115878982A CN 115878982 A CN115878982 A CN 115878982A CN 202211571572 A CN202211571572 A CN 202211571572A CN 115878982 A CN115878982 A CN 115878982A
- Authority
- CN
- China
- Prior art keywords
- frequency
- target
- time
- underwater
- neural network
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/30—Assessment of water resources
Landscapes
- Measurement Of Velocity Or Position Using Acoustic Or Ultrasonic Waves (AREA)
Abstract
The invention discloses an underwater target identification method and system based on double-frequency echo signal characteristics. The method comprises the following steps: extracting the time-frequency characteristics of each sample from the acquired underwater sound target echo signal data samples of different frequency bands; respectively putting the time-frequency characteristic graphs of two frequency bands of the same target echo into two channels for combination to form a time-frequency characteristic graph of a double-frequency echo signal; the time-frequency characteristics of each sample are used as sample data, the underwater acoustic echo data samples of the same target are labeled with the same label, and the underwater acoustic echo data samples of different targets are labeled with different labels to form a training set; training by taking the training set as the input of a convolutional neural network to obtain an identification model based on the convolutional neural network; and carrying out target recognition on the underwater acoustic echo data subjected to the feature extraction by using the trained recognition model. The invention has high identification accuracy, simple network structure, less parameters and convenient application.
Description
Technical Field
The invention relates to the field of underwater acoustic signal processing, in particular to an underwater target identification method and system based on double-frequency echo signal characteristics, and belongs to the field of identification of underwater target echo signals.
Background
With the rapid development of underwater acoustic technology, the observation data of underwater targets are rapidly increased, and how to identify underwater acoustic targets by using the observation data is a research hotspot in the field of underwater safety. Nowadays, underwater sound countermeasure technology is continuously developed, and the countermeasure environment is increasingly complicated. A common countermeasure is to interfere with the opponent's target location by simulating the echo of an underwater target. Therefore, a target identification method is urgently needed to be found to deal with various underwater bait interferences, and further, the underwater target is identified in a complex environment.
The traditional underwater target recognition method usually depends on shallow learning, such as a support vector machine model, a Gaussian mixture model and the like, is suitable for solving the problems of easy feature extraction and specific constraint classification, and has poor data classification effect for difficult feature engineering. In addition, the parameters of shallow learning cannot be adaptive, so the self-correction capability is not good enough. Meanwhile, the existing underwater target identification method is based on target echo of a single frequency band for processing, and due to the lack of characteristic information of a target body under different frequency bands, the identification effect is greatly influenced under the interference of an acoustic bait.
Therefore, an underwater target identification scheme capable of identifying targets in a complex underwater acoustic environment needs to be established for the interference of the acoustic bait in the environment.
Disclosure of Invention
The invention aims to overcome the defects of the prior art, provides a method for realizing underwater target identification under the interference of sound bait, and improves the identification effect.
In order to achieve the purpose, the invention adopts the following technical scheme:
an underwater target identification method based on dual-frequency echo signal characteristics comprises the following steps:
(1) Extracting the time-frequency characteristics of each sample from the acquired underwater sound target echo signal data samples of different frequency bands;
(2) Respectively putting the time-frequency characteristic graphs of two frequency bands of the same target echo into two channels for combination to form a time-frequency characteristic graph of a double-frequency echo signal;
(3) The time-frequency characteristics of each sample are used as sample data, the underwater acoustic echo data samples of the same target are labeled with the same label, and the underwater acoustic echo data samples of different targets are labeled with different labels to form a training set;
(4) Training by taking the training set as the input of the convolutional neural network to obtain a recognition model based on the convolutional neural network;
(5) And performing target recognition on the underwater acoustic echo data subjected to feature extraction by using the trained recognition model.
The invention also provides an underwater target identification system based on the difference of the dual-frequency echo signals, which comprises the following components:
the double-frequency signal time-frequency feature construction module is used for extracting the time-frequency features of the underwater acoustic echo data, combining the time-frequency features of two frequency bands of the same target and regarding the combined features as the double-frequency signal time-frequency features;
the convolutional neural network recognition model training module is used for taking the time-frequency characteristics of each sample as sample data, marking the same label on the underwater acoustic echo data sample of the same target, marking different labels on the underwater acoustic echo data samples of different targets to form a training set, taking the training set as the input of the convolutional neural network, and training to obtain a recognition model based on the convolutional neural network;
and the target identification module is used for inputting the underwater acoustic echo data subjected to the feature extraction as an identification model to realize the classification and identification of the target.
The present invention also provides a computer apparatus comprising: one or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the programs when executed by the processors implement the steps of the dual frequency echo signal difference based underwater object recognition method as described above.
The present invention also provides a computer-readable storage medium having stored thereon a computer program which, when being executed by a processor, carries out the steps of the method for underwater object identification based on difference of dual-frequency echo signals as described above.
The invention provides an underwater target recognition method based on double-frequency echo signal characteristics, which enhances the recognition effect by combining echo signals of two different frequency bands and analyzing the characteristics and is applied to the field of underwater sound recognition for the first time. The convolutional neural network model can break through the dependence on prior knowledge and a feature extraction method in the existing underwater acoustic target recognition, well overcome the defect of the lack of shallow learning self-adaption, can perform parameter self-adaption learning, can represent high-dimensional complex parameters, and can extract deep features of the target. According to experimental demonstration, compared with the traditional single-frequency signal identification method, the accuracy of the double-frequency echo signal identification method provided by the invention is improved by 8.9% in a test set and reaches 84.6%, and the method is simple in network structure, less in parameters, provides a new idea for an underwater sound target identification method in a complex environment, and has a wide application prospect.
Drawings
FIG. 1 is a schematic flow diagram of the process of the present invention.
FIG. 2 is a diagram of a class recognition confusion matrix according to the method of the present invention.
FIG. 3 is a diagram of a convolutional neural network architecture of the present invention.
Detailed Description
The technical scheme of the invention is further explained by combining the attached drawings.
The invention provides a method for realizing underwater target identification based on dual-frequency echo signal characteristics, which mainly comprises two stages with reference to fig. 1: constructing dual-frequency signal time-frequency characteristics and identifying a convolutional neural network. After extracting the time-frequency characteristics of the underwater acoustic echo data, the dual-frequency signal time-frequency characteristic part combines the time-frequency characteristics of the same target under different frequency bands, and the combined characteristics are regarded as the dual-frequency signal time-frequency characteristics. The convolutional neural network identification part takes the constructed dual-frequency signal time-frequency characteristics as input samples, utilizes a linear rectification function as an activation function, and takes a binary cross entropy function as a target function to carry out training, thereby realizing the classification identification of the input samples. In the embodiment, after underwater sound target echo signals are obtained, the underwater sound target echo signals are divided into training samples and testing samples according to a certain proportion, and a training set and a testing set are formed through feature extraction, dual-frequency signal time-frequency feature construction and label labeling; and taking the training set as the input of the convolutional neural network to obtain an identification model based on the convolutional neural network, and taking the test set as the input of the identification model to verify the identification performance of the convolutional neural network.
The following will discuss the implementation and effect of the present invention with a test example on lake. The method specifically comprises the following steps:
step 001, the test conditions of the invention are as follows: a cylinder model of 55cm diameter and 4.2m length was selected as the target and an acoustic source was used instead of the acoustic bait to provide simulated echo. A lake with good hydrological conditions is selected in the test field, the collected data can be guaranteed to be effective, and echo data of a target body are obtained through a sound source emitting detection signals in the test. False echoes of the acoustic bait are transmitted directly from the acoustic source to the hydrophone. And respectively arranging the standard hydrophone, the transmitting transducer and the target body to the same depth, wherein the standard hydrophone is arranged between the transmitting transducer and the target body. The transmitting transducer is 3m away from the standard hydrophone, the standard hydrophone is 3m away from the target body, and the laying depth is 4m. A signal source is used for driving a transmitting transducer to transmit a single-frequency pulse signal, the signal frequency is respectively 20KHz and 50KHz, the pulse width is 0.5ms, the wave number is respectively 10 and 25, and the number of sampling points is 256000. And rotating the target body by 45 degrees and 90 degrees and then repeating the steps respectively to acquire the echo characteristics of the target body in different postures.
In step 002, the experimental data includes two types, namely, target echo and non-target false echo, each type includes two frequency bands of 20KHz and 50KHz, the pulse width of the detection signal is 1ms, and the number of samples in each frequency band is 1110, which is 4400 in total. And part of data with poor acquisition effect is removed, blank sections left during acquisition are manually removed from the rest audio, the audio is subjected to filtering and denoising, and the problem of small sound is solved by enhancing the data of part of the data.
And step 003, performing fast Fourier transform on the data, extracting time-frequency characteristics of each sample, and obtaining 1110 time-frequency graphs with the sizes (198 and 138).
And step 004, putting the time-frequency characteristic graphs of the two different frequency bands into different channels to form a new time-frequency characteristic graph of the dual-channel dual-frequency signal, wherein each channel comprises the time-frequency characteristic of one frequency band, the combined image size is (198, 138,2), and 2 is the number of the channels.
The dual-frequency time-frequency characteristic is constructed based on an underwater acoustic echo bright spot model. A large number of theoretical analyses and experimental researches prove that the echo of any complex target is formed by overlapping a plurality of wavelets, each wavelet can be regarded as a wave emitted from a certain scattering point, and the scattering point is a bright point. The target echo is generated by a target under the excitation of incident sound waves, carries characteristic information of the target and can be equivalent to the superposition of a series of scattered bright spots.
The echo is related to both the natural vibration characteristics of the target and the incident acoustic wave characteristics. Let the single-frequency wave edge of frequency omega
The direction is propagated to the target, and the transfer function of a single bright spot can be summarized as:
wherein the content of the first and second substances,
is an amplitude reflection factor, usually a function of frequency, the value of the center frequency can be taken for the narrowband signal; tau is time delay and is determined by a sound path xi of the equivalent scattering center relative to a certain reference point, tau =2 xi/C, and C is sound velocity; />
Is the phase jump at the time of echo formation,the nature of the bright spot, i.e. the shape of the target, the angle of incidence of the sound wave.
According to the linear superposition principle, the target can be regarded as the superposition of a plurality of bright spots, N represents the total number of target bright spots, and the total transfer function can be expressed as:
the object model is composed of a set of different orientations
Three parameters.
The single bright spot echo signal can be expressed as:
wherein, P i (ω) is the input signal, P, which is the detection signal S (ω) is the echo signal(s),
for far field correction, k is the wavenumber, j is the imaginary unit, and r is the propagation distance.
Because a large volume difference generally exists between the target body and the acoustic bait, according to the parameter setting of the bright spot model, the characteristic differences included in the underwater acoustic echo signals of different frequency bands are mainly reflected in far field correction and amplitude factors. For targets of different structures and materials, the reflected intensity of the targets is different corresponding to acoustic signals of different frequencies. In the invention, the amplitude factor and the amplitude reflection factor are the same meaning and both represent correction coefficients for the amplitude, and the reflection means that the echo generated by reflection is influenced by the coefficient. Wherein the amplitude factor
For some corner and edge spots and elastic spots, the signal frequency is related. Setting the radius of the curved surface of the target body as R, and the incident signal and the targetThe volume angle is θ, and the amplitude factor can be expressed as:
according to the invention, echo signals are analyzed through an underwater acoustic echo bright spot model, echo differences caused by incident signals of different frequency bands are determined, and underwater targets are identified according to characteristic differences of double-frequency echo signals.
The final dual-frequency time-frequency signature is an image matrix of size (198, 138,2).
And 005, marking a sample according to the time-frequency characteristic diagram of the dual-frequency signal obtained in the step 002-004. And taking the double-frequency time-frequency characteristic graph of each behavior sample as sample data, labeling each sample data with a label, labeling the underwater acoustic echo data samples of the same target with the same label, labeling the underwater acoustic echo data samples of different targets with different labels, and forming a training set and a test set. In the embodiment of the invention, according to 8:2 into training and test sets.
Step 006, using the training set as an input of the convolutional neural network, the structure of which is shown in fig. 3. In order to retain as much dual-frequency time-frequency feature information as possible, it is necessary to avoid the continuous use of convolution operations, and therefore, the convolutional neural network structure of this example is composed of two convolutional layers, two pooling layers, and two fully-connected layers, where the pooling method of the pooling layers is maximum pooling. In addition, in the time-frequency image sample, most of the information of the underwater target is hidden in the frequency band distribution and the amplitude of the frequency band distribution, and the boundary characteristics of the image are not particularly critical, so a padding layer is not added in the model. And in order to avoid the overfitting problem caused by the increase of the number of network layers, a Dropout method is introduced, and the Dropout coefficient is 0.6. And training by taking a linear rectification function (ReLU) as an activation function and a binary cross entropy function as a target function according to the characteristic of the dual-frequency echo signal, and storing the trained model. The linear characteristic of the ReLU function when the linear characteristic is larger than 0 can well solve the problem of gradient reduction, so that the characteristic can be more efficient in the calculation process. And its overall non-linearity in turn enables fitting of any complex continuous function in a neural network. The cross entropy loss function has the characteristic that the loss is small when the predicted value is close to the label value, and the loss is large when the predicted value is far away from the label value, so that the model learning is facilitated.
The specific parameters of the model are shown in table 1 below.
TABLE 1 neural network model parameters
And step 007, verifying the recognition performance of the convolutional neural network by taking the test set as the input of the training model.
Referring to FIG. 2, a confusion matrix may be used to present a visualization of the performance of the algorithm, with the total number of columns representing the number of data predicted for that category, each row representing the true attribution category of data, and the total number of data in each row representing the number of data instances for that category. The identification method of the time-frequency characteristics of the double-frequency signals can be seen through the confusion matrix, and the identification accuracy rates of the two types of signals are close. 482 of the 577 target echo signals were accurately identified as target echoes, and 95 were identified as simulated echoes of the target. 494 out of the 577 target volume simulated echoes were identified as target volume simulated echoes, and 83 were identified as target volume echoes.
As can be seen from the verification result of the test set, the method has good identification effect. In practical application, the actually acquired underwater sound target echo data can be input into the model after feature extraction based on the trained recognition model, and target recognition is carried out.
The accuracy of the double-frequency echo signal identification method provided by the invention is improved by 8.9% in a test set and reaches 84.6% compared with the traditional single-frequency signal identification method, and the method has the advantages of simple network structure and few parameters, provides a new idea for an underwater acoustic target identification method in a complex environment, and has wide application prospect.
The invention also provides an underwater target recognition system based on the difference of the double-frequency echo signals, which comprises the following components:
the double-frequency signal time-frequency feature construction module is used for extracting the time-frequency features of the underwater acoustic echo data, combining the time-frequency features of two frequency bands of the same target and regarding the combined features as the double-frequency signal time-frequency features;
the convolutional neural network recognition model training module is used for taking the time-frequency characteristics of each sample as sample data, marking the same label on the underwater acoustic echo data sample of the same target, marking different labels on the underwater acoustic echo data samples of different targets to form a training set, taking the training set as the input of the convolutional neural network, and training to obtain a recognition model based on the convolutional neural network;
and the target identification module is used for inputting the underwater acoustic echo data subjected to the feature extraction as an identification model to realize the classification and identification of the target.
It should be understood that the underwater target recognition system in the embodiment of the present invention may implement all technical solutions in the above method embodiments, and the functions of each functional module may be implemented according to the method in the above method embodiments, for example, when implemented specifically, the dual-frequency signal time-frequency feature building module may be implemented according to the description of steps 001-004, the convolutional neural network recognition model training module may be implemented according to the description of steps 005-006, the target recognition module may be implemented according to the description of step 007, and the specific implementation process may refer to the relevant description in the above method embodiments, which is not described herein again.
The present invention also provides a computer apparatus comprising: one or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the programs when executed by the processors implement the steps of the dual frequency echo signal difference based underwater object recognition method as described above.
The invention also provides a computer-readable storage medium, on which a computer program is stored, which computer program, when being executed by a processor, carries out the steps of the method for underwater object identification based on difference of dual-frequency echo signals as described above.
The invention provides a method for realizing underwater target identification based on dual-frequency echo signal characteristics, and the method and the way for realizing the technical scheme are many, and the method is only a preferred embodiment of the invention. It should be noted that, for those skilled in the art, without departing from the principle of the present invention, several improvements and modifications can be made, and these improvements and modifications should also be construed as the protection scope of the present invention. All the components not specified in the present embodiment can be realized by the prior art.
Claims (8)
1. An underwater target identification method based on dual-frequency echo signal difference is characterized by comprising the following steps:
(1) Extracting the time-frequency characteristics of each sample from the acquired underwater sound target echo signal data samples of different frequency bands;
(2) Respectively putting the time-frequency characteristic graphs of two frequency bands of the same target echo into two channels for combination to form a time-frequency characteristic graph of a double-frequency echo signal;
(3) The time-frequency characteristics of each sample are used as sample data, the underwater acoustic echo data samples of the same target are labeled with the same label, and the underwater acoustic echo data samples of different targets are labeled with different labels to form a training set;
(4) Training by taking the training set as the input of a convolutional neural network to obtain an identification model based on the convolutional neural network;
(5) And carrying out target recognition on the underwater acoustic echo data subjected to the feature extraction by using the trained recognition model.
2. The method of claim 1, wherein the time-frequency features of the sample are extracted using a fast fourier transform.
3. The method of claim 1, wherein the differences between the time-frequency characteristics of the dual-frequency echo signals comprise a far-field modification and an amplitude factor, the far-field modification being
Wherein k is the wave number, j is the imaginary unit, r is the incident wave propagation distance; the amplitude factor is expressed as: />
Wherein R is the radius of the curved surface of the target body, and theta is the included angle between the incident signal and the target body.
4. The method of claim 1, wherein the convolutional neural network consists of two convolutional layers, two pooling layers plus two fully-connected layers, wherein the pooling method of the pooling layers is max-pooling and a Dropout layer is placed after the max-pooling layer.
5. The method of claim 1, wherein training the convolutional neural network uses a linear rectification function ReLU as an activation function and a binary cross entropy function as a target loss function.
6. An underwater target identification system based on dual-frequency echo signal difference, comprising:
the double-frequency signal time-frequency feature construction module is used for extracting the time-frequency features of the underwater acoustic echo data, combining the time-frequency features of two frequency bands of the same target and regarding the combined features as the double-frequency signal time-frequency features;
the convolutional neural network recognition model training module is used for taking the time-frequency characteristics of each sample as sample data, labeling the same label on the underwater acoustic echo data sample of the same target, labeling different labels on the underwater acoustic echo data samples of different targets to form a training set, taking the training set as the input of the convolutional neural network, and training to obtain a recognition model based on the convolutional neural network;
and the target identification module is used for inputting the underwater acoustic echo data subjected to the feature extraction as an identification model to realize the classification and identification of the target.
7. A computer device, comprising:
one or more processors;
a memory; and
one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, which when executed by the processors, implement the steps of the dual frequency echo signal difference based underwater object recognition method of any one of claims 1-5.
8. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the method for underwater object identification based on difference of dual-frequency echo signals according to any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211571572.6A CN115878982B (en) | 2022-12-08 | 2022-12-08 | Underwater target identification method and system based on double-frequency echo signal characteristics |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211571572.6A CN115878982B (en) | 2022-12-08 | 2022-12-08 | Underwater target identification method and system based on double-frequency echo signal characteristics |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115878982A true CN115878982A (en) | 2023-03-31 |
| CN115878982B CN115878982B (en) | 2023-09-22 |
Family
ID=85766542
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211571572.6A Active CN115878982B (en) | 2022-12-08 | 2022-12-08 | Underwater target identification method and system based on double-frequency echo signal characteristics |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115878982B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB201201027D0 (en) * | 2011-01-25 | 2012-03-07 | Subsea Asset Location Tech Ltd | Identification, detection and positioning of underwater accoustic reflectors |
| CN104777474A (en) * | 2015-04-29 | 2015-07-15 | 中国科学院声学研究所 | Echo phase feature extracting method used for underwater target recognition |
| CN107748862A (en) * | 2017-09-21 | 2018-03-02 | 清华大学 | A kind of unmanned plane sorting technique and device based on dual-frequency radar signal time-frequency distributions |
| CN109100710A (en) * | 2018-06-26 | 2018-12-28 | 东南大学 | A kind of Underwater targets recognition based on convolutional neural networks |
| CN112257521A (en) * | 2020-09-30 | 2021-01-22 | 中国人民解放军军事科学院国防科技创新研究院 | CNN underwater acoustic signal target identification method based on data enhancement and time-frequency separation |
| CN112990082A (en) * | 2021-04-06 | 2021-06-18 | 舟山精卫智能科技有限公司 | Detection and identification method of underwater sound pulse signal |
| CN113435276A (en) * | 2021-06-16 | 2021-09-24 | 中国电子科技集团公司第五十四研究所 | Underwater sound target identification method based on antagonistic residual error network |
| CN113452637A (en) * | 2021-09-01 | 2021-09-28 | 中国海洋大学 | Underwater acoustic communication signal modulation identification method based on feature selection and support vector machine |
-
2022
- 2022-12-08 CN CN202211571572.6A patent/CN115878982B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB201201027D0 (en) * | 2011-01-25 | 2012-03-07 | Subsea Asset Location Tech Ltd | Identification, detection and positioning of underwater accoustic reflectors |
| CN104777474A (en) * | 2015-04-29 | 2015-07-15 | 中国科学院声学研究所 | Echo phase feature extracting method used for underwater target recognition |
| CN107748862A (en) * | 2017-09-21 | 2018-03-02 | 清华大学 | A kind of unmanned plane sorting technique and device based on dual-frequency radar signal time-frequency distributions |
| CN109100710A (en) * | 2018-06-26 | 2018-12-28 | 东南大学 | A kind of Underwater targets recognition based on convolutional neural networks |
| CN112257521A (en) * | 2020-09-30 | 2021-01-22 | 中国人民解放军军事科学院国防科技创新研究院 | CNN underwater acoustic signal target identification method based on data enhancement and time-frequency separation |
| CN112990082A (en) * | 2021-04-06 | 2021-06-18 | 舟山精卫智能科技有限公司 | Detection and identification method of underwater sound pulse signal |
| CN113435276A (en) * | 2021-06-16 | 2021-09-24 | 中国电子科技集团公司第五十四研究所 | Underwater sound target identification method based on antagonistic residual error network |
| CN113452637A (en) * | 2021-09-01 | 2021-09-28 | 中国海洋大学 | Underwater acoustic communication signal modulation identification method based on feature selection and support vector machine |
Non-Patent Citations (2)
| Title |
|---|
| ZHANG SHAO-KANG等: "Intelligent Recognition of Underwater Acoustic Target Noise on Underwater Glider Platform", 《2018 CHINESE AUTOMATION CONGRESS (CAC)》, pages 4189 - 4193 * |
| 夏峙: "水下目标主动声呐回波信号特征研究", 《中国博士学位论文全文数据库工程科技Ⅱ辑》, no. 11, pages 028 - 9 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115878982B (en) | 2023-09-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Liu et al. | Underwater target recognition using convolutional recurrent neural networks with 3-D Mel-spectrogram and data augmentation | |
| Thode et al. | Automated detection and localization of bowhead whale sounds in the presence of seismic airgun surveys | |
| CN106772331B (en) | Target identification method and Target Identification Unit | |
| Zeng et al. | Maximum margin classification based on flexible convex hulls for fault diagnosis of roller bearings | |
| Peso Parada et al. | Using Gaussian mixture models to detect and classify dolphin whistles and pulses | |
| CN115114949A (en) | Intelligent ship target identification method and system based on underwater acoustic signals | |
| CN110850420B (en) | Fisher SVM sonar signal discrimination method based on marble loss | |
| Jin et al. | Offshore ship recognition based on center frequency projection of improved EMD and KNN algorithm | |
| White et al. | More than a whistle: Automated detection of marine sound sources with a convolutional neural network | |
| CN107727749A (en) | A kind of ultrasonic quantitative detection method based on wavelet packet fusion feature extraction algorithm | |
| Adam | The use of the Hilbert-Huang transform to analyze transient signals emitted by sperm whales | |
| Bellafkir et al. | Bat echolocation call detection and species recognition by transformers with self-attention | |
| CN101644768B (en) | Torpedo target recognition method based on cepstrum analysis | |
| CN115878982A (en) | Underwater target identification method and system based on dual-frequency echo signal characteristics | |
| Zhou et al. | An attention-based multi-scale convolution network for intelligent underwater acoustic signal recognition | |
| CN113109794A (en) | Deep sea sound source depth setting method based on deep neural network in strong noise environment | |
| Küsel et al. | Single-sensor, cue-counting population density estimation: Average probability of detection of broadband clicks | |
| CN108269566A (en) | A kind of thorax mouth wave recognition methods based on multiple dimensioned sub-belt energy collection feature | |
| CN115047448A (en) | Indoor target rapid detection method and system based on acoustic-electromagnetic intermodulation | |
| CN113109795B (en) | Deep sea direct sound zone target depth estimation method based on deep neural network | |
| CN105156901B (en) | A kind of optical fiber early warning system noise-reduction method and device based on wavelet analysis | |
| Neilsen | Localization of multiple acoustic sources in the shallow ocean | |
| Ashok et al. | Absorption of echo signal for underwater acoustic signal target system using hybrid of ensemble empirical mode with machine learning techniques | |
| Nanda et al. | SETI (Search for Extra Terrestrial Intelligence) signal classification using machine learning | |
| Wang et al. | Auditory-based multi-scale amplitude-aware permutation entropy as a measure for feature extraction of ship radiated noise |
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 |