CN112926504A - Acoustic emission signal denoising method based on noise reduction self-encoder - Google Patents
Acoustic emission signal denoising method based on noise reduction self-encoder Download PDFInfo
- Publication number
- CN112926504A CN112926504A CN202110308357.6A CN202110308357A CN112926504A CN 112926504 A CN112926504 A CN 112926504A CN 202110308357 A CN202110308357 A CN 202110308357A CN 112926504 A CN112926504 A CN 112926504A
- Authority
- CN
- China
- Prior art keywords
- acoustic emission
- denoising
- emission signal
- signal
- dae
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 33
- 230000009467 reduction Effects 0.000 title claims abstract description 9
- 210000002569 neuron Anatomy 0.000 claims abstract description 18
- 238000002474 experimental method Methods 0.000 claims abstract description 5
- 238000009826 distribution Methods 0.000 claims description 21
- 238000012549 training Methods 0.000 claims description 13
- 238000012360 testing method Methods 0.000 claims description 10
- 238000005070 sampling Methods 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 5
- 230000008030 elimination Effects 0.000 claims description 3
- 238000003379 elimination reaction Methods 0.000 claims description 3
- 210000005036 nerve Anatomy 0.000 claims description 3
- 230000001537 neural effect Effects 0.000 claims description 3
- 238000002759 z-score normalization Methods 0.000 claims description 2
- 230000007797 corrosion Effects 0.000 abstract description 7
- 238000005260 corrosion Methods 0.000 abstract description 7
- 238000012545 processing Methods 0.000 abstract description 7
- 230000000694 effects Effects 0.000 abstract description 4
- 230000006870 function Effects 0.000 description 16
- 238000010586 diagram Methods 0.000 description 8
- 230000008569 process Effects 0.000 description 8
- 238000001514 detection method Methods 0.000 description 6
- 230000004913 activation Effects 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 3
- 238000003062 neural network model Methods 0.000 description 3
- 239000000243 solution Substances 0.000 description 3
- 238000012546 transfer Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 230000007423 decrease Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000035945 sensitivity Effects 0.000 description 2
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000012267 brine Substances 0.000 description 1
- 238000010219 correlation analysis Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000003628 erosive effect Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 210000002364 input neuron Anatomy 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- HPALAKNZSZLMCH-UHFFFAOYSA-M sodium;chloride;hydrate Chemical compound O.[Na+].[Cl-] HPALAKNZSZLMCH-UHFFFAOYSA-M 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000011426 transformation method Methods 0.000 description 1
- 230000001052 transient effect Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Signal Processing (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The invention discloses an acoustic emission signal Denoising method based on a Denoising Autoencoder (DAE), which trains The Denoising Autoencoder to learn more stable invariance characteristics through unsupervised learning so that The error between a reconstructed signal and an original signal is converged to a minimum value, thereby achieving The purpose of Denoising. And by carrying out a denoising experiment on the basis of processing 3011 corrosion acoustic emission signal samples, an experimental result shows that the denoising model has a good denoising effect when the number of hidden neurons is 300, and the denoising model of the noise reduction self-encoder has better denoising performance and generalization performance than a wavelet threshold denoising method. The denoising model of the denoising self-encoder is applied to denoising acoustic emission signals, can effectively remove noise, and has important significance for subsequent acoustic emission signal identification processing.
Description
Technical Field
The invention belongs to the technical field of signal processing, and particularly relates to an acoustic emission signal denoising method based on a denoising autoencoder.
Background
When a material is deformed or cracks propagate under the action of external force or internal force, the phenomenon that strain energy is released in the form of elastic waves is called acoustic emission. The technique of analyzing acoustic emission signals using instrumental detection and inferring the source of the acoustic emission using the acoustic emission signals is known as acoustic emission detection. The acoustic emission detection technology is different from a conventional nondestructive detection method, is a passive dynamic detection method, does not need to enter a detected object for detection, has the unique advantages of instantaneity, integrity, high sensitivity and the like, and can dynamically monitor the health of a structure.
The collected acoustic emission signals are complex in components and contain a large amount of mechanical noise and electromagnetic noise, so that the elimination of the noise of the acoustic emission signals is the premise of analysis and identification of the acoustic emission signals. The early acoustic emission signal denoising technology comprises the traditional filtering denoising and Fourier transform methods, and the problem of loss of useful signals caused by smooth transient components of original signals exists; in the later period, a wavelet analysis method is applied to the field of noise removal of acoustic emission signals to obtain a certain achievement, a fixed threshold is set by Donoho according to the variance and the length of the noise signals, threshold processing is carried out on all high-frequency coefficients after wavelet decomposition, and Tang carries out noise reduction on vibration signals of a gear box of wind power equipment and vibration signals of a rolling bearing by utilizing Morlet wavelet transformation and a continuous wavelet transformation method, so that a certain noise removal effect is obtained. Although the wavelet transform has the characteristic of multi-resolution, the problems of wavelet basis function selection, stationarity assumption, parameter sensitivity and the like exist. Analysis of the acquired AE signals in actual engineering measurement to infer the types and degrees of defects and damages in the structure is a typical nonlinear pattern recognition problem and is difficult to complete through a traditional threshold setting and AE characteristic parameter correlation analysis method.
Disclosure of Invention
Aiming at the existing problems, the invention provides a noise-reduction self-encoder-based acoustic emission signal denoising method, which comprises the steps of extracting deep-level representation characteristics from a noise-containing acoustic emission signal through an encoder, reconstructing the signal by using a decoder for the extracted characteristics, and enabling a network to learn more robust invariance characteristics through unsupervised learning training DAE so as to obtain more effective input expression and enable the error between the reconstructed signal and an original signal to be converged to a minimum value, thereby achieving the purpose of denoising.
In order to achieve the purpose, the technical solution adopted by the invention is as follows:
a noise reduction self-encoder-based acoustic emission signal denoising method is characterized by comprising the following steps:
step 1: collecting an original acoustic emission signal by using an acoustic emission monitor;
step 2: establishing a DAE-based acoustic emission signal denoising model;
and step 3: inputting the acquired acoustic emission signals into a DAE-based acoustic emission signal denoising model, and reconstructing signals;
and 4, step 4: training a DAE-based acoustic emission signal denoising model through unsupervised learning, and enabling errors between a reconstructed signal and an original signal to be converged to a minimum value;
and 5: finally, the acoustic emission signal after denoising is obtained.
Further, the acoustic emission signal denoising model of the DAE in step 2 includes an input layer, a hidden layer, and an output layer, and the number of neurons in the output layer is equal to that in the input layer.
Further, the specific operation steps of step 3 include:
step 31: inputting the collected acoustic emission signals;
step 32: to the acoustic emission signal who gathers, damage of making an uproar according to binomial distribution probability carries out, generates the damage acoustic emission signal that makes an uproar, and its binomial distribution formula is:
wherein n is the number of signal sampling points, p is the probability of occurrence of an event A in each test, and the event A refers to zero element generation; x represents the occurrence frequency of the event A in the n-fold Bernoulli test, and the possible value is 0, 1, …, n; for each k (k is more than or equal to 0 and less than or equal to n), the event { X ═ k } is that the event a happens k times in n times of experiments;
the two-term distribution depends on the parameters n and p, and a one-dimensional data sequence with the same number as the sampling points of the input acoustic emission signals is generated through a distribution expected value;
step 33: inputting the noise-added damaged acoustic emission signal to an input layer of a noise elimination model of an acoustic emission signal of the DAE, coding the signal, and extracting deep representation characteristics;
step 34: the extracted feature utilization is decoded for signal reconstruction.
Further, the formula of the encoding described in step 33 is:
the decoding formula in step 34 is:
wherein z is the output of a certain nerve unit of the DAE hidden layer, and the calculation formula is as follows:
wherein n is the number of input units connected to the neural unit, wiFor corresponding connection weights, xiFor input to the corresponding input unit.
Further, in step 32, z-score normalization is firstly performed on the collected acoustic emission signals, and the calculation formula is as follows:
wherein, x' is the data after the acoustic emission signal is standardized, mu is the mean value of the acoustic emission signal, sigma is the standard deviation of the acoustic emission signal, and x is the original data of the acoustic emission signal.
Compared with the prior art, the method has the following beneficial effects:
firstly, the method provided by the invention trains the noise reduction self-encoder to reconstruct a noise-added emission signal based on an unsupervised learning mode so as to achieve the purpose of denoising, an experimental result shows that the DAE denoising model has an advantage in denoising effect when the number of hidden neurons is 300, and the provided DAE denoising model has better denoising performance and generalization compared with a wavelet threshold denoising method, thereby providing a basis for next step of acoustic emission signal identification.
Drawings
FIG. 1 is a schematic diagram of AE;
FIG. 2 is a de-noising schematic diagram of the DAE;
FIG. 3 is a schematic block diagram of a DAE-based acoustic emission signal denoising model;
FIG. 4 is a schematic diagram of a sample of an acoustic emission signal of corrosion in an embodiment;
FIG. 5 is a schematic diagram of a sample of an acoustic emission signal after normalization by z-score in an embodiment;
FIG. 6 is a schematic diagram of an erosion acoustic emission signal with Gaussian white noise added in the embodiment;
FIG. 7 is a schematic illustration of a sample of an acoustic emission signal of corrosion after noise damage in an embodiment;
FIGS. 8(a) - (e) are diagrams of DAE model training processes with hidden neuron numbers 500, 400, 300, 200, and 100, respectively;
FIG. 9 is a fitting curve of SNR distribution histograms of DAEs with different numbers of hidden neurons after denoising an acoustic emission signal;
FIG. 10 is a schematic diagram of a DAE neural network model architecture;
FIG. 11 shows SNR distributions of the DAE denoising model and the wavelet threshold denoising method in the embodiment.
Detailed Description
In order to make those skilled in the art better understand the technical solution of the present invention, the following further describes the technical solution of the present invention with reference to the drawings and the embodiments.
1. Denoising principle of denoising autoencoder
The DAE is a derivative model of an Auto Encoder (AE), and the AE structure is shown in fig. 1. The AE is similar to an unsupervised learning neural network model in nature, the number of hidden nodes is usually smaller than that of input nodes, the number of output nodes is the same as that of input nodes, and an approximate reconstruction signal is obtained by training the AE.
AE includes encoding (encoder) and decoding (decoder) processes, with input X ═ X1,x2,…,xn) If the output is Y ═ Y1,y2,…,yn)。
The encoding (encoder) process is as follows:
h=σ(WX+b) (1),
the decoding (decoder) process is:
Y=σ(W'h+b') (2),
wherein, W, b, W ', b' are corresponding weight and bias, and sigma is activation function;
defining the loss function as a least squares function as:
the core idea of DAE is to add noise to the input data X and damage it in such a way that the input data X is set to zero according to a certain probability, the probability distribution generally adopts binomial distribution, as shown in fig. 2, and the rest is similar to AE. The essence of the DAE is that in the process of encoding and decoding the signal, the input noise is "erased" in the process of reconstructing the signal, thereby achieving the purpose of denoising.
2. DAE-based acoustic emission signal denoising model
The DAE-based acoustic emission signal denoising model principle structure is shown in figure 3, the number of hidden layer neurons is generally smaller than that of input neurons, and in order to achieve the reconstruction purpose, the number of output layer neurons is equal to that of input layers.
And (3) carrying out noise adding damage on the acquired acoustic emission signals according to a binomial distribution probability, wherein the binomial distribution is shown as a formula (4), and the expected value of the distribution is shown as a formula (5).
E(X)=np (5),
The two-term distribution depends on parameters n and p, a one-dimensional data sequence with the same number as the sampling points of the input acoustic emission signals is generated through the distribution, 0 element index in the sequence is used as a basis for setting zero of corresponding input data of the acoustic emission signals, and in order to ensure that certain zero elements are generated, n is 10, and p is 0.1.
The noise reduction self-encoder is a single hidden layer, the encoder and decoder activation functions can select a logical sigmoid function or a Positive designing linear transfer function, the logical sigmoid function is determined by equation (6), and the Positive designing linear transfer function is determined by equation (7):
z in the formula (6) is determined by the formula (8), and z is the output of a certain nerve unit of the hidden layer of the DAE;
n in the formula (8) is the number of input units connected to the neural unit, wiFor corresponding connection weights, xiIs the input of the corresponding input unit.
3. Acquisition and processing of acoustic emission signals
3011 corrosion acoustic emission signal samples are collected through a 5% brine corrosion simulation test at the early stage, and the number of sampling points is 8192. The corrosion acoustic emission signal sample is shown in figure 4, in order to eliminate the difference between signal data and improve the accurate identification rate and the convergence rate of the DAE model, z-score standardization processing is carried out on each acquired acoustic emission signal, and calculation is carried out according to the public display (9).
Wherein x' is the data after the acoustic emission signal is standardized, mu is the mean value of the acoustic emission signal, sigma is the standard deviation of the acoustic emission signal, and x is the original data of the acoustic emission signal;
the acoustic emission signal sample after z-score standardization is shown in fig. 5, and in order to facilitate the subsequent experiment to check the denoising performance, gaussian white noise is added to the standardized acoustic emission signal, the noise power is 10dBW, and the acoustic emission signal added with the gaussian white noise is shown in fig. 6.
Before the DAE is input after Gaussian white noise is added to an acoustic emission signal, the acoustic emission signal needs to be subjected to noise adding and damage, namely, input data is set to be zero according to binomial distribution probability, n is 10 and p is 0.1 for related parameters of binomial distribution. A sample of the noise corrupted acoustic emission signal is shown in fig. 7.
Examples
1. Examples of the examples
(1) Determination of DAE hidden layer neuron number
The DAE is a neural network with a single hidden layer structure, the number of hidden layer neurons needs to be determined according to the DAE according to an actual denoising application scene, and the denoising performance is measured by adopting a signal-to-noise ratio, as shown in formula (10):
wherein, Psignal,PnoisePower of the signal and noise, respectively; and P issignalNormalized corrosion acoustic emission signal power, P, as shown in FIG. 5noiseFor noise-eliminated acoustic emission signal power and PsignalThe difference between them.
The DAE coding activation function selects a Logistic sigmoid function, the decoding activation function selects a Positive designing linear transfer function, and the loss function adopts a mean square error function MSE (mean Squared error), as shown in formula (11):
wherein x isiIn order to be the true value of the value,is a predicted value, where m is the number of sampling points;
the neural network training function adopts a quantized conjugate gradient back propagation algorithm (Scaled conjugate gradient back propagation), the training algorithm can be adaptively trained, parameters do not need to be set, and the training efficiency is high.
30% of the 3011 samples of the processed corrosive acoustic emission signals were extracted as DAE test data, 70% were extracted as DAE training data, and the maximum number of training rounds MaxEpochs was 20. The training process of the DAE model with the hidden neuron numbers of 500, 400, 300, 200 and 100 is shown in fig. 8(a) - (e) in sequence. As can be seen from FIG. 8, the performance of the model is worse as the number of hidden neurons decreases, but the performance gap is smaller.
And (3) carrying out denoising test on 904 samples of extracted 30% acoustic emission signal test data after DAE denoising model training convergence, wherein an MATLAB 2019a is adopted in the test, fitting curves of SNR distribution histograms of DAEs with different numbers of hidden neurons after denoising of the acoustic emission signals are shown in the attached drawing 9, and M in the drawing is the number of hidden neurons, and values of M are respectively 500, 400, 300, 200 and 100. As can be seen from fig. 9, with the decrease of the number of hidden neurons, the noise cancellation capability of the DAE is generally weakened, a large span exists between 300 and 200 hidden neurons, and the number of the hidden neurons of the DAE is set to 300 in the present embodiment, considering that the DAE input is 8192 sampling points, the model training convergence time is long, and the noise cancellation performance and the model running time are comprehensively considered, and the structure of the DAE neural network model is as shown in fig. 10.
(2) Comparing denoising performance with wavelet threshold denoising method
Then, comparing denoising performances of the DAE denoising model of the acoustic emission signal with a Donoho wavelet threshold denoising method, and analyzing the denoising performances of the DAE denoising model and the Donoho wavelet threshold denoising method:
1000 samples are randomly extracted from the processed acoustic emission signal as test data, and the SNR distribution of the DAE denoising model and the wavelet threshold denoising method is shown in fig. 11. As can be seen in the figure, the SNR mean value of 1000 acoustic emission signal samples after being denoised by DAE is 6.12, the variance is 0.25, the SNR mean value after being denoised by wavelet threshold is 5.23, and the variance is 0.45.
In the acoustic emission signal denoising performance, the DAE denoising model denoising performance is generally superior to that of a wavelet threshold denoising method, the variance is small, and the robustness is better.
2. Conclusion of the experiment
Through the verification process, the invention provides a DAE-based acoustic emission signal denoising model, and a denoising self-encoder is trained to reconstruct a denoising emission signal based on an unsupervised learning mode so as to denoise. Experimental results show that the DAE denoising model has advantages in denoising effect and model convergence speed when the number of hidden neurons is 300, and compared with a Donoho wavelet threshold denoising method, the provided DAE denoising model has better denoising performance and robustness. The DAE denoising model is applied to denoising acoustic emission signals, can effectively remove noise, and has important significance for subsequent acoustic emission signal identification processing.
Those not described in detail in this specification are within the skill of the art. Although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that various changes in the embodiments and modifications of the invention can be made, and equivalents of some features of the invention can be substituted, and any changes, equivalents, improvements and the like, which fall within the spirit and principle of the invention, are intended to be included within the scope of the invention.
Claims (5)
1. A noise reduction self-encoder-based acoustic emission signal denoising method is characterized by comprising the following steps:
step 1: collecting an original acoustic emission signal by using an acoustic emission monitor;
step 2: establishing a DAE-based acoustic emission signal denoising model;
and step 3: inputting the acquired acoustic emission signals into a DAE-based acoustic emission signal denoising model, and reconstructing signals;
and 4, step 4: training a DAE-based acoustic emission signal denoising model through unsupervised learning, and enabling errors between a reconstructed signal and an original signal to be converged to a minimum value;
and 5: finally, the acoustic emission signal after denoising is obtained.
2. The method of claim 1, wherein the noise reduction model of DAE comprises an input layer, a hidden layer and an output layer, and the number of neurons in the output layer is equal to that in the input layer.
3. The method for denoising acoustic emission signals based on a denoising self-encoder as claimed in claim 2, wherein the specific operation step of step 3 comprises:
step 31: inputting the collected acoustic emission signals;
step 32: to the acoustic emission signal who gathers, damage of making an uproar according to binomial distribution probability carries out, generates the damage acoustic emission signal that makes an uproar, and its binomial distribution formula is:
wherein n is the number of signal sampling points, p is the probability of occurrence of an event A in each test, and the event A refers to zero element generation; x represents the occurrence frequency of the event A in the n-fold Bernoulli test, and the possible value is 0, 1, …, n; for each k (k is more than or equal to 0 and less than or equal to n), the event { X ═ k } is that the event a happens k times in n times of experiments;
the two-term distribution depends on the parameters n and p, and a one-dimensional data sequence with the same number as the sampling points of the input acoustic emission signals is generated through a distribution expected value;
step 33: inputting the noise-added damaged acoustic emission signal to an input layer of a noise elimination model of an acoustic emission signal of the DAE, coding the signal, and extracting deep representation characteristics;
step 34: the extracted feature utilization is decoded for signal reconstruction.
4. The method of claim 3, wherein the formula of the encoding in step 33 is:
the decoding formula in step 34 is:
wherein z is the output of a certain nerve unit of the DAE hidden layer, and the calculation formula is as follows:
wherein n is the number of input units connected to the neural unit, wiFor corresponding connection weights, xiFor input to the corresponding input unit.
5. The method as claimed in claim 3, wherein z-score normalization is performed on the acoustic emission signals collected in step 32, and the calculation formula is:
wherein, x' is the data after the acoustic emission signal is standardized, mu is the mean value of the acoustic emission signal, sigma is the standard deviation of the acoustic emission signal, and x is the original data of the acoustic emission signal.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110308357.6A CN112926504A (en) | 2021-03-23 | 2021-03-23 | Acoustic emission signal denoising method based on noise reduction self-encoder |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110308357.6A CN112926504A (en) | 2021-03-23 | 2021-03-23 | Acoustic emission signal denoising method based on noise reduction self-encoder |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112926504A true CN112926504A (en) | 2021-06-08 |
Family
ID=76175635
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110308357.6A Pending CN112926504A (en) | 2021-03-23 | 2021-03-23 | Acoustic emission signal denoising method based on noise reduction self-encoder |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112926504A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114039781A (en) * | 2021-11-10 | 2022-02-11 | 湖南大学 | Slow denial of service attack detection method based on reconstruction abnormity |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109784692A (en) * | 2018-12-29 | 2019-05-21 | 重庆大学 | A kind of fast and safely constraint economic load dispatching method based on deep learning |
CN109965885A (en) * | 2019-04-24 | 2019-07-05 | 中国科学院电子学研究所 | A kind of BCG signal de-noising method and device based on denoising autocoder |
CN110044554A (en) * | 2019-04-16 | 2019-07-23 | 重庆大学 | A kind of online test method of the metal pressure container leakage based on acoustic emission signal |
CN110879254A (en) * | 2018-09-05 | 2020-03-13 | 哈尔滨工业大学 | Steel rail crack acoustic emission signal detection method based on improved least square generation type countermeasure network |
CN111144303A (en) * | 2019-12-26 | 2020-05-12 | 华北电力大学(保定) | Power line channel transmission characteristic identification method based on improved denoising autoencoder |
-
2021
- 2021-03-23 CN CN202110308357.6A patent/CN112926504A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110879254A (en) * | 2018-09-05 | 2020-03-13 | 哈尔滨工业大学 | Steel rail crack acoustic emission signal detection method based on improved least square generation type countermeasure network |
CN109784692A (en) * | 2018-12-29 | 2019-05-21 | 重庆大学 | A kind of fast and safely constraint economic load dispatching method based on deep learning |
CN110044554A (en) * | 2019-04-16 | 2019-07-23 | 重庆大学 | A kind of online test method of the metal pressure container leakage based on acoustic emission signal |
CN109965885A (en) * | 2019-04-24 | 2019-07-05 | 中国科学院电子学研究所 | A kind of BCG signal de-noising method and device based on denoising autocoder |
CN111144303A (en) * | 2019-12-26 | 2020-05-12 | 华北电力大学(保定) | Power line channel transmission characteristic identification method based on improved denoising autoencoder |
Non-Patent Citations (1)
Title |
---|
周俊: "基于降噪自编码器的声发射信号去噪研究", 重庆理工大学学报, 31 December 2020 (2020-12-31) * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114039781A (en) * | 2021-11-10 | 2022-02-11 | 湖南大学 | Slow denial of service attack detection method based on reconstruction abnormity |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108318249B (en) | Fault diagnosis method for rotary mechanical bearing | |
Zhou et al. | Novel synthetic index-based adaptive stochastic resonance method and its application in bearing fault diagnosis | |
Lin et al. | Rolling bearing fault diagnosis using impulse feature enhancement and nonconvex regularization | |
CN113094993B (en) | Modulation signal denoising method based on self-coding neural network | |
CN110795843B (en) | Method and device for identifying faults of rolling bearing | |
Chen et al. | Fault feature extraction and diagnosis of gearbox based on EEMD and deep briefs network | |
Liu et al. | Fault diagnosis of rotor using EMD thresholding-based de-noising combined with probabilistic neural network | |
CN116405109A (en) | Optical module communication self-adaptive modulation method based on linear direct drive | |
CN111580161A (en) | Earthquake random noise suppression method based on multi-scale convolution self-coding neural network | |
CN115758082A (en) | Fault diagnosis method for rail transit transformer | |
CN112926504A (en) | Acoustic emission signal denoising method based on noise reduction self-encoder | |
CN113159088B (en) | Fault monitoring and diagnosis method based on multi-feature fusion and width learning | |
CN117473414A (en) | Bearing fault position identification method based on low-noise time-frequency image | |
Cui et al. | A GAN noise modeling based blind denoising method for guided waves | |
CN117111155A (en) | Microseism data denoising method based on integrated framework | |
CN112539887A (en) | WT-LCD-WD-based pipeline leakage signal denoising method | |
CN115859054B (en) | MIC-CEEMDAN-based hydroelectric generating set draft tube pressure pulsation signal denoising method | |
Tran et al. | Denoising induction motor sounds using an autoencoder | |
CN115563480A (en) | Gear fault identification method for screening octave geometric modal decomposition based on kurtosis ratio coefficient | |
CN113409213B (en) | Method and system for enhancing noise reduction of time-frequency diagram of fault signal of plunger pump | |
CN113567129A (en) | CEEMD-based noise reduction method for train bearing vibration signal | |
Zhou et al. | Research on acoustic emission signal denoising based on autoencoder | |
Ma et al. | Fault diagnosis of motor bearing equipment based on vibration signals | |
CN115114955B (en) | Equipment fault detection method based on sound and vibration signals | |
CN117334203A (en) | Dog variety identification method using MFCC-DSRU |
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 |