CN109344751B - Reconstruction method of noise signal in vehicle - Google Patents
Reconstruction method of noise signal in vehicle Download PDFInfo
- Publication number
- CN109344751B CN109344751B CN201811102322.1A CN201811102322A CN109344751B CN 109344751 B CN109344751 B CN 109344751B CN 201811102322 A CN201811102322 A CN 201811102322A CN 109344751 B CN109344751 B CN 109344751B
- Authority
- CN
- China
- Prior art keywords
- component
- signal
- reconstruction
- noise
- vehicle
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Biophysics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Signal Processing (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
- Fittings On The Vehicle Exterior For Carrying Loads, And Devices For Holding Or Mounting Articles (AREA)
Abstract
The invention relates to a reconstruction method of noise signals in a vehicle, which comprises the following steps: 1) signal decomposition analysis: carrying out signal decomposition analysis on the source signal to obtain three stable signal component categories, namely a high-frequency component, a medium-frequency component and a low-frequency component; 2) calculating component fitness: constructing a BP neural network model for respective training, and taking the weight and the threshold of the BP neural network model as a component fitness value to obtain an optimal component fitness value; 3) and (3) signal reconstruction model: and according to the category of the input signal components, giving the optimal component fitness value to a noise reconstruction BP network as an initial weight and a threshold value, training, acquiring a reconstruction algorithm model corresponding to each type of signal component after convergence, and performing reconstruction superposition according to the reconstruction algorithm model to complete reconstruction of the noise signal at the ear side of the passenger. Compared with the prior art, the method has the advantages of reducing the non-stationarity and modeling difficulty of the signal, improving the reconstruction precision and the like.
Description
Technical Field
The invention relates to the field of signal processing and information fusion, in particular to a reconstruction method of an in-vehicle noise signal.
Background
In order to realize active control (ANC) of ear-side noise of an occupant in a vehicle, a primary reference signal is first provided for a control system. For the pickup of the primary reference signal, the traditional method is to install a microphone near the ear side of the passenger to acquire the primary reference signal, and the method inevitably introduces secondary pollution of a secondary sound source and is not beneficial to the rapid convergence of the system. Therefore, it is of certain significance to research a reconstruction method of ear-side noise of an occupant in a vehicle and obtain a reference signal of ANC.
Currently, the main methods for sound field reconstruction include near-field acoustic holography (NAH), multi-sensor data fusion (MSDF), and the like. The time-domain NAH method needs to be based on the free sound field assumption, but is often difficult to satisfy in practical applications. In consideration of the reconstruction of the ear-side noise of the passenger, the ear-side signal of the passenger is reconstructed in order to realize the utilization of the noise source signal outside the vehicle. The data fusion method provides a theoretical basis for solving the problems of multi-source data feature extraction and fusion modeling. The multi-source data fusion is integrated from multi-sensor data and information through established rules and analysis methods, and the knowledge of the consistency of the observed target is obtained on the basis.
The MSDF method has been developed from traditional methods such as kalman filtering, bayesian estimation, maximum likelihood estimation, cluster analysis, etc. to intelligence based on self-organizing maps (SOM), Adaptive Weighted Fusion (AWF), Fuzzy Logic Identification (FLI), Support Vector Machines (SVM), and Artificial Neural Networks (ANN). Among them, ANN is widely used, and most of them employ a BP (back propagation) algorithm. The BP neural network establishes a model according to the intrinsic relation of the data, automatically extracts related knowledge from the data, and has self-learning, self-organizing and self-adapting capabilities.
However, in-vehicle noise signals belong to mechanical vibration and acoustic signals and have strong nonlinearity and non-stationarity, and although the BP neural network has good nonlinear fitting capability and can be well applied to a multi-source data fusion technology, errors caused by processing signals with the characteristic of non-stationarity cannot be avoided.
Disclosure of Invention
The present invention is directed to a method for reconstructing a noise signal in a vehicle to overcome the above-mentioned drawbacks of the prior art.
The purpose of the invention can be realized by the following technical scheme:
a reconstruction method of an in-vehicle noise signal includes the following steps:
1) signal decomposition analysis: acquiring normal noise source signal data, preprocessing the data, and performing signal decomposition analysis on a source signal to obtain three stable signal component categories, namely a high-frequency component, an intermediate-frequency component and a low-frequency component;
2) calculating component fitness: respectively taking the noise source signal components of the three categories as input, taking the noise signal component in the vehicle as expected output, constructing a BP neural network model for respective training, and taking the weight and the threshold of the BP neural network model as component fitness values to obtain an optimal component fitness value;
3) and (3) signal reconstruction model: and according to the category of the input signal components, giving the optimal component fitness value to a noise reconstruction BP network as an initial weight and a threshold value, training by an error back propagation method, acquiring a reconstruction algorithm model corresponding to each type of signal components after convergence, and reconstructing and superposing according to the reconstruction algorithm model to complete the reconstruction of the noise signal at the ear side of the passenger.
The step 1) specifically comprises the following steps:
11) decomposing the source signal x (t) based on eigenmode decomposition, then:
wherein, ci(t) is the i-th IMF component of the signal, n is the total number of IMF components, rnIs the residual component of the output signal;
12) acquiring the number of extreme points of each IMF component in each second and the energy ratio of each IMF component in the original signal;
13) setting an extreme point number threshold (N, M), dividing the IMF component into a high-frequency component, a medium-frequency component and a low-frequency component according to the number of the extreme points of each component, and then:
X(t)=d1+d2+d3
wherein d is1,d2,d3Respectively representing a high frequency component, a medium frequency component, and a low frequency component.
In the step 2), the constructed BP neural network model is a three-layer BP network.
In step 2), the expression of the component fitness value is as follows:
wherein the content of the first and second substances,is diThe component fitness value for a component, and i is 1,2,3,for the weight matrix of the input layer to the hidden layer,is a weight matrix from hidden layer to output layer,in order to be a hidden layer threshold value,is the output layer threshold.
In the step 2), when the output learning error of the BP neural network model is smaller than the expected output error or the iteration step number reaches the maximum set step number, the corresponding component fitness value is the optimal component fitness value, and then:
wherein the content of the first and second substances,in order to optimize the value of the component fitness,the component fitness value updated for the iterative r step.
In the step 13), the value of the threshold value (N, M) of the number of extreme points is (80, 20).
Compared with the prior art, the invention has the following advantages:
the signal reconstruction method can realize reconstruction of signals in the vehicle, and through the SDA method, the non-stationarity of the signals is reduced to a great extent, and the modeling difficulty is reduced; meanwhile, the optimal fitness value obtained by CFC is endowed to a noise reconstruction BP network as an initial weight and a threshold value, so that the problem of random generation of the weight and the threshold value of the BP network can be solved, and the reconstruction precision is improved
Drawings
Fig. 1 is a flow chart of a signal reconstruction method according to the present invention.
Fig. 2 is a diagram of a neural network architecture for noise reconstruction.
Fig. 3 is a diagram for analyzing the contribution amount of each key point.
Fig. 4 is an energy ratio diagram of each IMF component.
Fig. 5(a) is a graph comparing the time domain results of the signal reconstruction with the original signal for the algorithm proposed herein.
Fig. 5(b) is a plot of the signal reconstruction versus the frequency domain results of the original signal for the algorithm proposed herein.
Fig. 6(a) is a comparison graph of the time domain result of the signal reconstruction and the original signal of the BP reconstruction algorithm.
Fig. 6(b) is a comparison graph of the signal reconstruction of the BP reconstruction algorithm and the frequency domain result of the original signal.
Detailed Description
Based on the defects of the prior art, the technical problem to be solved by the present invention is to provide a method for reconstructing an in-vehicle signal, in order to reconstruct an in-vehicle noise signal, a key point signal affecting the in-vehicle noise is used as input data, and a noise reconstruction model is established based on a BP neural network to obtain a noise signal on the ear side of a passenger. Considering the characteristics of a high-speed working condition noise signal and the limitation of a BP neural network, firstly, decomposing and reconstructing the signal by an SDA method, reducing the non-stationarity of the original signal, obtaining three signal components with relatively high stationarity, guiding the motion of different components according to respective categories based on a neural network model, and defining CFC to update the fitness value of the components, wherein the fitness value represents the weight and the threshold value of the neural network. And then, training the three types of signal component data by using a large amount of signal component data to obtain an optimal component fitness value, and carrying out initialization assignment on the weight and the threshold of the network. And finally, obtaining an intelligent in-vehicle noise signal reconstruction algorithm model based on the BP neural network by using the test data. Through experimental verification, the algorithm can effectively reconstruct the noise signal on the ear side of the passenger.
In order to achieve the purpose, the invention adopts the following technical scheme:
a multi-source signal-based in-vehicle noise signal reconstruction method comprises the following steps:
in step 1:
in a semi-anechoic chamber, acquiring sound-sound transfer functions of all transfer paths based on a reciprocity method, ensuring that a test position is consistent with an acquired road test position, and selecting a key measuring point signal with the most correlation to reconstruct the noise in the vehicle;
calculating the contribution of each noise source signal to the in-vehicle signal by using an acoustic transfer function (TPA) contribution analysis method, wherein the formula is as follows:
wherein, PMNThe total sound pressure of the ear side of the passenger in the vehicle;contribution amount of sound pressure on the transmission path i;is a function of the transfer characteristic on path i;is the working input on the transfer path i.
the data signal is processed by EMD to obtain finite relatively stable IMF components and a trend term rn;
By using the EED method provided by the invention, the number of extreme points and the energy ratio of each IMF component are solved, and an appropriate (N, M) value is determined to obtain 3 components, namely a high-frequency component, a medium-frequency component and a low-frequency component.
X(t)=d1+d2+d3 (3)
an i-th noise source signal component is used as an input signal (i is 1,2 and 3), an i-th in-vehicle noise signal component is used as a desired output signal, and an initial fitness valueRandomly generating, wherein the weight value and the threshold value of the neural network are used as component fitness values;
the method adopts an n-m-l three-layer neural network basic structure, wherein n, m and l respectively represent the neuron numbers of an input layer, a hidden layer and an output layer, and the weight value and the threshold matrix of the network are expressed as follows;
the initial value r of the iteration times is 0; comparing the minimum learning error to meet the expected error;
computing a forward propagated output signal O of neuron j using input dataj;
Wherein, wijAnd thetajRespectively representing the weight and the threshold value of the neuron j, wherein f is an excitation function, and the excitation function uses a hyperbolic tangent sigmoid function.
Calculating an error E;
wherein y isd,kAnd ykRespectively representing expected outputs of neurons in the output layerAn output value and a true output value, N is the number of data samples.
If E is less than or equal to gamma or r is more than or equal to max _ times, the signal component d is outputiIs optimized to the fitness value
decomposing and reconstructing a noise test signal, and performing a flow step 1;
determining the signal component diAnd optimum moderation valueBelong to the same class, handleThe noise is assigned to reconstruct the BP network as initial weights and thresholds in matrices 4, 5, 6, 7, and the signal component d is usediTraining a noise reconstruction BP network;
the initial value r of the iteration times is 0; comparing the minimum learning error to meet the expected error;
computing a forward propagated output signal O of neuron j using input dataj;
Calculating error E, if E is less than or equal to gamma or r is greater than or equal to max _ times, obtaining signal component diThe noise of (3) reconstructs the BP network.
Repeating the above process to obtain the whole signal reconstruction model;
and 7, inputting each noise source signal into the signal reconstruction model, and outputting the in-vehicle noise reconstruction signal.
The following detailed description of the embodiments, taken in conjunction with the accompanying drawings, illustrate by way of example and, together with the description, further serve to explain the principles of the invention.
Examples
Next, using the acquired normal noise source signal data under the high-speed working condition, as shown in fig. 3, the contribution amount analysis diagram of each key point is used to determine the key noise source signal affecting the noise signal in the vehicle. And preprocessing the data, acquiring a noise source signal sample library, determining that the number of neurons in an input layer of the BP network is n-4, and determining that the number of neurons in an output layer of the BP network is 1.
As shown in fig. 1, which is a schematic flow chart of signal reconstruction according to an embodiment of the present invention,
firstly, decomposing and reconstructing a data signal by an SDA method, and performing EMD decomposition processing on the data signal to obtain finite relatively stable IMF components and a trend term rn;
With the EMD method proposed herein, the number of extreme points and the energy ratio of each IMF component are solved, see fig. and table 1, and a suitable (N, M) ═ 80, 20 is determined to yield 3 components, i.e., high frequency component, medium frequency component, low frequency component.
X(t)=d1+d2+d3 (2)
TABLE 1 number of extreme points (/ s) and energy ratio of IMF component of in-vehicle noise signal
Secondly, determining that a 4-m-1 3-layer BP neural network basic structure is adopted, and based on the signal source m being 128, as shown in fig. 2, taking a noise source signal as an input signal and taking an in-vehicle noise signal as an expected output value;
the weight and threshold matrix of the network are represented as follows;
the weight and threshold of the neural network are used as component fitness value and initial component fitness valueCan be expressed as:
thirdly, calculating an optimal fitness value:
an i-th noise source signal component is used as an input signal (i is 1,2 and 3), an i-th in-vehicle noise signal component is used as a desired output signal, and an initial fitness valueRandomly generating;
the initial value r of the iteration times is 0; comparing the minimum learning error to meet the expected error;
computing a forward propagated output signal O of neuron j using input dataj;
Wherein, wijAnd thetajRespectively representing the weight and the threshold value of the neuron j, wherein f is an excitation function, and the excitation function uses a hyperbolic tangent sigmoid function.
Calculating an error E;
wherein y isd,kAnd ykRepresenting the expected output value and the real output value of the output layer neuron, respectively, and N is the number of data samples.
If E is less than or equal to gamma or r is more than or equal to max _ times, the signal component d is outputiIs optimized to the fitness value
Fourthly, establishing a signal reconstruction model:
decomposing and reconstructing the noise test signal;
determining the signal component diAnd optimum moderation valueBelong to the same class, handleAssigning noise to reconstruct the BP network as initial weights and thresholds in matrices 3, 4, 5, 6, and using the signal component diTraining a noise reconstruction BP network;
the initial value r of the iteration times is 0; comparing the minimum learning error to meet the expected error;
computing a forward propagated output signal O of neuron j using input dataj;
Calculating error E, if E is less than or equal to gamma or r is greater than or equal to max _ times, obtaining signal component diThe noise of (3) reconstructs the BP network.
And repeating the above processes to obtain the whole signal reconstruction model.
And verifying the effectiveness of the in-vehicle noise signal reconstruction algorithm model.
Fig. 5 shows the result of the signal reconstruction in comparison with the original signal. As can be seen from the figure, the neural network algorithm provided by the invention can correctly reconstruct the position and the amplitude of the original signal for the noise signal in the vehicle, and the result shows that the method has higher signal reconstruction performance.
Performance comparison of different algorithms
Fig. 5 and 6 compare the performance of the signal reconstruction BP network and the network algorithm model proposed herein. As can be seen from the figure, the two reconstruction algorithms can correctly reconstruct the position and the amplitude of the original signal for the noise signal in the vehicle, and the result shows that the method has higher signal reconstruction performance.
In order to further analyze the accuracy of the reconstruction result, the mean square error E is used for characterization, and the formula is shown.
Wherein y isd,iAnd yiRespectively representing a reconstruction value and a true value, and N is the number of samples.
To better illustrate the results, for yd,i、yiAnd (6) carrying out normalization processing. Shown in the formula.
Wherein, ymin,ymaxRespectively representing the minimum and maximum values of the vector y. y is1The normalized result is represented.
The normalized result is substituted in equation (11) to obtain the reconstructed root mean square relative error shown in table 2.
TABLE 2 reconstructed RMS relative error
As can be seen from Table 2, for the reconstruction problem of the noise signal at the ear side of the passenger under the high-speed working condition, the algorithm model provided by the invention is superior to the BP algorithm model, wherein the accuracy of the result without normalization is improved by 39.33%, the accuracy of the result with normalization is improved by 68.92%, wherein the error of the result with normalization is 0.0023 and is less than the expected error of 0.005, which indicates that the reconstruction model for the noise signal in the vehicle provided by the invention can meet the accuracy requirement
In conclusion, the reconstruction of the noise signal in the vehicle can avoid the secondary pollution of the secondary sound source. Meanwhile, the reconstruction of the signals in the vehicle can be realized through the signal reconstruction method, and the instability of the signals is reduced to a great extent and the modeling difficulty is reduced through the SDA method; meanwhile, the optimal fitness value obtained by CFC is given to the noise reconstruction BP network as the initial weight and the threshold value, so that the problem that the weight and the threshold value of the BP network are randomly generated can be solved, and the reconstruction precision is improved.
Claims (6)
1. A method for reconstructing a noise signal in a vehicle, comprising the steps of:
1) signal decomposition analysis: acquiring normal noise source signal data, preprocessing the data, and performing signal decomposition analysis on a source signal to obtain three stable signal component categories, namely a high-frequency component, an intermediate-frequency component and a low-frequency component;
2) calculating component fitness: respectively taking the noise source signal components of the three categories as input, taking the noise signal component in the vehicle as expected output, constructing a BP neural network model for respective training, and taking the weight and the threshold of the BP neural network model as component fitness values to obtain an optimal component fitness value;
3) and (3) signal reconstruction model: and according to the category of the input signal components, giving the optimal component fitness value to a noise reconstruction BP network as an initial weight and a threshold value, training by an error back propagation method, acquiring a reconstruction algorithm model corresponding to each type of signal components after convergence, and reconstructing and superposing according to the reconstruction algorithm model to complete the reconstruction of the noise signal at the ear side of the passenger.
2. The method for reconstructing an in-vehicle noise signal according to claim 1, wherein the step 1) specifically includes the steps of:
11) decomposing the source signal x (t) based on eigenmode decomposition, then:
wherein, ci(t) is the i-th IMF component of the signal, n is the total number of IMF components, rnIs the residual component of the output signal;
12) acquiring the number of extreme points of each IMF component in each second and the energy ratio of each IMF component in the original signal;
13) setting an extreme point number threshold (N, M), dividing the IMF component into a high-frequency component, a medium-frequency component and a low-frequency component according to the number of the extreme points of each component, and then:
X(t)=d1+d2+d3
wherein d is1,d2,d3Respectively representing a high frequency component, a medium frequency component, and a low frequency component.
3. The method for reconstructing noise signals in a vehicle according to claim 1, wherein the BP neural network model constructed in the step 2) is a three-layer BP network.
4. The method for reconstructing an in-vehicle noise signal according to claim 3, wherein in the step 2), the expression of the component fitness value is as follows:
wherein the content of the first and second substances,is diThe component fitness value for a component, and i is 1,2,3,as the weight moment from input layer to hidden layerThe number of the arrays is determined,is a weight matrix from hidden layer to output layer,in order to be a hidden layer threshold value,is the output layer threshold.
5. The method according to claim 4, wherein in the step 2), when the output learning error of the BP neural network model is smaller than the expected output error or the iteration step number reaches the maximum set step number, the corresponding component fitness value is the optimal component fitness value, and then:
6. The method for reconstructing noise signals in a vehicle according to claim 2, wherein in the step 13), the threshold value of the number of extreme points (N, M) is (80, 20).
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811102322.1A CN109344751B (en) | 2018-09-20 | 2018-09-20 | Reconstruction method of noise signal in vehicle |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811102322.1A CN109344751B (en) | 2018-09-20 | 2018-09-20 | Reconstruction method of noise signal in vehicle |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109344751A CN109344751A (en) | 2019-02-15 |
CN109344751B true CN109344751B (en) | 2021-10-08 |
Family
ID=65306251
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811102322.1A Active CN109344751B (en) | 2018-09-20 | 2018-09-20 | Reconstruction method of noise signal in vehicle |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109344751B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111487063B (en) * | 2020-03-20 | 2021-05-14 | 天津大学 | Parameter adaptive internal combustion engine noise source separation method |
CN111721401B (en) * | 2020-06-17 | 2022-03-08 | 广州广电计量检测股份有限公司 | Low-frequency noise analysis system and method |
CN115035911B (en) * | 2022-05-31 | 2023-05-02 | 中国第一汽车股份有限公司 | Noise generation model training method, device, equipment and medium |
CN117437931B (en) * | 2023-12-20 | 2024-04-02 | 深圳市新厚泰电子科技有限公司 | Sound signal optimized transmission method for microphone |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103049798A (en) * | 2012-12-05 | 2013-04-17 | 浙江大学城市学院 | Short-period electric generation power forecasting method applied to photovoltaic electric generation system |
CN103077267A (en) * | 2012-12-28 | 2013-05-01 | 电子科技大学 | Parameter sound source modeling method based on improved BP (Back Propagation) neural network |
CN103474066A (en) * | 2013-10-11 | 2013-12-25 | 福州大学 | Ecological voice recognition method based on multiband signal reconstruction |
CN104168131A (en) * | 2014-06-05 | 2014-11-26 | 国家电网公司 | Flow generation method of power dispatching exchange network based on multicast communication |
CN106769040A (en) * | 2016-12-14 | 2017-05-31 | 上海工程技术大学 | A kind of method of the sparse reconstruct of bearing vibration signal |
-
2018
- 2018-09-20 CN CN201811102322.1A patent/CN109344751B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103049798A (en) * | 2012-12-05 | 2013-04-17 | 浙江大学城市学院 | Short-period electric generation power forecasting method applied to photovoltaic electric generation system |
CN103077267A (en) * | 2012-12-28 | 2013-05-01 | 电子科技大学 | Parameter sound source modeling method based on improved BP (Back Propagation) neural network |
CN103474066A (en) * | 2013-10-11 | 2013-12-25 | 福州大学 | Ecological voice recognition method based on multiband signal reconstruction |
CN104168131A (en) * | 2014-06-05 | 2014-11-26 | 国家电网公司 | Flow generation method of power dispatching exchange network based on multicast communication |
CN106769040A (en) * | 2016-12-14 | 2017-05-31 | 上海工程技术大学 | A kind of method of the sparse reconstruct of bearing vibration signal |
Non-Patent Citations (3)
Title |
---|
A study on the emulsifier fault diagnosis of the BP neural network based on EMD and KPCA;Yue Sheng Wang等;《Proceedings of The 2016 2nd International Conference on Energy Equipment Science and Engineering》;20161112;第303-307页 * |
基于振动噪声信息融合的齿轮箱齿轮故障诊断研究;薛金亮;《中国优秀硕士学位论文全文数据库》;20131215;C029-22页 * |
芮国胜等. 一种基于基追踪压缩感知信号重构的改进算法.《电子测量技术》.2010, * |
Also Published As
Publication number | Publication date |
---|---|
CN109344751A (en) | 2019-02-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109344751B (en) | Reconstruction method of noise signal in vehicle | |
CN109993280B (en) | Underwater sound source positioning method based on deep learning | |
CN110728360B (en) | Micro-energy device energy identification method based on BP neural network | |
CN109800700B (en) | Underwater acoustic signal target classification and identification method based on deep learning | |
US20200285900A1 (en) | Power electronic circuit fault diagnosis method based on optimizing deep belief network | |
Weymaere et al. | On the initialization and optimization of multilayer perceptrons | |
CN107703486A (en) | A kind of auditory localization algorithm based on convolutional neural networks CNN | |
CN112733447B (en) | Underwater sound source positioning method and system based on domain adaptive network | |
CN111723701A (en) | Underwater target identification method | |
CN112331224A (en) | Lightweight time domain convolution network voice enhancement method and system | |
CN106656357B (en) | Power frequency communication channel state evaluation system and method | |
CN113378652B (en) | Disturbance classification method based on EWT-MPE-PSO-BP | |
CN113676266B (en) | Channel modeling method based on quantum generation countermeasure network | |
CN112949820A (en) | Cognitive anti-interference target detection method based on generation of countermeasure network | |
CN114595732A (en) | Radar radiation source sorting method based on depth clustering | |
Luo et al. | Performance evaluation of selective fixed-filter active noise control based on different convolutional neural networks | |
Dash et al. | Prediction of financial time series data using hybrid evolutionary Legendre neural network: Evolutionary LENN | |
Li et al. | Data augmentation method for underwater acoustic target recognition based on underwater acoustic channel modeling and transfer learning | |
CN112505628B (en) | Self-adaptive wave beam forming method based on graph convolution neural network | |
CN111859241B (en) | Unsupervised sound source orientation method based on sound transfer function learning | |
Chauhan et al. | Speech recognition and separation system using deep learning | |
CN114363262A (en) | Chaotic dynamic congestion prediction system and method under air-space-ground integrated network | |
CN111931412A (en) | Underwater target noise LOFAR spectrogram simulation method based on generative countermeasure network | |
Africa et al. | Development of an adaptive finite impulse response filter optimization algorithm using rough set theory | |
Zhao et al. | Underwater target perception algorithm based on pressure sequence generative adversarial network |
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 |