CN113962256A - Intelligent fault diagnosis method and system for electromechanical actuator - Google Patents
Intelligent fault diagnosis method and system for electromechanical actuator Download PDFInfo
- Publication number
- CN113962256A CN113962256A CN202111155991.7A CN202111155991A CN113962256A CN 113962256 A CN113962256 A CN 113962256A CN 202111155991 A CN202111155991 A CN 202111155991A CN 113962256 A CN113962256 A CN 113962256A
- Authority
- CN
- China
- Prior art keywords
- electromechanical actuator
- scale
- fault diagnosis
- neural network
- training
- 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
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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Molecular Biology (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Signal Processing (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention discloses an intelligent fault diagnosis method and system for an electromechanical actuator, which comprises the following steps: collecting current signal samples of the electromechanical actuator under different health conditions; preprocessing the collected current signal sample to obtain a training sample; establishing a one-dimensional convolution neural network model based on multi-scale feature fusion; training a one-dimensional convolution neural network model based on multi-scale feature fusion by using a training sample; the intelligent fault diagnosis of the electromechanical actuator is carried out by utilizing the trained one-dimensional convolution neural network model based on multi-scale feature fusion, and the method and the system can carry out accurate fault diagnosis on the electromechanical actuator.
Description
Technical Field
The invention belongs to the field of fault diagnosis of electromechanical actuators, and relates to an intelligent fault diagnosis method and system for an electromechanical actuator.
Background
Electromechanical actuators (EMA), a form of electric actuator, are one of the key technologies in the development of modern aircraft to multi-electric and all-electric drives. Compared with the traditional hydraulic system, the hydraulic system has the advantages of small volume, light weight, simple structure, high safety, convenience in maintenance and the like, and is widely applied to various fields of aviation, navigation, industrial process control and the like. Since the health condition of the electromechanical actuator directly influences the flight state and safety of the airplane, the research on the fault diagnosis method of the electromechanical actuator has important significance for guaranteeing the safe operation and economic maintenance of the airplane, however, similar disclosures are not given in the prior art.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides an intelligent fault diagnosis method and system for an electromechanical actuator, and the method and system can be used for accurately diagnosing faults of the electromechanical actuator.
In order to achieve the above object, the intelligent fault diagnosis method for an electromechanical actuator according to the present invention comprises:
collecting current signal samples of the electromechanical actuator under different health conditions;
preprocessing the collected current signal sample to obtain a training sample;
establishing a one-dimensional convolution neural network model based on multi-scale feature fusion;
training a one-dimensional convolution neural network model based on multi-scale feature fusion by using a training sample;
and carrying out intelligent fault diagnosis on the electromechanical actuator by using the trained one-dimensional convolution neural network model based on multi-scale feature fusion.
At a sampling frequency fsCollecting current signals { x [ n ] of electromechanical actuator under different health conditions]i}。
Preprocessing a collected current signal sample to obtain a training sample by the specific process:
and carrying out normalization processing on the collected current signal samples, and amplifying the number of the samples in an overlapping sampling mode to obtain a training sample.
The one-dimensional convolution neural network model based on multi-scale feature fusion comprises a multi-scale transformation module, a feature learning and fusion module and a fault classification module.
The multi-scale transformation module respectively carries out time-frequency scale transformation on the input samples through 3 different scales according to the formula (3);
y(k,s)＝φ(k,s)(x[l]′) (3)
wherein, y(k,s)As a result of scaling the input samples, phi(k,s)(g) For scale transformation operation, k and s are scale factors, k is used for controlling frequency scale, and s is used for controlling time scale.
The feature learning and fusion module comprises 1 trunk part and 3 branch parts, wherein the trunk part comprises 3 groups of stacked convolution layers C1-C3, pooling layers P1-P3 and 1 full-connection layer FC1, and each support rod part comprises 1 convolution layer and 1 global average pooling layer GAP.
The input of the branch part is a characteristic diagram T output by the pooling layers P1-P31、T2And T3Output feature map G1、G2And G3；
And splicing and fusing the feature maps output by the trunk part and the feature maps output by each branch part by using a Concat method to obtain a high-low layer feature fusion result Q under a preset scale, and finally fusing R with the feature fusion results under 3 scales.
The intelligent fault diagnosis system for the electromechanical actuator comprises:
the acquisition module is used for acquiring current signal samples of the electromechanical actuator under different health conditions;
the preprocessing module is used for preprocessing the collected current signal sample to obtain a training sample;
the building module is used for building a one-dimensional convolution neural network model based on multi-scale feature fusion;
the training module is used for training the one-dimensional convolution neural network model based on multi-scale feature fusion by utilizing the training samples;
and the diagnosis module is used for performing intelligent fault diagnosis on the electromechanical actuator by using the trained one-dimensional convolution neural network model based on multi-scale feature fusion.
The invention has the following beneficial effects:
when the intelligent fault diagnosis method and the intelligent fault diagnosis system for the electromechanical actuator are specifically operated, multi-scale transformation is introduced into a traditional convolutional neural network to establish a multi-scale feature fusion-based one-dimensional convolutional neural network model, so that the network model obtains features of different receptive fields at the same level to extract time and frequency features hidden in signals at different scales, and the trained multi-scale feature fusion-based one-dimensional convolutional neural network model is used for fault diagnosis, so that the sensitivity to various uncertain sources can be reduced, the intelligent fault diagnosis method and the intelligent fault diagnosis system for the electromechanical actuator have good robustness and high diagnosis accuracy, and the recognition capability of the model to different health conditions of the electromechanical actuator is greatly improved.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a frame diagram of a one-dimensional convolutional neural network model based on multi-scale feature fusion;
FIG. 3 is a block diagram of trunk and limb modules;
FIG. 4 is a block diagram of an electromechanical actuator;
FIG. 5 is a diagram of the results of model training and test classification;
FIG. 6 is a graph of the results of 10 tests of the present invention and various methods;
FIG. 7 is a graph showing the results of 3 different SNR tests according to the present invention.
Detailed Description
In order to make the technical solutions of the present invention better understood, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, not all of the embodiments, and are not intended to limit the scope of the present disclosure. Moreover, in the following description, descriptions of well-known structures and techniques are omitted so as to not unnecessarily obscure the concepts of the present disclosure. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
There is shown in the drawings a schematic block diagram of a disclosed embodiment in accordance with the invention. The figures are not drawn to scale, wherein certain details are exaggerated and possibly omitted for clarity of presentation. The shapes of various regions, layers and their relative sizes and positional relationships shown in the drawings are merely exemplary, and deviations may occur in practice due to manufacturing tolerances or technical limitations, and a person skilled in the art may additionally design regions/layers having different shapes, sizes, relative positions, according to actual needs.
Referring to fig. 1, the intelligent fault diagnosis method for the electromechanical actuator according to the invention comprises the following steps:
the method comprises the following steps of preprocessing collected current signals to form training samples, constructing a multi-scale feature fusion one-dimensional convolution neural network based on multi-scale feature fusion learning, training the multi-scale feature fusion one-dimensional convolution neural network by using the training samples, and diagnosing the fault of an electromechanical actuator by using the trained multi-scale feature fusion one-dimensional convolution neural network, and specifically comprises the following steps:
1) at a sampling frequency fsCollecting current signals x [ n ] of electromechanical actuator under different health conditions]iAnd form the original data set { x [ n ]]iThen sample each current signal x [ n ]]iCarrying out normalization treatment according to the formula (1);
wherein, x [ n ]]iminIs the minimum of the data points contained in the sequence of fragments, x [ n ]]imaxIs the maximum of the data points contained in the sequence of fragments;
2) according to the formula (2), dividing the normalized data in an overlapped sampling mode to amplify the current signal sample;
where m is the current signal sample x [ n ]]’iThe maximum divisible number of (l) is each divisionSegment signal x [ l ]]’jLength of (d), λ is the overlap ratio, x [ l ]]’jFor the j-th segmented signal, j belongs to [1, m ∈]；
3) Constructing a one-dimensional convolution neural network model based on multi-scale feature fusion, and performing layer-by-layer training and testing on the one-dimensional convolution neural network model based on multi-scale feature fusion by using training samples based on a supervised learning method;
4) and performing fault diagnosis by using the trained one-dimensional convolution neural network model based on multi-scale feature fusion.
The framework based on the multi-scale feature fusion one-dimensional convolution neural network model is shown in fig. 2 and comprises the following steps:
a) multi-scale transformation module
The time-frequency scale transformation is respectively carried out on the input samples through 3 different scales, and the transformation operation is as shown in formula (3):
y(k,s)＝φ(k,s)(x[l]′) (3)
wherein, y(k,s)As a result of scaling the input samples, phi(k,s)(g) For scale transformation operation, k and s are scale factors, k is used for controlling frequency scale, and s is used for controlling time scale.
b) Feature learning and fusion module
The feature learning and fusion module comprises 1 trunk part and 3 branch parts, as shown in fig. 3, wherein the trunk part comprises 3 stacked convolutional layers C1-C3, pooling layers P1-P3 and 1 full-link layer FC1, and the input of the result y is converted by different scales(k,s)The output characteristic diagram is T, each branch part consists of 1 convolutional layer and 1 global average pooling layer (GAP), and the input of the branch part is the characteristic diagram T output by the pooling layers (P1-P3)1、T2And T3The output characteristic graphs are respectively G1、G2And G3Then, splicing and fusing the feature graph output by the trunk part and the feature graph output by each branch part by using a Concat method to obtain a result Q of high-low layer feature fusion under a certain scale, finally fusing R of the feature fusion results under 3 scales, and performing mathematical descriptionThe method comprises the following steps:
Q＝F(T,G1,G2,G3) (4)
R＝F(Q1,Q2,Q3) (5)
c) fault classification module
Flattening the final feature fusion result, randomly discarding the flattened result by using a Dropout technology with a preset probability p, then performing fault classification through a Softmax layer, and outputting a fault diagnosis result.
The training process based on the multi-scale feature fusion one-dimensional convolution neural network is shown in figure 1 and comprises the following steps:
11) setting hyper-parameters in the model, wherein the hyper-parameters comprise scale transformation factors k and s, the number of convolution layers, the number and the size of each layer of convolution kernel, the size m of each iteration batch, a training algebra e and a learning rate eta;
12) initializing training parameters in the model, wherein the training parameters comprise a weight parameter W and a bias parameter b of the model;
13) performing iterative training on the model by using a training sample, calculating the error between the output value and the true value of the model by network forward propagation, and updating the weight by using error backward propagation and an Adam optimization algorithm;
14) judging whether the error between the output value and the true value meets the precision requirement or whether the training process reaches the set iteration times, if so, saving the parameters of the model, otherwise, jumping to the step 13);
15) inputting the test sample into the trained model, and outputting a fault diagnosis result;
16) and judging whether the diagnosis result meets the performance requirement of the fault diagnosis of the electromechanical actuator, if so, saving the fault diagnosis model, and otherwise, jumping to the step 11).
Example one
In the embodiment, a direct-drive type electromechanical actuator is used as a support, and the specific structure is shown in fig. 4. In view of the components of the electromechanical actuator, the faults that may occur during operation include four types, namely motor faults, electrical faults, mechanical faults and sensor faults. The method comprehensively considers three factors of the frequency of the fault occurrence, the degree of influence and the similarity of the fault expression of the electromechanical actuator, and selects 4 faults of the turn-to-turn short circuit of the motor winding, the open circuit of the power tube, the motor stalling and the sensor deviation for research, which are shown in table 1 specifically.
TABLE 1
The training and verification classification results of the present invention are shown in FIG. 5. 6000 training samples are divided into two parts, wherein 4800 samples are used for training the model, 1200 samples are used for verifying the quality of the model, the training algebra is 20, the batch size is 60, and an Adam optimization algorithm is adopted. It can be seen from fig. 5 that the loss function value is continuously decreased with the increase of the training algebra, when the training algebra of the model is about 9, the accuracy of the sample for verifying the model has reached 99.5%, it can be seen that this training process of the model is very successful, and it can be preliminarily determined that the model should have good performance.
The test results of the present invention, WPD-SVM (conventional method), 1DCNN, 2DCNN, LSTM, and WDCNN are shown in fig. 6. Wherein, three comparison methods of 1DCNN, 2DCNN and WDCNN can be referred to three documents in turn: 1. rolling bearing self-adaptive fault diagnosis algorithm [ J ] based on one-dimensional convolutional neural network, instrument and meter report, 2018,39(07) 134-; 2. xiaoxiong, Wang Jianxiang, Zhang Yongjun, etc. A two-dimensional convolutional neural network optimization method for bearing fault diagnosis [ J ] China Motor engineering report, 2019,39(15):4558 + 4568; zhang Wei, Peng Gaoliang, Li Chuanhao, et al. A new deep learning model for fault diagnosis with a good adaptation availability on raw simulation signals [ J ]. Sensors,2017,17(2): 425.
As can be seen in fig. 6: the performance of the invention is best, while the performance of the WPD-SVM is worst, and the performance of the other methods is in the middle. The reasons why the results of the conventional diagnostic methods are not ideal are as follows 3: (1) the feature extraction and classification are separately designed and carried out, but not completely unified, so that the feature extraction and classification cannot be simultaneously optimized, and the upper bound of the final diagnosis performance is limited; (2) the features can not be extracted in a self-adaptive manner, usually depend on manual design, require certain signal processing technology and diagnosis professional knowledge, and are time-consuming and labor-consuming; (3) most of the existing methods are specific fields, cannot be updated on line along with the change of application equipment or fields, namely, the universality is poor, and the method cannot be well popularized to new diagnosis fields, so the effect is not ideal. The inputs of the three fault diagnosis models based on LSTM, 1DCNN and WDCNN are all one-dimensional signals, but the average accuracy of the LSTM model is lower than that of the latter two models, and the reason is that the LSTM structure is more suitable for language and text processing, so the accuracy of state identification of the electromechanical actuator is relatively low; the input of the 2DCNN model is a two-dimensional image, but the current output by the electromechanical actuator is a one-dimensional signal, so that the one-dimensional signal needs to be converted into the two-dimensional image, and a certain information deviation may exist in the conversion process, so that the identification accuracy of the 2DCNN model is also low.
However, the network model designed by the invention fuses the low-level detail information and the high-level abstract features of the samples under different scales through a plurality of fusion structures of different scales and high-level and low-level features, so that the diversity and complementarity of the features can be effectively enriched, the identifiability of the fusion features is enhanced, and the identification capability of the model on different health conditions of the electromechanical actuator is greatly improved.
Fig. 7 is a graph of test results of the present invention under 3 different signal-to-noise ratios (SNRs), and it can be seen from fig. 7 that the present invention has a diagnostic performance exceeding 90% under 3 different SNRs, and the diagnostic performance increases with the increase of the SNR, for example, when the SNR is 15, the fault identification accuracy approaches 100%, which indicates that the present invention has good robustness to noise.
Claims (8)
1. An intelligent fault diagnosis method for an electromechanical actuator is characterized by comprising the following steps:
collecting current signal samples of the electromechanical actuator under different health conditions;
preprocessing the collected current signal sample to obtain a training sample;
establishing a one-dimensional convolution neural network model based on multi-scale feature fusion;
training a one-dimensional convolution neural network model based on multi-scale feature fusion by using a training sample;
and carrying out intelligent fault diagnosis on the electromechanical actuator by using the trained one-dimensional convolution neural network model based on multi-scale feature fusion.
2. The intelligent fault diagnosis method for an electromechanical actuator according to claim 1, characterized in that the sampling frequency f is setsCollecting current signals { x [ n ] of electromechanical actuator under different health conditions]i}。
3. The intelligent fault diagnosis method for the electromechanical actuator as claimed in claim 1, wherein the specific process of preprocessing the collected current signal samples to obtain training samples comprises:
and carrying out normalization processing on the collected current signal samples, and amplifying the number of the samples in an overlapping sampling mode to obtain a training sample.
4. The intelligent fault diagnosis method for the electromechanical actuator according to claim 1, wherein the multi-scale feature fusion one-dimensional convolution neural network model comprises a multi-scale transformation module, a feature learning and fusion module and a fault classification module.
5. The intelligent fault diagnosis method for the electromechanical actuator according to claim 4, wherein the multi-scale transformation module performs scale transformation on the input samples through 3 different scales according to equation (3);
y(k,s)＝φ(k,s)(x[l]′) (3)
wherein, y(k,s)As a result of scaling the input samples, phi(k,s)(g) For scale transformation operation, k and s are scale factors, k is used for controlling frequency scale, and s is used for controlling time scale.
6. The intelligent fault diagnosis method for the electromechanical actuator as claimed in claim 4, wherein the feature learning and fusion module comprises 1 trunk portion and 3 branch portions, wherein the trunk portion comprises 3 sets of stacked convolutional layers C1-C3, pooling layers P1-P3 and 1 full-connection layer FC1, and each branch portion comprises 1 convolutional layer and 1 global average pooling layer GAP.
7. The intelligent fault diagnosis method for the electromechanical actuator according to claim 6, wherein the inputs of the branch parts are the characteristic diagrams T output by the pooling layers P1-P3 respectively1、T2And T3Output feature map G1、G2And G3；
And splicing and fusing the feature maps output by the trunk part and the feature maps output by each branch part by using a Concat method to obtain a high-low layer feature fusion result Q under a preset scale, and finally fusing R with the feature fusion results under 3 scales.
8. An intelligent fault diagnosis system for an electromechanical actuator, comprising:
the acquisition module is used for acquiring current signal samples of the electromechanical actuator under different health conditions;
the preprocessing module is used for preprocessing the collected current signal sample to obtain a training sample;
the building module is used for building a one-dimensional convolution neural network model based on multi-scale feature fusion;
the training module is used for training the one-dimensional convolution neural network model based on multi-scale feature fusion by utilizing the training samples;
and the diagnosis module is used for performing intelligent fault diagnosis on the electromechanical actuator by using the trained one-dimensional convolution neural network model based on multi-scale feature fusion.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111155991.7A CN113962256A (en) | 2021-09-29 | 2021-09-29 | Intelligent fault diagnosis method and system for electromechanical actuator |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111155991.7A CN113962256A (en) | 2021-09-29 | 2021-09-29 | Intelligent fault diagnosis method and system for electromechanical actuator |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113962256A true CN113962256A (en) | 2022-01-21 |
Family
ID=79462813
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111155991.7A Pending CN113962256A (en) | 2021-09-29 | 2021-09-29 | Intelligent fault diagnosis method and system for electromechanical actuator |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113962256A (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109299705A (en) * | 2018-10-24 | 2019-02-01 | 电子科技大学 | Rotary machinery fault diagnosis method based on one-dimensional depth residual error convolutional neural networks |
| CN110596506A (en) * | 2019-10-30 | 2019-12-20 | 福州大学 | Converter fault diagnosis method based on time convolution network |
| CN111340238A (en) * | 2020-03-12 | 2020-06-26 | 中南大学 | Fault diagnosis method, device, equipment and storage medium of industrial system |
| WO2020244134A1 (en) * | 2019-06-05 | 2020-12-10 | 华南理工大学 | Multi-task feature sharing neural network-based intelligent fault diagnosis method |
| CN112254964A (en) * | 2020-09-03 | 2021-01-22 | 太原理工大学 | Rolling bearing fault diagnosis method based on rapid multi-scale convolution neural network |
| CN112949823A (en) * | 2021-02-05 | 2021-06-11 | 杭州电子科技大学 | Industrial process performance diagnosis method based on one-dimensional multi-scale depth convolution neural network |
-
2021
- 2021-09-29 CN CN202111155991.7A patent/CN113962256A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109299705A (en) * | 2018-10-24 | 2019-02-01 | 电子科技大学 | Rotary machinery fault diagnosis method based on one-dimensional depth residual error convolutional neural networks |
| WO2020244134A1 (en) * | 2019-06-05 | 2020-12-10 | 华南理工大学 | Multi-task feature sharing neural network-based intelligent fault diagnosis method |
| CN110596506A (en) * | 2019-10-30 | 2019-12-20 | 福州大学 | Converter fault diagnosis method based on time convolution network |
| CN111340238A (en) * | 2020-03-12 | 2020-06-26 | 中南大学 | Fault diagnosis method, device, equipment and storage medium of industrial system |
| CN112254964A (en) * | 2020-09-03 | 2021-01-22 | 太原理工大学 | Rolling bearing fault diagnosis method based on rapid multi-scale convolution neural network |
| CN112949823A (en) * | 2021-02-05 | 2021-06-11 | 杭州电子科技大学 | Industrial process performance diagnosis method based on one-dimensional multi-scale depth convolution neural network |
Non-Patent Citations (1)
| Title |
|---|
| 孙曙光;李勤;杜太行;崔景瑞;王景芹;: "基于一维卷积神经网络的低压万能式断路器附件故障诊断", 电工技术学报, no. 12 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111340238B (en) | Fault diagnosis method, device, equipment and storage medium of industrial system | |
| CN109765053B (en) | Rolling bearing fault diagnosis method using convolutional neural network and kurtosis index | |
| CN111914883B (en) | Spindle bearing state evaluation method and device based on deep fusion network | |
| CN114092832B (en) | High-resolution remote sensing image classification method based on parallel hybrid convolutional network | |
| CN111650453B (en) | Power equipment diagnosis method and system based on windowing characteristic Hilbert imaging | |
| CN113567131B (en) | Bearing fault diagnosis method based on S transformation and miniature convolutional neural network model | |
| CN113203566B (en) | Motor bearing fault diagnosis method based on one-dimensional data enhancement and CNN | |
| CN111368885A (en) | Aero-engine gas circuit fault diagnosis method based on deep learning and information fusion | |
| CN110991471B (en) | Fault diagnosis method for high-speed train traction system | |
| CN110954326A (en) | Rolling bearing online fault diagnosis method capable of automatically learning feature expression | |
| CN112926728B (en) | Small sample turn-to-turn short circuit fault diagnosis method for permanent magnet synchronous motor | |
| CN112859898A (en) | Aircraft trajectory prediction method based on two-channel bidirectional neural network | |
| CN112541510A (en) | Intelligent fault diagnosis method based on multi-channel time series data | |
| CN113343760A (en) | Human behavior recognition method based on multi-scale characteristic neural network | |
| CN112596016A (en) | Transformer fault diagnosis method based on integration of multiple one-dimensional convolutional neural networks | |
| CN111368648B (en) | Individual identification method and device for radar radiation source, electronic equipment and storage medium thereof | |
| CN113255591A (en) | Bearing fault diagnosis method based on random forest and fusion characteristics | |
| CN114818579A (en) | Analog circuit fault diagnosis method based on one-dimensional convolution long-short term memory network | |
| CN117809164A (en) | Substation equipment fault detection method and system based on multi-mode fusion | |
| CN115221973A (en) | Aviation bearing fault diagnosis method based on enhanced weighted heterogeneous ensemble learning | |
| CN116578833A (en) | IGBT module aging fault diagnosis system based on optimized random forest model | |
| CN110956617A (en) | Brain nuclear magnetic resonance abnormal image visualization method based on circulation attention model | |
| Zhao et al. | Bearing fault diagnosis based on mel frequency cepstrum coefficient and deformable space-frequency attention network | |
| CN114357372A (en) | Aircraft fault diagnosis model generation method based on multi-sensor data driving | |
| CN111783335B (en) | Transfer learning-based few-sample structure frequency response dynamic model correction method |
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 |