CN116738339A - Multi-classification deep learning recognition detection method for small-sample electric signals - Google Patents
Multi-classification deep learning recognition detection method for small-sample electric signals Download PDFInfo
- Publication number
- CN116738339A CN116738339A CN202310687001.7A CN202310687001A CN116738339A CN 116738339 A CN116738339 A CN 116738339A CN 202310687001 A CN202310687001 A CN 202310687001A CN 116738339 A CN116738339 A CN 116738339A
- Authority
- CN
- China
- Prior art keywords
- classification
- signal
- module
- data
- sample
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 11
- 238000013135 deep learning Methods 0.000 title claims description 11
- 238000000034 method Methods 0.000 claims abstract description 24
- 238000007637 random forest analysis Methods 0.000 claims abstract description 22
- 238000000605 extraction Methods 0.000 claims abstract description 20
- 238000003066 decision tree Methods 0.000 claims abstract description 13
- 238000005070 sampling Methods 0.000 claims abstract description 7
- 238000012549 training Methods 0.000 claims description 21
- 238000004422 calculation algorithm Methods 0.000 claims description 18
- 230000006870 function Effects 0.000 claims description 11
- 238000003745 diagnosis Methods 0.000 claims description 8
- 230000000007 visual effect Effects 0.000 claims description 8
- 210000002569 neuron Anatomy 0.000 claims description 7
- 238000013528 artificial neural network Methods 0.000 claims description 5
- 238000012952 Resampling Methods 0.000 claims description 2
- 230000004913 activation Effects 0.000 claims description 2
- 238000007621 cluster analysis Methods 0.000 claims description 2
- 238000002372 labelling Methods 0.000 claims description 2
- 238000007781 pre-processing Methods 0.000 claims description 2
- 230000008034 disappearance Effects 0.000 abstract description 3
- 230000036541 health Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 238000010801 machine learning Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000007635 classification algorithm Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/259—Fusion by voting
-
- 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
- G06N3/047—Probabilistic or stochastic networks
-
- 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)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Probability & Statistics with Applications (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a multi-classification non-invasive identification detection method for small-sample electric signals. Comprising the following steps: and the characteristic extraction module and the signal classification module. When the aircraft signal is subjected to feature extraction (201), aircraft signal samples containing labels are sent to a feature extraction module (202), deep features (211) of the extracted samples are input into a random forest module for classification, N samples are extracted by adopting a self-help sampling method (303), m features are extracted without being put back in a feature set (304), a decision tree is generated (305), a cycle number judgment module (306) is entered, the generated decision tree is formed into a random forest (307) for classification (308), voting (309), and a classification result (310) is output according to a majority vote principle. By the method, the problems of shallow feature extraction, gradient disappearance, single feature scale and the like are effectively solved, and the accuracy of multi-scale signal classification and identification of the aircraft is remarkably improved.
Description
Technical Field
The invention relates to a multi-classification deep learning identification detection method for small-sample electric signals.
Background
The identification and classification of the aircraft signals are the core of the aircraft health management strategy, and the current health state of the complex system is judged to effectively find out the occurrence cause and source of the fault and provide a series of suggestions or decisions related to maintenance and guarantee. The aircraft health management strategy is widely accepted and applied in the field of aerospace industry, provides great guarantee for the safety and reliability of the aircraft, and becomes a necessary trend of the development of the aerospace industry.
The fault diagnosis systems developed for the complex equipment at present have certain limitations, most of the diagnosis systems are still in an experimental verification stage, and a plurality of technologies are required to be further researched and perfected when the systems are applied to actual engineering, so that the important requirements of the spacecraft fault diagnosis technology are broken through, and better guarantee and maintenance are provided for the spacecraft. Modern equipment tends to be large, complex, automatic and intelligent, particularly large equipment with more and more complex functions, and traditional fault diagnosis technology shows great limitation and cannot meet the requirements of the current complex equipment on real-time performance and accuracy of fault diagnosis.
Disclosure of Invention
The invention aims to solve the problem of signal scale recognition limitation in the existing aircraft signal classification and recognition method, and provides a small-sample electric signal multi-classification deep learning recognition detection method. The problems of shallow feature extraction, gradient disappearance, single feature scale and the like are effectively solved, and the accuracy of aircraft signal classification and identification is remarkably improved.
Traditional machine learning algorithms in aircraft signal recognition and classification methods lack feature extraction capability for high-dimensional data, and classification accuracy, classification speed and other performance indexes in aircraft signal classification problems are limited.
The invention adopts a deep learning feature extraction and random forest small sample electric signal multi-classification non-invasive identification detection method, is used as a new leading edge machine learning method, is widely applied to the fields of computer vision, natural language processing and the like related to data science, and has stronger robustness and universality. The deep learning method is based on a deep neural network model, and features can be automatically extracted from complex aircraft signal samples. The invention solves the problem of high-dimension complex data processing difficulty of the traditional machine learning method by a deep message network method, and improves the feature extraction capability and generalization capability.
The invention relates to a deep message network feature extraction algorithm and a random forest signal classification algorithm.
Drawings
FIG. 1 shows an algorithm flow diagram for aircraft signal classification and identification according to one embodiment of the invention;
FIG. 2 shows a feature extraction flow chart of an aircraft signal according to one embodiment of the invention;
fig. 3 shows a flow chart for classifying aircraft signals based on random forests according to an embodiment of the invention.
Detailed Description
A flowchart of a small sample electrical signal multi-classification deep learning recognition detection method according to an embodiment of the present invention is shown in fig. 1, which includes:
when the aircraft signal is to perform fault discrimination (101), firstly, data of complex equipment in the aircraft are subjected to signal acquisition and transmitted to a sensor (102), and the original signal obtained by the sensor is subjected to signal preprocessing (103).
Then, judging the signal source (104), if the signal source is historical data, entering a historical data reading link (105), and performing signal cluster analysis (106) on the data, so as to effectively assist expert labeling work (107), and finally constructing the historical data and corresponding labels into an expert database (108); if the source is real-time data, a real-time data reading step (109) is entered.
The historical data and the real-time data are converted into two-dimensional data (110) and are input into a neural network together, the deep features obtained through extraction are sent into a random forest classification algorithm (112) through feature extraction (111) of a deep belief network, and aircraft signal classification results (114) are obtained through a voting module (113). Then convergence judgment (115) is carried out, if the loss function of the classification result reaches convergence in training, a final link is entered to output a real-time fault diagnosis result (116); if the loss function of the classification result does not converge in training, returning to a random forest algorithm (112), updating voting weight parameters in the random forest algorithm (113), and performing a signal classification (114) process again until the loss function converges. When the training is judged to be converged, the training is completed, and a real-time fault diagnosis result is output (116).
The deep neural network is a probability generation model, and is formed by establishing joint distribution between one sample data and category labels and combining a plurality of hidden variables. The deep belief network is formed by stacking and combining a series of limited boltzmann machines (Restrict Boltzmann Machine), wherein the limited boltzmann machines are graphical models formed by visible layers and hidden layers, and the two layers meet the boltzmann distribution.
Neurons between the visual layer and the hidden layer belong to full connection, but no connection exists between the neuron nodes between the same layers, the neurons are independent and irrelevant, and the two layers meet the Boltzmann distribution, namely:
p(h|v)＝p(h 1 |v)，p(h 2 |v)，…p(h n |v) (1)
according to the formula, the visual layer v and the hidden layer h are mutually reconstructed through p (h|v), and when the difference between the reconstructed vector of the hidden layer and an original input sample is smaller than a set minimum value through adjusting parameters, the minimum value of an energy function is found, and training is finished;
the purpose of training is to obtain three parameters, respectively network weights W mn Offset B n And offset C m ；
The energy equation for the joint configuration can be expressed as:
the joint probability distribution for a certain configuration can be determined by boltzmann distribution and energy function as shown in the following formula:
the joint probability density of the joint probability distribution is:
giving an electric signal sample set to meet independent identical distribution requirements: d= { v (1) ，v (2) ，…，v (N) The learning parameter θ= { W, b, c } is required. Since this probability satisfies a special gibbs probability distribution, there are:
thus, it is possible to obtain:
c i ＝c i +α(p(h i ＝1|v (1) )-p(h i ＝1|v (k) )) (10)
where α is the learning rate, k represents the kth cycle, and there are:
the deep belief network is a deep learning method formed by superposition of a plurality of limited boltzmann machines, the deep belief network method is adopted to perform feature learning on an expert database, a proper network layer number is selected, signals are reconstructed through layer-by-layer training, data dimension is reduced, and finally output of an electric signal sample after dimension reduction is obtained. The expert database is used for training the online classifier, so that the accuracy of online classification can be improved, and the time is shortened.
The feature extraction process of the deep belief network comprises the following steps:
when the aircraft signal is subjected to feature extraction (201), an aircraft signal sample containing a label is sent to a feature extraction module (202), greedy training according to layers is started to be carried out from top to bottom (203), a positive gradient is calculated (204), V '(205) of a visible unit is sampled, hidden activation h' (206) is resampled by a Gibbs method, a negative gradient is calculated (207), a neural network parameter is updated according to the negative gradient (208), whether the number of loops is reached or not is judged (209), if the number of loops is not reached, the visible unit module is resampled (205), if the number of loops is reached, an iteration step number judgment module (210) is started, and if the number of iteration steps is not reached, the training set is retrained to the training set return module (203); if the iteration step number reaches the iteration number, training (211) of the neuron convolution network is finished, and the deep sample features are extracted.
The Random Forest algorithm is a combined classifier formed by combining decision trees, has mature theory and application, is a mode identification algorithm with excellent performance, is mainly applied to classification identification problems, regression and the like of samples, particularly classification problems, has good performance on most data sets and has a very stable algorithm model. The random forest algorithm uses resampling technology (bootstrap) to randomly extract a plurality of samples from an original sample data set to form a new sample data set, repeatedly operates to form a plurality of sample data sets, constructs a decision tree for each resampled sample set as a training sample set, then combines the constructed decision tree to form a random forest, the decision tree is a multi-stage decision system, the classification process is sequentially carried out until finally acceptable classes are obtained, and the decision tree model formula is { h (X, theta } k ) K=1, 2, …, K }, where θ k K is the number of decision trees for independent co-distributed random variables. The classification and identification of the load electric signals of the spacecraft power supply system based on a random forest algorithm are mainly studied, the health state of the load electric signals of the spacecraft power supply system is judged, and in a given electric signal variable X, after each decision tree classifies the electric signals in sequence, voting is carried out to determine a final classification result. The random forest algorithm has mature theory and a lot of practical engineering applications, and the applications prove that the random forest algorithm has higher classification accuracy in the identification of complex data, has better tolerance to noise existing in the data, and does not have the occurrence of overfitting phenomenon.
The algorithm flow of the random forest is shown in fig. 3:
the random forest module starts operation (301), data are collected, features are set (302), N samples are extracted by adopting a self-help sampling method (303), m features are extracted in a feature set without replacement (304), m features are used as nodes to generate decision trees (305), a cycle number judgment module (306) is entered, if the cycle number is not reached, the self-help sampling method is returned to the sample extraction module (303), if the cycle number is reached, the generated decision trees are formed into random forests (307), the data are input for classification (308), the voting module (309) is entered, the statistics is carried out to obtain a vote, and a classification result (310) is output according to a majority vote principle.
The beneficial effects of the invention include:
compared with the existing method, the small-sample electric signal multi-classification non-invasive identification detection method provided by the invention has the following beneficial effects:
1. compared with the traditional method, the method converts one-dimensional data into two-dimensional data, adopts a deep belief network to extract the characteristics, and has higher training speed;
2. the method has good robustness and universality;
3. the method adopts a random forest algorithm, has high training speed, is easy to be made into a parallelization method, and further improves the operation efficiency;
4. the method effectively solves the problems of shallow feature extraction, gradient disappearance, single feature scale and the like, remarkably improves the accuracy of aircraft signal classification and identification, and makes excellent contribution to core fault identification of aircraft health management strategies.
Claims (3)
1. The multi-classification deep learning recognition detection method for the small-sample electric signals is characterized by comprising the following steps of:
a3 Determining (104) the source of the signal, wherein:
if the signal source is historical data, entering a historical data reading link (105), and performing signal cluster analysis (106) on the data, so as to effectively assist expert labeling work (107), and finally constructing the historical data and labels corresponding to the historical data into an expert database (108);
if the signal source is real-time data, a real-time data reading step (109) is entered,
a4 Converts the historical data and the real-time data into two-dimensional data (110) to be input into the neural network together,
a5 Feature extraction (111) using a deep belief network,
a6 The extracted deep features are sent into a random forest classification algorithm (112), a classification result (114) of the aircraft signal is obtained through a voting module (113),
a7 Performing convergence judgment (115), wherein:
if the loss function of the classification result reaches convergence in training, entering a final link to output a real-time fault diagnosis result (116);
if the loss function of the classification result does not converge in training, returning to a random forest classification algorithm (112), updating voting weight parameters in a voting module (113), performing signal classification (114) again until the loss function converges,
wherein:
the deep belief network is a probability generation model, which is formed by establishing joint distribution between one sample data and category labels and combining a plurality of hidden variables,
the deep belief network is formed by stacking and combining a series of limited boltzmann machines, wherein the limited boltzmann machines are graphical models consisting of visible layers and hidden layers,
the neurons between the visual layer and the hidden layer are fully connected, the nodes of the neurons between the same layer are not connected, are independent and irrelevant,
both the visual layer and the hidden layer satisfy the boltzmann distribution, namely:
p(h|v)＝p(h 1 |v)，p(h 2 |v)，…p(h n |v) (1)，
where h represents the state of the hidden layer, v represents the state of the visual layer, P represents the conditional probability,
namely: the visual layer v and the hidden layer h are mutually reconstructed through p (h|v), and when the difference between the reconstructed vector of the hidden layer and the original input sample is smaller than a set minimum value through adjusting parameters, the minimum value of the energy function is found, and training is finished;
the purpose of training is to obtain three parameters, respectively network weights W mn Offset B n And offset C m ；
The energy equation of the limited boltzmann machine is expressed as:
the joint probability distribution for a certain configuration can be determined by boltzmann distribution and energy function as shown in the following formula:
the joint probability density of the joint probability distribution is:
wherein a given set of electrical signal samples is made to satisfy independent co-distribution requirements:
D＝{v (1) ，v (2) ，…，v (N) the learning parameter θ= { W, b, c } is required, and since the joint probability satisfies a special gibbs probability distribution, there are:
thus, it is obtained:
c i ＝c i +α(p(h i ＝1|v (1) )-p(h i ＝1|v (k) )) (10)
where α is the learning rate, k represents the kth cycle, and there are:
the feature extraction operation of the deep belief network in the step A5 comprises the following steps:
a51 The aircraft signal samples containing the tags are fed into a feature extraction module (202),
a52 Performing a greedy training by layer from top to bottom (203),
a53 A positive gradient is calculated (204),
a54 A state V' (205) of the visual layer of the visual unit is sampled,
a55 Resampling the hidden activation h' (206) using the gibbs method,
a56 A negative gradient is calculated (207),
a57 Updating the neural network parameters according to the negative gradient (208),
a58 Judging whether the cycle number is reached (209), if the cycle number is not reached, returning to the step A54 to sample the visible unit module (205), if the cycle number is reached, entering an iteration step number judging module (210), if the iteration step number is not reached to the preset iteration number, returning to the step A52), and if the iteration step number is reached to the preset iteration number, ending training of the neuron convolutional network (211), and extracting the deep features of the sample.
2. The small sample electrical signal multi-classification deep learning identification detection method according to claim 1, wherein the method comprises the following steps:
the random forest classification algorithm comprises the following steps:
b2 Sampling N samples (303) by self-help sampling method,
b3 Extracting m features without replacement in the feature set (304),
b4 Generating a decision tree (305) with m features as nodes,
b5 If the number of the loops is not reached, returning to the step B2), sampling (303) by a self-help sampling method, if the number of the loops is reached, forming a generated decision tree into a random forest (307),
b6 The input data is classified (308),
b7 Voting (309),
b8 Statistics of tickets and output classification results according to majority ticket principles (310).
3. The small sample electrical signal multi-classification deep learning identification detection method of claim 1, further comprising performing, prior to step A3:
a1 Data signals of complex devices in the aircraft are acquired by means of the sensors and transmitted to the sensors (102),
a2 Signal preprocessing (103) is performed on the raw signals obtained by the sensor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310687001.7A CN116738339A (en) | 2023-06-09 | 2023-06-09 | Multi-classification deep learning recognition detection method for small-sample electric signals |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310687001.7A CN116738339A (en) | 2023-06-09 | 2023-06-09 | Multi-classification deep learning recognition detection method for small-sample electric signals |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116738339A true CN116738339A (en) | 2023-09-12 |
Family
ID=87916261
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310687001.7A Pending CN116738339A (en) | 2023-06-09 | 2023-06-09 | Multi-classification deep learning recognition detection method for small-sample electric signals |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116738339A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117079059A (en) * | 2023-10-13 | 2023-11-17 | 云南师范大学 | Tree species automatic classification method based on multi-source satellite image |
| CN117498400A (en) * | 2024-01-03 | 2024-02-02 | 长峡数字能源科技(湖北)有限公司 | Distributed photovoltaic and energy storage data processing method and system |
| CN117808703A (en) * | 2024-02-29 | 2024-04-02 | 南京航空航天大学 | Multi-scale large-scale component assembly gap point cloud filtering method |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107766883A (en) * | 2017-10-13 | 2018-03-06 | 华中师范大学 | A kind of optimization random forest classification method and system based on weighted decision tree |
| WO2019144521A1 (en) * | 2018-01-23 | 2019-08-01 | 杭州电子科技大学 | Deep learning-based malicious attack detection method in traffic cyber physical system |
| CN111414932A (en) * | 2020-01-07 | 2020-07-14 | 北京航空航天大学 | Classification identification and fault detection method for multi-scale signals of aircraft |
| WO2021022970A1 (en) * | 2019-08-05 | 2021-02-11 | 青岛理工大学 | Multi-layer random forest-based part recognition method and system |
| CN112926547A (en) * | 2021-04-13 | 2021-06-08 | 北京航空航天大学 | Small sample transfer learning method for classifying and identifying aircraft electric signals |
| CN113673346A (en) * | 2021-07-20 | 2021-11-19 | 中国矿业大学 | Motor vibration data processing and state recognition method based on multi-scale SE-Resnet |
| CN115277113A (en) * | 2022-07-06 | 2022-11-01 | 国网山西省电力公司信息通信分公司 | Power grid network intrusion event detection and identification method based on ensemble learning |
-
2023
- 2023-06-09 CN CN202310687001.7A patent/CN116738339A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107766883A (en) * | 2017-10-13 | 2018-03-06 | 华中师范大学 | A kind of optimization random forest classification method and system based on weighted decision tree |
| WO2019144521A1 (en) * | 2018-01-23 | 2019-08-01 | 杭州电子科技大学 | Deep learning-based malicious attack detection method in traffic cyber physical system |
| WO2021022970A1 (en) * | 2019-08-05 | 2021-02-11 | 青岛理工大学 | Multi-layer random forest-based part recognition method and system |
| CN111414932A (en) * | 2020-01-07 | 2020-07-14 | 北京航空航天大学 | Classification identification and fault detection method for multi-scale signals of aircraft |
| CN112926547A (en) * | 2021-04-13 | 2021-06-08 | 北京航空航天大学 | Small sample transfer learning method for classifying and identifying aircraft electric signals |
| CN113673346A (en) * | 2021-07-20 | 2021-11-19 | 中国矿业大学 | Motor vibration data processing and state recognition method based on multi-scale SE-Resnet |
| CN115277113A (en) * | 2022-07-06 | 2022-11-01 | 国网山西省电力公司信息通信分公司 | Power grid network intrusion event detection and identification method based on ensemble learning |
Non-Patent Citations (5)
| Title |
|---|
| KE LI等: "Hierarchical multi-class classification in multimodal spacecraft data using DNN and weighted support vector machine", NEUROCOMPUTING, 16 February 2017 (2017-02-16), pages 2 * |
| KE LI等: "Multi-label spacecraft electrical signal classification method based on DBN and random forest", PLOS ONE, 9 May 2017 (2017-05-09) * |
| 伏为峰;杨辉华;刘振丙;冯艳春;: "基于深度信念网络与随机森林的药品鉴别方法", 计算机仿真, no. 04, 15 April 2018 (2018-04-15) * |
| 曲文龙;宋晓明;曲嘉一;: "基于深度置信网络与梯度提升决策树的糖尿病检测方法", 吉林师范大学学报(自然科学版), no. 03, 10 August 2020 (2020-08-10) * |
| 武复宇;王雪;丁建伟;杜培军;谭琨;: "高光谱遥感影像多级联森林深度网络分类算法", 遥感学报, no. 04, 25 April 2020 (2020-04-25) * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117079059A (en) * | 2023-10-13 | 2023-11-17 | 云南师范大学 | Tree species automatic classification method based on multi-source satellite image |
| CN117079059B (en) * | 2023-10-13 | 2023-12-19 | 云南师范大学 | Tree species automatic classification method based on multi-source satellite image |
| CN117498400A (en) * | 2024-01-03 | 2024-02-02 | 长峡数字能源科技(湖北)有限公司 | Distributed photovoltaic and energy storage data processing method and system |
| CN117498400B (en) * | 2024-01-03 | 2024-03-15 | 长峡数字能源科技(湖北)有限公司 | Distributed photovoltaic and energy storage data processing method and system |
| CN117808703A (en) * | 2024-02-29 | 2024-04-02 | 南京航空航天大学 | Multi-scale large-scale component assembly gap point cloud filtering method |
| CN117808703B (en) * | 2024-02-29 | 2024-05-10 | 南京航空航天大学 | Multi-scale large-scale component assembly gap point cloud filtering method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116738339A (en) | Multi-classification deep learning recognition detection method for small-sample electric signals | |
| CN108712404B (en) | Internet of things intrusion detection method based on machine learning | |
| CN109800875A (en) | Chemical industry fault detection method based on particle group optimizing and noise reduction sparse coding machine | |
| CN106599797B (en) | A kind of infrared face recognition method based on local parallel neural network | |
| CN105447473B (en) | A kind of any attitude facial expression recognizing method based on PCANet-CNN | |
| CN111368885B (en) | Gas circuit fault diagnosis method for aircraft engine | |
| CN111832647A (en) | Abnormal flow detection system and method | |
| Bi et al. | An evolutionary deep learning approach using genetic programming with convolution operators for image classification | |
| CN112270345B (en) | Clustering algorithm based on self-supervision dictionary learning | |
| CN110334756A (en) | Power system monitor alarm event knows method for distinguishing, terminal installation, equipment and medium | |
| CN112732921B (en) | False user comment detection method and system | |
| CN108647707B (en) | Probabilistic neural network creation method, failure diagnosis method and apparatus, and storage medium | |
| CN111368926B (en) | Image screening method, device and computer readable storage medium | |
| CN113988215B (en) | Power distribution network metering cabinet state detection method and system | |
| CN110851654A (en) | Industrial equipment fault detection and classification method based on tensor data dimension reduction | |
| CN106874963A (en) | A kind of Fault Diagnosis Method for Distribution Networks and system based on big data technology | |
| CN112580588A (en) | Intelligent flutter signal identification method based on empirical mode decomposition | |
| CN116628605A (en) | Method and device for electricity stealing classification based on ResNet and DSCAttention mechanism | |
| CN112163474A (en) | Intelligent gearbox diagnosis method based on model fusion | |
| Yun et al. | Fault diagnosis of diesel generator set based on deep believe network | |
| CN116032790A (en) | Method, device and system for identifying, diagnosing and predicting massive data flow anomalies of dispatching automation system | |
| Luqin | A survey of facial expression recognition based on convolutional neural network | |
| CN112765148A (en) | Network intrusion detection method based on improved SVM multi-classification | |
| Wang et al. | An Interpretable Deep Bayesian Model for Facial Micro-Expression Recognition | |
| Wang et al. | Intelligent Diagnosis of Gearbox Based on Spatial Attention Convolutional Neural 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 |