CN115081318B - Artificial bata ray based on antagonistic neural network method for predicting experimental data of aircraft - Google Patents
Artificial bata ray based on antagonistic neural network method for predicting experimental data of aircraft Download PDFInfo
- Publication number
- CN115081318B CN115081318B CN202210652320.XA CN202210652320A CN115081318B CN 115081318 B CN115081318 B CN 115081318B CN 202210652320 A CN202210652320 A CN 202210652320A CN 115081318 B CN115081318 B CN 115081318B
- Authority
- CN
- China
- Prior art keywords
- neural network
- data
- ray
- aircraft
- bata
- 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
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 57
- 230000003042 antagnostic effect Effects 0.000 title claims abstract description 28
- 238000000034 method Methods 0.000 title claims abstract description 28
- 241001175904 Labeo bata Species 0.000 title claims abstract description 16
- 238000012549 training Methods 0.000 claims abstract description 26
- 238000002474 experimental method Methods 0.000 claims abstract description 20
- 230000006870 function Effects 0.000 claims description 16
- 230000004913 activation Effects 0.000 claims description 12
- 238000012360 testing method Methods 0.000 claims description 10
- 238000012795 verification Methods 0.000 claims description 9
- 230000017105 transposition Effects 0.000 claims description 6
- 230000000737 periodic effect Effects 0.000 claims description 4
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 3
- 239000011664 nicotinic acid Substances 0.000 claims description 2
- 210000000006 pectoral fin Anatomy 0.000 claims description 2
- 238000000638 solvent extraction Methods 0.000 claims description 2
- 238000013135 deep learning Methods 0.000 abstract description 6
- 238000011160 research Methods 0.000 abstract description 4
- 230000000694 effects Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 2
- 230000036544 posture Effects 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/06—Multi-objective optimisation, e.g. Pareto optimisation using simulated annealing [SA], ant colony algorithms or genetic algorithms [GA]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/02—Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA]
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
According to the experimental data prediction method of the simulated bata ray aircraft based on the antagonistic neural network, the limited hydrodynamic parameters of the simulated bata ray aircraft corresponding to different motion parameters obtained through experiments are combined, and the internal interpolation prediction is carried out on the original data through a deep learning method, so that the experimental result can be effectively expanded to support the subsequent researches such as the accurate control of the simulated bata ray aircraft, and meanwhile, the labor and time cost required by the experiments are greatly reduced. Secondly, the countermeasure neural network can automatically generate false data samples in the training process, so that the model has higher prediction precision on a small data set of an experiment.
Description
Technical Field
The invention belongs to the field of deep learning, and particularly relates to a experimental data prediction method of a ray-simulated ray-a-tree aircraft based on an antagonistic neural network.
Background
The 21 st century is the century of the ocean, and the ocean's national strategy has put higher demands on the performance of underwater vehicles. The traditional propeller propelling underwater vehicle has the characteristics of large running noise and weak maneuverability, and researches find that the ray organisms such as the ray in water have low noise, high efficiency and high maneuverability on fin flapping movement, so researchers start to invest into the research of the ray-like vehicle. The accurate control of the simulated bated ray aircraft requires hydrodynamic experiments on various flapping postures of the aircraft, and the control of the motion postures usually involves a plurality of parameter combinations such as left-right amplitude frequency, phase and the like. Because of the time and labor cost limitations of the experiment, it is difficult to perform a fine combination experiment on all parameter variations. It is therefore necessary to fit the result of the non-experimental interpolated input data with limited experimental data.
The deep learning is an algorithm for performing feature recognition, rule exploration and data regression based on data samples, and is widely used in the aspect of interpolation prediction of data. The final goal of deep learning is to enable a machine to analyze learning capabilities like a person, recognize text, image, sound, etc., and migrate learned rules to other models. The deep learning algorithms commonly used at present are: fully connected neural networks, convolutional neural networks, recurrent neural networks, and the like. The above are all single neural networks, and the antagonistic neural network is actually a combination of two networks, one of which generates analog data, and the other of which judges whether the generated data is authentic.
The specific algorithm ideas against the neural network are: the network generating the simulation data continuously optimizes itself, so that the generated tensor can confuse the judging network, and the judging network continuously optimizes so that the judgment is more accurate. The neural network has the capability of autonomous learning and searching for an optimal solution at a high speed. At the same time, the generator of the antagonistic neural network can generate a large amount of false data during the training process, and thus has great advantages in terms of prediction of small data sets. The data set obtained by the experiment is usually small in data quantity, the single neural network is poor in prediction effect, sometimes even the result diverges, and the original data set cannot be fitted. Therefore, the experimental data of the simulated bata ray aircraft are predicted by the antagonistic neural network, so that a more accurate prediction result is obtained.
Disclosure of Invention
Technical problem to be solved
In order to avoid the defects of the prior art, the invention provides an experimental data prediction method of an artificial bata ray aircraft based on an antagonistic neural network, which predicts experimental results corresponding to input motion parameters through the neural network and obtains fine interpolation test results through a small amount of experiments so as to save labor, expenses and time cost in the experimental process. Meanwhile, the self-learning characteristic of the antagonistic neural network is combined, and the reliability of the prediction result is effectively improved.
Technical proposal
A simulated ray aircraft experimental data prediction method based on an antagonistic neural network is characterized by comprising the following steps:
s1: data acquisition
The six-dimensional force time sequence data of the prototype periodic variation corresponding to different input motion parameter labels is measured through a hydrodynamic force experiment of the bionic aircraft; the motion parameter label comprises a left amplitude, a right amplitude, a frequency and a phase difference;
s2: data set partitioning
Setting different input motion parameter labels as input x of the neural network, and setting corresponding six-dimensional force time series data as output y; and all data were taken at 8:1:1, dividing a training set, a verification set and a test set in proportion;
s3: training neural networks
Setting up an antagonistic neural network, taking x in a training set and a verification set as input variables, and taking y as output variables to be input into the neural network, wherein parameters and super parameters of the training network;
s4: predicting experimental results
Inputting variable labels x in the test set into a trained neural network, and predicting corresponding sequence data y;
s5: verifying the experimental results
And comparing the predicted result with the experimental result, calculating the relative percentage error of the predicted result, and comparing the relative percentage error with the predicted result of the original data set.
The invention further adopts the technical scheme that: s1, experimental data are obtained through a hydrodynamic experiment of the simulated ray-light aircraft, and the experiment is carried out by setting different motion parameter combinations on the simulated ray-light aircraft, and measuring the sequence data of the changes of stress and moment of the simulated ray-light aircraft along the time by six-dimensional force sensors in the simulated ray-light aircraft along the spanwise direction, the chordwise direction and the plumb direction.
The invention further adopts the technical scheme that: in the step S1, the data of the pectoral fin flapping movement period of the 10 simulated ray aircraft are obtained through one experiment, and the obtained six-dimensional force time series data have the same strong periodicity.
The invention further adopts the technical scheme that: all data in S2 are as per 8:1:1, before dividing the training set, the verification set and the test set in proportion, the original data set needs to be randomly disturbed according to the same random seed.
The invention further adopts the technical scheme that: s3, the countermeasure neural network consists of two neural networks of a generator and a discriminator, wherein a generator model consists of four one-dimensional transposition convolutional layers and a full-connection layer, the activation functions of the transposition convolutional layers are Relu, and the activation functions of the full-connection layer are tanh so as to keep nonlinear characteristics of the model; the arbiter model consists of three one-dimensional convolution layers and one full-connection layer, wherein the activation functions of the convolution layers are Relu, and the activation functions of the full-connection layer are Sigmoid.
The invention further adopts the technical scheme that: optimizing a weight parameter set of a generator model of the antagonistic neural network by adopting an Adam optimizer, and setting the initial learning rate to be 0.001 and the gradient attenuation rate to be 0.9; the loss function of the generator is set as the mean square error and the loss function of the arbiter is set as the two-to-cross entropy.
The invention further adopts the technical scheme that: the antagonistic neural network dataset was trained in a batch of every 32 divisions.
The invention further adopts the technical scheme that: and S5, the prediction result is visualized through a matplotlib module in the neural network program, and a predicted line graph is output for comparison with the line graph of the original data every 20 training rounds, so that the state of model training is conveniently determined, and parameters are conveniently adjusted at any time.
Advantageous effects
According to the experimental data prediction method of the simulated bata ray aircraft based on the antagonistic neural network, the limited hydrodynamic parameters of the simulated bata ray aircraft corresponding to different motion parameters obtained through experiments are combined, and the internal interpolation prediction is carried out on the original data through a deep learning method, so that the experimental result can be effectively expanded to support the subsequent researches such as the accurate control of the simulated bata ray aircraft, and meanwhile, the labor and time cost required by the experiments are greatly reduced. Secondly, the countermeasure neural network can automatically generate false data samples in the training process, so that the model has higher prediction precision on a small data set of an experiment.
Drawings
The drawings are only for purposes of illustrating particular embodiments and are not to be construed as limiting the invention, like reference numerals being used to refer to like parts throughout the several views.
FIG. 1 is a schematic diagram of an experimental apparatus for obtaining data according to the present invention;
FIG. 2 is a schematic representation of periodic hydrodynamic data measured experimentally in accordance with the present invention;
FIG. 3 is a block diagram of an antagonistic neural network generator and arbiter according to the present invention;
FIG. 4 is a flow chart of the prediction of the time series of the hydrodynamic test results of the simulated ray of the present invention;
FIG. 5 shows the prediction effect of the countermeasure neural network (left: raw experimental data; middle: neural network prediction result; right: prediction result error) according to the present invention.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention. In addition, technical features of the embodiments of the present invention described below may be combined with each other as long as they do not collide with each other.
The method for acquiring the original data and dividing the data set of the prediction method is described with reference to fig. 1 and 2. A method for predicting experimental data of a simulated ray of a ray aircraft based on an antagonistic neural network comprises the steps of obtaining an original data set through experimental means. Firstly, a six-dimensional force sensor is carried on the mass center position of the simulated ray of the aircraft, the six-dimensional force sensor is fixedly connected with a bracket of the flume through a threaded hole of the force sensor, and the aircraft is placed in the flume. And secondly, connecting the aircraft, the sensor signal receiver, the control module and the computer through a data line. The motion parameter instruction is input to the simulated bated ray aircraft through a computer, and periodic hydrodynamic data, including time series data of spanwise direction, chordwise direction and vertical direction stress and moment, are measured by a six-dimensional force sensor. One set of experimental data typically includes about 10 full cycles, each set of time series data is taken from the beginning of the cycle, one full cycle data at a time, along with the corresponding input motion parameters, as data to be fed into the neural network for training. The second step requires dividing the raw data into training data sets. Different input motion parameter labels (left amplitude, right amplitude, frequency, phase difference) are set as input tensors x of the neural network, and corresponding single-period six-dimensional force time series data are set as output tensors y. And after randomly scrambling all the x and y corresponding to each other, the method comprises the following steps of: 1:1, and sending the training set, the verification set and the test set into the neural network for every 32 data in a batch.
The specific operation process of training the neural network model and predicting the result is described with reference to fig. 3 and 4: firstly, constructing an antagonistic neural network, wherein a generator model of the network consists of four one-dimensional transposition convolutional layers and a full-connection layer, the activation functions of the transposition convolutional layers are Relu, and the activation functions of the full-connection layer are tanh; the network discriminant model consists of three one-dimensional convolution layers and a full-connection layer, wherein the activation functions of the convolution layers are Relu, and the activation functions of the full-connection layer are Sigmoid. After the network is built, the data of the training set and the verification set are fed into the nerve network, the model is trained through the training set, and the verification set ensures that the model is not fitted and has good mobility. The neural network generator adopts an adam optimizer to train model parameters of the neural network generator, so that the obtained false data sequence is gradually close to the primary vision data. The discriminator is trained by a two-to-cross entropy optimizer to generate fitting performance scoring of false data, and the generator is continuously driven to generate more accurate sequence data. Thirdly, in the training process of the model, the matplotlib module is used for visualizing the comparison of the predicted data and the original data once after every 20 training steps so as to determine the training state and adjust the parameters in time. And finally, feeding x in the test set into the trained neural network, and judging the accuracy of the predicted result and the actual experimental data through relative percentage errors.
The effect of the data enhancement method on the prediction result is described with reference to fig. 4 and 5: typically, experimental data that needs to be obtained for aircraft control are: left and right amplitude [0 °,40 ° ], a set of data is tested every 5 °; a set of data was tested every 0.1Hz at frequencies [0.3Hz,0.7Hz ]; the phase difference [0 degree, 40 degree ], one group of data is tested every 10 degrees, 1600 groups of working conditions are needed for the full arrangement, and the difficulty is great for experiments. By the method, partial working conditions can be measured according to larger step length, and hydrodynamic time-varying data corresponding to the working conditions of the interpolation value variable are fitted by using the antagonistic neural network, so that the experimental cost is greatly reduced. Meanwhile, as can be obtained from fig. 5, the prediction result of the interpolation value has better fitting degree than the force and moment values of the original experimental result.
While the invention has been described with reference to certain preferred embodiments, it will be understood by those skilled in the art that various changes and substitutions of equivalents may be made without departing from the spirit and scope of the invention.
Claims (7)
1. A simulated ray aircraft experimental data prediction method based on an antagonistic neural network is characterized by comprising the following steps:
s1: data acquisition
The six-dimensional force time sequence data of the prototype periodic variation corresponding to different input motion parameter labels is measured through a hydrodynamic force experiment of the bionic aircraft; the motion parameter label comprises a left amplitude, a right amplitude, a frequency and a phase difference;
s1, experimental data are obtained through a hydrodynamic experiment of the simulated ray-light aircraft, wherein the experiment is to set different motion parameter combinations for the simulated ray-light aircraft, and the six-dimensional force sensor in the simulated ray-light aircraft is used for measuring the sequence data of the changes of the stress and the moment of the simulated ray-light aircraft along the direction of the spanwise direction, the chord direction and the plumb direction;
s2: data set partitioning
Setting different input motion parameter labels as input x of the neural network, and setting corresponding six-dimensional force time series data as output y; and all data were taken at 8:1:1, dividing a training set, a verification set and a test set in proportion;
s3: training neural networks
Setting up an antagonistic neural network, taking x in a training set and a verification set as input variables, and taking y as output variables to be input into the neural network, wherein parameters and super parameters of the training network;
s4: predicting experimental results
Inputting variable labels x in the test set into a trained neural network, and predicting corresponding sequence data y;
s5: verifying the experimental results
And comparing the predicted result with the experimental result, calculating the relative percentage error of the predicted result, and comparing the relative percentage error with the predicted result of the original data set.
2. The experimental data prediction method for the artificial bata ray aircraft based on the antagonistic neural network according to claim 1, which is characterized in that: in the step S1, the data of the pectoral fin flapping movement period of the 10 simulated ray aircraft are obtained through one experiment, and the obtained six-dimensional force time series data have the same strong periodicity.
3. The experimental data prediction method for the artificial bata ray aircraft based on the antagonistic neural network according to claim 1, which is characterized in that: all data in S2 are as per 8:1:1, before dividing the training set, the verification set and the test set in proportion, the original data set needs to be randomly disturbed according to the same random seed.
4. The experimental data prediction method for the artificial bata ray aircraft based on the antagonistic neural network according to claim 1, which is characterized in that: s3, the countermeasure neural network consists of two neural networks of a generator and a discriminator, wherein a generator model consists of four one-dimensional transposition convolutional layers and a full-connection layer, the activation functions of the transposition convolutional layers are Relu, and the activation functions of the full-connection layer are tanh so as to keep nonlinear characteristics of the model; the arbiter model consists of three one-dimensional convolution layers and one full-connection layer, wherein the activation functions of the convolution layers are Relu, and the activation functions of the full-connection layer are Sigmoid.
5. The experimental data prediction method for the artificial bata ray aircraft based on the antagonistic neural network according to claim 4, which is characterized in that: optimizing a weight parameter set of a generator model of the antagonistic neural network by adopting an Adam optimizer, and setting the initial learning rate to be 0.001 and the gradient attenuation rate to be 0.9; the loss function of the generator is set as the mean square error and the loss function of the arbiter is set as the two-to-cross entropy.
6. The experimental data prediction method for the artificial bata ray aircraft based on the antagonistic neural network according to claim 5, which is characterized in that: the antagonistic neural network dataset was trained in a batch of every 32 divisions.
7. The experimental data prediction method for the artificial bata ray aircraft based on the antagonistic neural network according to claim 1, which is characterized in that: and S5, the prediction result is visualized through a matplotlib module in the neural network program, and a predicted line graph is output for comparison with the line graph of the original data every 20 training rounds, so that the state of model training is conveniently determined, and parameters are conveniently adjusted at any time.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210652320.XA CN115081318B (en) | 2022-06-10 | 2022-06-10 | Artificial bata ray based on antagonistic neural network method for predicting experimental data of aircraft |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210652320.XA CN115081318B (en) | 2022-06-10 | 2022-06-10 | Artificial bata ray based on antagonistic neural network method for predicting experimental data of aircraft |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115081318A CN115081318A (en) | 2022-09-20 |
| CN115081318B true CN115081318B (en) | 2024-01-26 |
Family
ID=83252283
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210652320.XA Active CN115081318B (en) | 2022-06-10 | 2022-06-10 | Artificial bata ray based on antagonistic neural network method for predicting experimental data of aircraft |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115081318B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7607610B1 (en) * | 2007-04-23 | 2009-10-27 | Robert Sterchak | Ornithopter having a wing structure and a mechanism for imparting realistic, bird-like motion thereto |
| CN113188757A (en) * | 2021-01-27 | 2021-07-30 | 西北工业大学 | Hydrodynamic performance test platform and test method of simulated bat ray pectoral fin prototype |
| CN113325858A (en) * | 2021-06-08 | 2021-08-31 | 西北工业大学 | Simulated bat aircraft course control method based on flapping wing amplitude |
| CN113804398A (en) * | 2021-08-23 | 2021-12-17 | 西北工业大学 | Cluster large-scale three-dimensional flow field and hydrodynamic force synchronous measurement system and test method |
| CN114201930A (en) * | 2021-11-24 | 2022-03-18 | 南京航空航天大学 | Transition point detection method for flow field analysis of underwater vehicle |
-
2022
- 2022-06-10 CN CN202210652320.XA patent/CN115081318B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7607610B1 (en) * | 2007-04-23 | 2009-10-27 | Robert Sterchak | Ornithopter having a wing structure and a mechanism for imparting realistic, bird-like motion thereto |
| CN113188757A (en) * | 2021-01-27 | 2021-07-30 | 西北工业大学 | Hydrodynamic performance test platform and test method of simulated bat ray pectoral fin prototype |
| CN113325858A (en) * | 2021-06-08 | 2021-08-31 | 西北工业大学 | Simulated bat aircraft course control method based on flapping wing amplitude |
| CN113804398A (en) * | 2021-08-23 | 2021-12-17 | 西北工业大学 | Cluster large-scale three-dimensional flow field and hydrodynamic force synchronous measurement system and test method |
| CN114201930A (en) * | 2021-11-24 | 2022-03-18 | 南京航空航天大学 | Transition point detection method for flow field analysis of underwater vehicle |
Non-Patent Citations (4)
| Title |
|---|
| AUV智能化现状与发展趋势;黄炎;《机器人》;全文 * |
| Data-Driven Prediction of Experimental Hydrodynamic Data of the Manta Ray Robot Using Deep Learning Method;Jingyi Bai;《MDPI》;全文 * |
| Field Experiments for Dynamic Characteristics and Motion Control of a Manta-type Autonomous Underwater Vehicle;Jong Hyeon Park;《RS》;全文 * |
| 仿生型水下航行器研究现状及发展趋势;李成进;;鱼雷技术(第01期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115081318A (en) | 2022-09-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110378052A (en) | It is looked to the future the equipment method for predicting residual useful life of operating condition based on Recognition with Recurrent Neural Network | |
| CN109343505A (en) | Gear method for predicting residual useful life based on shot and long term memory network | |
| CN108596327B (en) | Seismic velocity spectrum artificial intelligence picking method based on deep learning | |
| CN108106844B (en) | A kind of Method for Bearing Fault Diagnosis of the automatic coding machine of self-tuning parameter adjustment | |
| CN109163911A (en) | A kind of fault of engine fuel system diagnostic method based on improved bat algorithm optimization ELM | |
| CN110297479B (en) | Hydroelectric generating set fault diagnosis method based on convolutional neural network information fusion | |
| CN111412977A (en) | Preprocessing method for vibration sensing data of mechanical equipment | |
| CN108564326A (en) | Prediction technique and device, computer-readable medium, the logistics system of order | |
| CN114969995A (en) | Rolling bearing early fault intelligent diagnosis method based on improved sparrow search and acoustic emission | |
| CN109934422A (en) | Neural network wind speed prediction method based on time series data analysis | |
| CN108960421A (en) | The unmanned surface vehicle speed of a ship or plane online forecasting method based on BP neural network of improvement | |
| CN111624522A (en) | Ant colony optimization-based RBF neural network control transformer fault diagnosis method | |
| CN115859077A (en) | Multi-feature fusion motor small sample fault diagnosis method under variable working conditions | |
| CN112287592A (en) | Industrial equipment fault diagnosis method and system based on deep confidence network | |
| CN111428788B (en) | Deep learning-based multi-fault diagnosis method and system for steam turbine generator set rotor | |
| CN111967308A (en) | Online road surface unevenness identification method and system | |
| CN113822201A (en) | Deep learning method for underwater object shape recognition based on flow field velocity component time course | |
| CN115587290A (en) | Aero-engine fault diagnosis method based on variational self-coding generation countermeasure network | |
| CN111191823A (en) | Production logistics prediction method based on deep learning | |
| CN115081318B (en) | Artificial bata ray based on antagonistic neural network method for predicting experimental data of aircraft | |
| CN113200086A (en) | Intelligent vehicle steering control system and control method thereof | |
| CN113076920A (en) | Intelligent fault diagnosis method based on asymmetric domain confrontation self-adaptive model | |
| CN110261108B (en) | CNN color characteristic diagram-based bearing fault identification method during rated operation | |
| CN117009841A (en) | Model training method, motor fault diagnosis method and microcontroller | |
| CN116910513A (en) | Centrifugal pump rotor fault diagnosis method based on CWGAN-GP and double-flow CNN model |
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 |