CN114169396A - Training data generation model construction method and application for aircraft fault diagnosis - Google Patents
Training data generation model construction method and application for aircraft fault diagnosis Download PDFInfo
- Publication number
- CN114169396A CN114169396A CN202111307993.3A CN202111307993A CN114169396A CN 114169396 A CN114169396 A CN 114169396A CN 202111307993 A CN202111307993 A CN 202111307993A CN 114169396 A CN114169396 A CN 114169396A
- Authority
- CN
- China
- Prior art keywords
- data
- aircraft
- sample data
- training
- fault diagnosis
- 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.)
- Granted
Links
- 238000012549 training Methods 0.000 title claims abstract description 97
- 238000003745 diagnosis Methods 0.000 title claims abstract description 89
- 238000010276 construction Methods 0.000 title claims abstract description 26
- 238000000034 method Methods 0.000 claims description 39
- 238000010801 machine learning Methods 0.000 claims description 7
- 230000008569 process Effects 0.000 claims description 6
- 238000012952 Resampling Methods 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 238000007476 Maximum Likelihood Methods 0.000 abstract description 2
- 238000002474 experimental method Methods 0.000 description 12
- 238000005070 sampling Methods 0.000 description 10
- 230000006870 function Effects 0.000 description 7
- 230000000052 comparative effect Effects 0.000 description 6
- 238000013135 deep learning Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 238000013136 deep learning model Methods 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000005094 computer simulation Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Images
Classifications
-
- 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
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Medical Informatics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention discloses a training data generation model construction method for aircraft fault diagnosis and application, belonging to the technical field of aircraft fault diagnosis, and comprising S1, constructing a training data generation model comprising a generator and a discriminator; s2, inputting a pre-collected aircraft sample data set into a training data generation model for training, and enabling a generator and a discriminator to play games with each other until Nash balance is achieved; the generator is based on a VAE model, input data are reconstructed by carrying out maximum likelihood estimation on data distribution, data similar to original data are generated, and meanwhile, the authenticity of the generated data is judged through a discriminator, so that the reliability and the authenticity of the data are improved; by adopting the training data generation model provided by the invention, real and accurate sample generation can be realized, so that the technical problems of data shortage and class unbalance are solved under the condition of ensuring the accuracy of fault diagnosis.
Description
Technical Field
The invention belongs to the technical field of aircraft fault diagnosis, and particularly relates to a training data generation model construction method for aircraft fault diagnosis and application.
Background
The failure of an aircraft during flight can result in serious economic loss and casualties. In order to provide a plan for a fault to be controlled when the fault occurs, a fault diagnosis technique is very important.
The fault diagnosis technology is a state identification technology which evaluates the state of equipment by using the current state information and the historical state of the equipment through a certain analysis method. Fault diagnosis is often divided into two categories: model-based methods and data-based methods; the method based on the model diagnoses the fault according to the abnormity of the state quantity of the aircraft by establishing an accurate mathematical model, but in the actual application process, a lot of model uncertainty and external disturbance cannot be modeled, so that the method is difficult to apply. In the data-based method, multiple parts are diagnosed for aircraft faults based on a deep learning model, and with the rapid development of deep learning and the wide migration in various fields, structures such as CNN, LSTM, Transformer and the like have been used in the field of fault diagnosis for a long time. However, the deep learning based approach has two fatal drawbacks, namely lack of data and class imbalance. On one hand, the lack of data is a common problem of all deep learning tasks, and on the other hand, in an industrial system represented by an aircraft, the occurrence of faults is avoided to the utmost extent, so that the fault data is extremely less than normal data, and the data of a specific part with faults is much less and less, which seriously affects the diagnosis effect of the deep learning diagnosis network. It is well known that the choice of data set affects the results far beyond the network structure itself.
In order to solve the defects, aircraft data are simulated in the existing research, but the data distribution difference still exists between the result of computer simulation and the actually acquired data, and the application of the data on the aircraft has risks. Therefore, aiming at the problems of data deficiency and unbalanced category, data is often expanded based on a few actually acquired sample data rather than being directly simulated, and an easily-conceivable method is to oversample the few sample data, which is called a SMOTE method. The other idea is to group data, namely, divide an unbalanced training data set into a plurality of balanced data sets according to a certain rule, and integrate a plurality of classifiers obtained by training on the balanced data sets according to a certain learning method, so as to eliminate the problem of class imbalance.
Disclosure of Invention
Aiming at the defects or the improvement requirements of the prior art, the invention provides a training data generation model construction method and application for aircraft fault diagnosis, which are used for solving the technical problems that the prior art cannot solve data shortage and class imbalance under the condition of ensuring the fault diagnosis accuracy.
In order to achieve the above object, the present invention provides a training data generation model construction method for aircraft fault diagnosis, comprising the following steps:
s1, building a training data generation model;
the training data generation model includes: a generator and a discriminator; the generator is based on a VAE model and is used for sequentially carrying out encoding and decoding operations on input aircraft sample data so as to reconstruct the aircraft sample data and obtain pseudo data similar to the aircraft sample data; the discriminator is used for discriminating whether the input data is real aircraft data;
s2, inputting a pre-collected aircraft sample data set into a training data generation model for training, and enabling a generator and a discriminator to play games with each other until Nash balance is achieved; the aircraft sample data set includes aircraft fault sample data.
Further preferably, the generator includes: a first MLP-Mixer, a VAE model, and a second MLP-Mixer in cascade;
the first MLP-Mixer is used for expanding dimensionality of input aircraft sample data and integrating features of different channels of the expanded aircraft sample data, so that the aircraft sample data is mapped into depth data features;
the VAE model is used for coding the depth data features through a coder to obtain a mean value and a variance corresponding to the depth data features, resampling in normal distribution corresponding to the obtained mean value and variance to obtain a hidden variable, and reconstructing the hidden variable through a decoder to obtain a pseudo data feature similar to the depth data feature;
and the second MLP-Mixer is used for mapping the pseudo data characteristics according to different channels, so that the pseudo data characteristics are reversely mapped into pseudo data similar to the sample data of the aircraft.
Further preferably, step S2 includes:
s21, respectively generating pseudo data corresponding to the sample data for each sample data in the pre-collected aircraft sample data set through a generator, and respectively inputting the sample data and the corresponding pseudo data into a discriminator to obtain the prediction probability that the sample data and the corresponding pseudo data are respectively discriminated as the real data of the aircraft;
s22, obtaining a loss value of the generator by calculating the sum of the expression difference and the distribution difference between the sample data and the corresponding pseudo data; obtaining a loss value of the discriminator by calculating the sum of differences between the predicted probability of the sample data and the corresponding true probability of the aircraft true data respectively judged by the sample data and the corresponding pseudo data;
s23, judging whether the sum of the loss value of the generator and the loss value of the discriminator reaches the minimum or whether the iteration number reaches the preset iteration number, if so, finishing training of the training data generation model, and ending the operation; otherwise, the parameters in the generator and the discriminator are updated, and the process goes to step S21.
Further preferably, the loss value of the generator is:
LG＝MSE+KLD
wherein MSE is the expression difference between the sample data and the corresponding dummy data; n is the number of sample data in the aircraft sample data set;
the data is the pseudo data corresponding to the ith sample data;
the ith sample data; KLD is KL divergence of sample data distribution and corresponding pseudo data distribution; mu.siAnd σiRespectively the mean and variance corresponding to the ith sample data output by the encoder.
Further preferably, the loss value of the discriminator is:
wherein,
judging the sample data as the prediction probability of the real data of the aircraft;
the sample data is the true probability of the true data of the aircraft;
and judging the pseudo data corresponding to the sample data as the prediction probability of the real data of the aircraft.
In a second aspect, the present invention provides a training data generation method for aircraft fault diagnosis, comprising:
expanding the aircraft sample data set by controlling the expansion times of the pre-collected sample data in the aircraft sample data set, so that the number of the sample data in the expanded aircraft sample data set reaches a preset number, and the number of the aircraft fault sample data and the number of the aircraft normal sample data reach balance; the obtained extended aircraft sample data set is a training data set for aircraft fault diagnosis;
the extended sample data is aircraft fault sample data and/or aircraft normal sample data;
the method for expanding the sample data comprises the following steps:
inputting sample data into a generator of a training data generation model constructed by adopting the training data generation model construction method provided by the first aspect of the invention to obtain pseudo data corresponding to the sample data so as to expand the sample data; the fault information tag of the dummy data is the same as the fault information tag of the sample data corresponding thereto.
In a third aspect, the invention provides a method for constructing an aircraft fault diagnosis model, which comprises the following steps:
and inputting a training data set generated by adopting the training data generation method provided by the second aspect of the invention into a machine learning model for training to obtain an aircraft fault diagnosis model.
Further preferably, the machine learning model is a Transformer model.
In a fourth aspect, the present invention provides an aircraft fault diagnosis method, comprising: and inputting the flight sampler data into the aircraft fault diagnosis model constructed by the aircraft fault diagnosis model construction method provided by the third aspect of the invention to obtain aircraft fault information.
In a fifth aspect, the present invention also provides a machine-readable storage medium storing machine-executable instructions that, when invoked and executed by a processor, cause the processor to implement one or more of the training data generation model construction method provided by the first aspect of the present invention, the training data generation method provided by the second aspect of the present invention, the aircraft fault diagnosis model construction method provided by the third aspect of the present invention, and the aircraft fault diagnosis method provided by the fourth aspect of the present invention.
Generally, by the above technical solution conceived by the present invention, the following beneficial effects can be obtained:
1. the invention provides a training data generation model construction method for aircraft fault diagnosis, wherein the constructed training data generation model comprises a generator and a discriminator; the generator is based on a VAE model, the reconstruction of input data is realized by carrying out maximum likelihood estimation on data distribution, data similar to original data is generated, and meanwhile, the authenticity of the generated data is judged through a discriminator; through the mutual game between the generator and the discriminator, the reliability and the authenticity of the data are increased; by adopting the training data generation model provided by the invention, real and accurate sample generation can be realized, so that the technical problems of data shortage and class unbalance are solved under the condition of ensuring the accuracy of fault diagnosis.
2. In the training data generation model construction method provided by the invention, the MLP-Mixer is introduced into the generator, and information in different channels of a spatial domain is fused during training in the network, so that the characteristic enhancement effect is achieved.
Drawings
Fig. 1 is a flowchart of a training data generation model construction method for aircraft fault diagnosis according to embodiment 1 of the present invention;
fig. 2 is a schematic structural diagram of a training data generation model provided in embodiment 1 of the present invention;
fig. 3 is a schematic structural diagram of a generator provided in embodiment 1 of the present invention;
fig. 4 is a schematic structural diagram of a VAE model provided in embodiment 1 of the present invention;
FIG. 5 shows the experimental results obtained by comparing the real fault diagnosis values of four fault actuators of an aircraft after training the same aircraft fault diagnosis model with the data set extended by the training data generation model provided by the invention and the actually acquired aircraft sample data set provided by comparative experiment 1;
fig. 6 is an experimental result obtained by comparing the real fault diagnosis value with the real fault diagnosis value after the same aircraft fault diagnosis model is trained by respectively adopting the model with the MLP-Mixer and the model without the MLP-Mixer provided by the comparative experiment 2 of the present invention, and fault diagnosis is performed on four fault actuators of the aircraft.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
Examples 1,
A training data generation model building method for aircraft fault diagnosis, as shown in fig. 1, includes the following steps:
s1, building a training data generation model;
as shown in fig. 2, the training data generation model includes: a generator and a discriminator; the generator is based on a VAE model and is used for sequentially carrying out encoding and decoding operations on input aircraft sample data so as to reconstruct the aircraft sample data and obtain pseudo data similar to the aircraft sample data; the discriminator is used for discriminating whether the input data is real aircraft data; in this embodiment, the training data generation model is referred to as an invariant self encoding Generator (AVAEG).
Preferably, in an alternative embodiment, as shown in fig. 3, the generator comprises: a first MLP-Mixer, a VAE model, and a second MLP-Mixer in cascade;
the first MLP-Mixer is used for expanding dimensionality of input aircraft sample data and integrating features of different channels of the expanded aircraft sample data, so that the aircraft sample data is mapped into depth data features; it should be noted that the aircraft sample data used in this embodiment includes attitude data and fault data of the aircraft, which have small dimensions but contain a very large amount of information. When the fault condition of the aircraft is judged by comprehensively considering the data of different state quantities, the data needs to be processed, and deeper characteristic information is extracted.
The VAE model is used for coding the depth data features through a coder to obtain a mean value and a variance corresponding to the depth data features, resampling in normal distribution corresponding to the obtained mean value and variance to obtain a hidden variable, and reconstructing the hidden variable through a decoder to obtain a pseudo data feature similar to the depth data feature;
specifically, the VAE model employed in the present embodiment is shown in fig. 4, in which the encoder includes a convolutional layer; the output of the encoder corresponding to the ith sample data is
Can be divided into two parts, i.e. mui,σi∈RhH represents a hidden variable dimension; mu.siAnd σiRespectively the mean and variance corresponding to the ith sample data output by the encoder.
The output of the encoder is resampled to obtain a hidden variable ziHidden variable ziIs the output z of the encoderiAnd a complex function of the random variable epsilon satisfying a gaussian distribution, namely:
wherein z isi∈Rh，
εi～N(0,1)，fh(. cndot.) is a hidden variable function.
In this embodiment, the decoder includes an deconvolution layer; hidden variable ziInputting the pseudo data characteristics into a decoder to obtain the pseudo data characteristics, wherein the dimensionality of the pseudo data characteristics is the same as the depth data characteristics of the input encoder;
and the second MLP-Mixer is used for mapping the pseudo data characteristics according to different channels, so that the pseudo data characteristics are reversely mapped into pseudo data similar to the sample data of the aircraft.
S2, inputting a pre-collected aircraft sample data set into a training data generation model for training, and enabling a generator and a discriminator to play games with each other until Nash balance is achieved; the aircraft sample data set includes aircraft fault sample data.
Specifically, in the embodiment, a small data set including aircraft fault sample data is collected in advance in a real aircraft scene and recorded as an aircraft sample data set; each sample data comprises aircraft state information, specifically comprises a state information vector consisting of an aircraft state quaternion and an attitude angle
Meanwhile, the fault information of each sample data is also acquired, and specifically comprises a fault information vector c formed by fault conditions of four actuating mechanisms of the aircraftr. Noting the aircraft sample data set as
Wherein,
n is the data volume of the data set, feature _ num is the element number of the state information vector, and label _ num is the element number of the fault information vector.
In the training phase, sample data in the aircraft sample data set
Obtaining corresponding pseudo data through a generator
Next, the sample data is processed
With corresponding dummy data
Respectively sent to a discriminator to make the discriminator distinguish true from false. The training process is the game of the generator and the discriminator, and the training is stopped when the two theoretically reach Nash equilibrium.
Specifically, the schematic diagram of the training phase of the AVAEG is shown in fig. 3, and in an alternative embodiment, the step S2 includes:
s21, respectively generating pseudo data corresponding to the sample data for each sample data in the pre-collected aircraft sample data set through a generator, and respectively inputting the sample data and the corresponding pseudo data into a discriminator to obtain the prediction probability that the sample data and the corresponding pseudo data are respectively discriminated as the real data of the aircraft;
s22, obtaining a loss value of the generator by calculating the sum of the expression difference and the distribution difference between the sample data and the corresponding pseudo data; obtaining a loss value of the discriminator by calculating the sum of differences between the predicted probability of the sample data and the corresponding true probability of the aircraft true data respectively judged by the sample data and the corresponding pseudo data;
specifically, the loss values of the generator are:
LG＝MSE+KLD
wherein MSE is the expression difference between the sample data and the corresponding dummy data; n is the number of sample data in the aircraft sample data set;
the data is the pseudo data corresponding to the ith sample data;
the ith sample data; KLD is KL divergence of sample data distribution and corresponding pseudo data distribution; mu.siAnd σiRespectively the mean and variance corresponding to the ith sample data output by the encoder.
Specifically, the loss value of the discriminator is:
wherein,
judging the sample data as the prediction probability of the real data of the aircraft;
the sample data is the true probability of the true data of the aircraft;
and judging the pseudo data corresponding to the sample data as the prediction probability of the real data of the aircraft.
S23, judging whether the sum of the loss value of the generator and the loss value of the discriminator reaches the minimum or whether the iteration number reaches the preset iteration number (the preset iteration number Epoch generally takes the value of 100), if so, finishing training the training data generation model, and ending the operation; otherwise, the parameters in the generator and the discriminator are updated, and the process goes to step S21.
Specifically, the loss function of the training data generation model is: l ═ LG+LD(ii) a And reversely updating parameters in the generator and the loss value according to the loss function value of the data generation model so as to train the training data generation model.
Examples 2,
A method of generating training data for aircraft fault diagnosis, comprising:
expanding the aircraft sample data set by controlling the expansion times of the pre-collected sample data in the aircraft sample data set, so that the number of the sample data in the expanded aircraft sample data set reaches a preset number, and the number of the aircraft fault sample data and the number of the aircraft normal sample data reach balance; the obtained extended aircraft sample data set is a training data set for aircraft fault diagnosis;
the extended sample data is aircraft fault sample data and/or aircraft normal sample data;
the method for expanding the sample data comprises the following steps:
inputting sample data into a generator of a training data generation model constructed by adopting the training data generation model construction method provided by the embodiment 1 to obtain pseudo data corresponding to the sample data so as to expand the sample data; the fault information tag of the dummy data is the same as the fault information tag of the sample data corresponding thereto.
Specifically, a scene that the collected aircraft sample data set contains less sample data and the number of fault sample data and normal sample data is unbalanced is taken as an example;
firstly, an aircraft sample dataset is collected
The sample data in (1) are respectively sent to a generator, each sample data is fed into the generator M (in the embodiment, M takes the value as 100) times to obtain M pseudo data, and each generated pseudo data is directly distributed with a fault information vector (fault information label) according to the original sample data. Aircraft sample data set
Generated data set obtained by this method
The data volume of the aircraft is expanded by 100 times, and an aircraft sample data set is merged
And generating a data set
Obtaining a new data set
Thereby effectively solving the problem that the data set is too small.
Furthermore, in order to solve the problem of category imbalance, the frequency of feeding the aircraft fault sample data into the generator can be controlled according to the quantity ratio Q of the aircraft normal sample data to the aircraft fault sample data in the new data set, so that the quantity of the aircraft normal sample data and the aircraft fault sample data is balanced; in this embodiment, each fault sample is fed to the generator Q times to achieve class balancing.
The related technical scheme is the same as embodiment 1, and is not described herein.
Examples 3,
A method for constructing an aircraft fault diagnosis model comprises the following steps:
the training data set generated by the training data generation method provided in embodiment 2 is input into a machine learning model (deep learning model) for training, and an aircraft fault diagnosis model is obtained.
Specifically, the loss function of the aircraft fault diagnosis model is:
wherein K is the number of training data in the training data set;
a predicted value of aircraft fault information of kth training data output by the aircraft fault diagnosis model; c. CkThe true value of the aircraft fault information for the kth training data.
And reversely updating parameters in the aircraft fault diagnosis model based on the loss function value of the aircraft fault diagnosis model, and repeating the process until the network converges, thereby minimizing the loss function value of the aircraft fault diagnosis model.
The aircraft fault diagnosis model can be a common machine learning model (deep learning model); preferably, in an alternative embodiment, the machine learning model is a Transformer model.
The related technical scheme is the same as embodiment 2, and is not described herein.
Examples 4,
An aircraft fault diagnosis method comprising: and inputting the flight sampler data into the aircraft fault diagnosis model constructed by the aircraft fault diagnosis model construction method provided in embodiment 3 to obtain aircraft fault information.
The related technical scheme is the same as embodiment 3, and is not described herein.
Examples 5,
A machine-readable storage medium storing machine-executable instructions that, when invoked and executed by a processor, cause the processor to implement one or more of the training data generation model construction method provided by embodiment 1, the training data generation method provided by embodiment 2, the aircraft fault diagnosis model construction method provided by embodiment 3, and the aircraft fault diagnosis method provided by embodiment 4.
The related technical solutions are the same as those in embodiments 1 to 4, and are not described herein.
To further illustrate the training data generation model provided by the present invention, details are given below in conjunction with comparative experiment 1 and comparative experiment 2:
In the experiment, the same aircraft fault diagnosis model is trained by respectively adopting a truly acquired aircraft sample data set (small data set) and a data set expanded by the training data generation model provided by the invention, and then four fault actuators (respectively marked as f) of the aircraft are carried outa1、f a2、f a3 and fa4) After fault diagnosis is carried out, comparing with a real fault diagnosis value to obtain an experimental result shown in fig. 5; wherein the abscissa is a sampling point obtained by sampling the data of the aircraft flying continuously for 1s for 100 times; the ordinate is the diagnostic value at the sampling point. In the experiment, fault diagnosis is respectively carried out on each sampling point based on the model, and then the sampling points are arranged to obtain a result shown in fig. 5; as can be seen from fig. 5, compared with the fault diagnosis result of the aircraft fault diagnosis model obtained by training a small data set, the fault diagnosis result of the aircraft fault diagnosis model obtained by training the data set after the training data generation model provided by the invention is extended is closer to the real fault diagnosis value. In addition, the mean square error MSE of the fault diagnosis result of the aircraft fault diagnosis model obtained by training the small data set and the fault diagnosis result of the aircraft fault diagnosis model obtained by training the data set after the training data generation model provided by the invention is calculated respectively, and the diagnosis MSE corresponding to the small data set is 0.025256619 and the diagnosis MSE after the data expansion is 0.0022036468. As can be seen from the above experiments, the training data generation model provided by the invention has an order of magnitude optimization on the accuracy of the aircraft diagnosis result.
In this experiment, the samples were divided intoAfter the same aircraft fault diagnosis model is trained by adopting the training data sets generated by the first experimental model and the second experimental model, four fault actuators (respectively marked as f) of the aircraft are carried outa1、f a2、f a3 and fa4) After fault diagnosis is carried out, comparing with a real fault diagnosis value to obtain an experimental result shown in fig. 6; the first experimental model and the second experimental model are respectively a training data generation model (model with MLP-Mixer added) provided by the invention and a model (model without MLP-Mixer added) with MLP-Mixer removed from the training data generation model provided by the invention. Similarly, the abscissa is a sampling point obtained by sampling the data of the aircraft flying for 1s for 100 times; the ordinate is the diagnostic value at the sampling point. In the experiment, fault diagnosis is respectively carried out on each sampling point based on the model, and then the sampling points are arranged to obtain a result shown in fig. 6; as can be seen from fig. 6, the fault diagnosis result obtained by generating the model (model with MLP-Mixer) based on the training data provided by the present invention is closer to the real fault diagnosis value than the fault diagnosis result obtained by generating the model without adding the MLP-Mixer. In addition, the mean square error MSE of the fault diagnosis result obtained based on the model without the MLP-Mixer and the fault diagnosis result obtained based on the training data generation model (the model with the MLP-Mixer) provided by the invention are respectively calculated, and the diagnosis MSE corresponding to the model without the MLP-Mixer is 0.0013495281 and the diagnosis MSE corresponding to the model with the MLP-Mixer is 0.0005102753. The experiment shows that the MLP-Mixer is introduced, so that the accuracy of the aircraft diagnosis result can be greatly improved, and the performance is better.
It will be understood by those skilled in the art that the foregoing is only a preferred embodiment of the present invention, and is not intended to limit the invention, and that any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (10)
1. A training data generation model construction method for aircraft fault diagnosis is characterized by comprising the following steps:
s1, building a training data generation model;
the training data generation model includes: a generator and a discriminator; the generator is based on a VAE model and is used for sequentially carrying out encoding and decoding operations on input aircraft sample data so as to reconstruct the aircraft sample data and obtain pseudo data similar to the aircraft sample data; the discriminator is used for discriminating whether the input data is real aircraft data;
s2, inputting a pre-collected aircraft sample data set into the training data generation model for training, and enabling the generator and the discriminator to play games with each other until Nash equilibrium is achieved; the aircraft sample data set includes aircraft fault sample data.
2. The training data generation model construction method according to claim 1, wherein the generator includes: a first MLP-Mixer, a VAE model, and a second MLP-Mixer in cascade;
the first MLP-Mixer is used for expanding dimensionality of input aircraft sample data and integrating features of different channels of the expanded aircraft sample data, so that the aircraft sample data is mapped into depth data features;
the VAE model is used for coding the depth data features through a coder to obtain a mean value and a variance corresponding to the depth data features, resampling is carried out in normal distribution corresponding to the obtained mean value and variance to obtain a hidden variable, and the hidden variable is reconstructed through a decoder to obtain pseudo data features similar to the depth data features;
the second MLP-Mixer is used for mapping the pseudo data characteristics according to different channels, so that the pseudo data characteristics are reversely mapped into pseudo data similar to the aircraft sample data.
3. The training data generative model construction method according to claim 1 or 2, wherein said step S2 comprises:
s21, respectively generating pseudo data corresponding to the sample data for each sample data in the aircraft sample data set through the generator, and respectively inputting the sample data and the corresponding pseudo data into the discriminator to obtain the prediction probability that the sample data and the corresponding pseudo data are respectively discriminated as the real data of the aircraft;
s22, calculating the sum of the expression difference and the distribution difference between the sample data and the corresponding pseudo data to obtain the loss value of the generator; obtaining a loss value of the discriminator by calculating the sum of differences between the predicted probability of the sample data and the corresponding true probability of the aircraft true data respectively judged by the sample data and the corresponding pseudo data;
s23, judging whether the sum of the loss value of the generator and the loss value of the discriminator reaches the minimum or whether the iteration number reaches the preset iteration number, if so, finishing training of the training data generation model, and ending the operation; otherwise, the parameters in the generator and the discriminator are updated, and the process goes to step S21.
4. The training data generation model construction method according to claim 3, wherein the loss value of the generator is:
LG＝MSE+KLD
wherein MSE is the expression difference between the sample data and the corresponding dummy data; n is the number of sample data in the aircraft sample data set;
the data is the pseudo data corresponding to the ith sample data;
the ith sample data; KLD is KL divergence of sample data distribution and corresponding pseudo data distribution; mu.siAnd σiRespectively, the mean and the variance corresponding to the ith sample data output by the encoder.
5. The training data generative model construction method according to claim 3, wherein the loss value of the discriminator is:
wherein,
judging the sample data as the prediction probability of the real data of the aircraft;
the sample data is the true probability of the true data of the aircraft;
and judging the pseudo data corresponding to the sample data as the prediction probability of the real data of the aircraft.
6. A method of generating training data for aircraft fault diagnosis, comprising:
expanding the aircraft sample data set by controlling the expansion times of the pre-collected sample data in the aircraft sample data set, so that the number of the sample data in the expanded aircraft sample data set reaches a preset number, and the number of the aircraft fault sample data and the number of the aircraft normal sample data reach a balance; the extended aircraft sample data set is a training data set for aircraft fault diagnosis;
the extended sample data is aircraft fault sample data and/or aircraft normal sample data;
the method for expanding the sample data comprises the following steps:
inputting the sample data into a generator of a training data generation model constructed by adopting the training data generation model construction method according to any one of claims 1 to 5 to obtain pseudo data corresponding to the sample data so as to expand the sample data; and the fault information label of the pseudo data is the same as the fault information label of the sample data corresponding to the pseudo data.
7. A method for constructing an aircraft fault diagnosis model is characterized by comprising the following steps:
inputting a training data set generated by the training data generation method of claim 6 into a machine learning model for training to obtain an aircraft fault diagnosis model.
8. The method of constructing an aircraft fault diagnosis model of claim 7, wherein the machine learning model is a Transformer model.
9. An aircraft fault diagnosis method, comprising: inputting flight sampler data into the aircraft fault diagnosis model constructed by the aircraft fault diagnosis model construction method according to claim 7 or 8 to obtain aircraft fault information.
10. A machine-readable storage medium having stored thereon machine-executable instructions that, when invoked and executed by a processor, cause the processor to perform one or more of the training data generation model construction method of any one of claims 1 to 5, the training data generation method of claim 6, the aircraft fault diagnosis model construction method of any one of claims 7 to 8, and the aircraft fault diagnosis method of claim 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111307993.3A CN114169396B (en) | 2021-11-05 | 2021-11-05 | Training data generation model construction method and application for aircraft fault diagnosis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111307993.3A CN114169396B (en) | 2021-11-05 | 2021-11-05 | Training data generation model construction method and application for aircraft fault diagnosis |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114169396A true CN114169396A (en) | 2022-03-11 |
| CN114169396B CN114169396B (en) | 2022-09-20 |
Family
ID=80478112
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111307993.3A Active CN114169396B (en) | 2021-11-05 | 2021-11-05 | Training data generation model construction method and application for aircraft fault diagnosis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114169396B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114963408A (en) * | 2022-05-25 | 2022-08-30 | 日照安泰科技发展有限公司 | Air conditioning system unbalanced fault diagnosis method based on data and feature enhancement |
| CN116774719A (en) * | 2023-08-18 | 2023-09-19 | 成都庆龙航空科技有限公司 | Unmanned aerial vehicle flight parameter self-diagnosis method and system |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110617966A (en) * | 2019-09-23 | 2019-12-27 | 江南大学 | Bearing fault diagnosis method based on semi-supervised generation countermeasure network |
| US20200110181A1 (en) * | 2018-10-04 | 2020-04-09 | The Boeing Company | Detecting fault states of an aircraft |
| CN112200244A (en) * | 2020-10-09 | 2021-01-08 | 西安交通大学 | Intelligent detection method for anomaly of aerospace engine based on hierarchical countermeasure training |
| CN112396088A (en) * | 2020-10-19 | 2021-02-23 | 西安交通大学 | Intelligent diagnosis method for mechanical fault of implicit excitation countertraining under small sample |
| US20210055719A1 (en) * | 2019-08-21 | 2021-02-25 | Hitachi, Ltd. | System for predictive maintenance using generative adversarial networks for failure prediction |
| CN112699952A (en) * | 2021-01-06 | 2021-04-23 | 哈尔滨市科佳通用机电股份有限公司 | Train fault image amplification method and system based on deep learning |
| CN113255078A (en) * | 2021-05-31 | 2021-08-13 | 南京信息工程大学 | Bearing fault detection method and device under unbalanced sample condition |
| CN113505477A (en) * | 2021-06-29 | 2021-10-15 | 西北师范大学 | Process industry soft measurement data supplementing method based on SVAE-WGAN |
| CN113569928A (en) * | 2021-07-13 | 2021-10-29 | 湖南工业大学 | Train running state detection data missing processing model and reconstruction method |
-
2021
- 2021-11-05 CN CN202111307993.3A patent/CN114169396B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200110181A1 (en) * | 2018-10-04 | 2020-04-09 | The Boeing Company | Detecting fault states of an aircraft |
| US20210055719A1 (en) * | 2019-08-21 | 2021-02-25 | Hitachi, Ltd. | System for predictive maintenance using generative adversarial networks for failure prediction |
| CN110617966A (en) * | 2019-09-23 | 2019-12-27 | 江南大学 | Bearing fault diagnosis method based on semi-supervised generation countermeasure network |
| CN112200244A (en) * | 2020-10-09 | 2021-01-08 | 西安交通大学 | Intelligent detection method for anomaly of aerospace engine based on hierarchical countermeasure training |
| CN112396088A (en) * | 2020-10-19 | 2021-02-23 | 西安交通大学 | Intelligent diagnosis method for mechanical fault of implicit excitation countertraining under small sample |
| CN112699952A (en) * | 2021-01-06 | 2021-04-23 | 哈尔滨市科佳通用机电股份有限公司 | Train fault image amplification method and system based on deep learning |
| CN113255078A (en) * | 2021-05-31 | 2021-08-13 | 南京信息工程大学 | Bearing fault detection method and device under unbalanced sample condition |
| CN113505477A (en) * | 2021-06-29 | 2021-10-15 | 西北师范大学 | Process industry soft measurement data supplementing method based on SVAE-WGAN |
| CN113569928A (en) * | 2021-07-13 | 2021-10-29 | 湖南工业大学 | Train running state detection data missing processing model and reconstruction method |
Non-Patent Citations (3)
| Title |
|---|
| ANDERS BOESEN LINDBO LARSEN ET AL.: "Autoencoding beyond pixels using a learned similarity metric", 《ARXIV:1512.09300V2》, 10 February 2016 (2016-02-10), pages 2 * |
| ILYA TOLSTIKHIN ET AL.: "MLP-Mixer: An all-MLP Architecture for Vision", 《ARXIV:2105.01601V4》, 11 July 2021 (2021-07-11), pages 1 - 16 * |
| 吴星原: "基于深度学习的草图分割算法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》, 15 September 2019 (2019-09-15), pages 4 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114963408A (en) * | 2022-05-25 | 2022-08-30 | 日照安泰科技发展有限公司 | Air conditioning system unbalanced fault diagnosis method based on data and feature enhancement |
| CN114963408B (en) * | 2022-05-25 | 2024-04-16 | 日照安泰科技发展有限公司 | Data and feature enhancement-based unbalanced fault diagnosis method for air conditioning system |
| CN116774719A (en) * | 2023-08-18 | 2023-09-19 | 成都庆龙航空科技有限公司 | Unmanned aerial vehicle flight parameter self-diagnosis method and system |
| CN116774719B (en) * | 2023-08-18 | 2023-11-07 | 成都庆龙航空科技有限公司 | Unmanned aerial vehicle flight parameter self-diagnosis method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114169396B (en) | 2022-09-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114169396B (en) | Training data generation model construction method and application for aircraft fault diagnosis | |
| Rad et al. | Surviving the barren plateau in variational quantum circuits with bayesian learning initialization | |
| CN114609994B (en) | Fault diagnosis method and device based on multi-granularity regularized rebalancing increment learning | |
| Huang et al. | 1DCNN fault diagnosis based on cubic spline interpolation pooling | |
| Lesort et al. | Continual learning in deep networks: an analysis of the last layer | |
| CN114116292B (en) | Hard disk fault prediction method fusing AP clustering and width learning system | |
| CN113537614A (en) | Construction method, system, equipment and medium of power grid engineering cost prediction model | |
| CN111177966B (en) | Guided missile structure uncertain load interval reconstruction method based on Bayesian theory | |
| CN118313265A (en) | Oil reservoir history fitting method, system, equipment and medium based on error learning double-agent model | |
| US11003823B2 (en) | Re-design of analog circuits | |
| Luo et al. | A novel method for remaining useful life prediction of roller bearings involving the discrepancy and similarity of degradation trajectories | |
| CN116205302A (en) | Task processing method and device based on quantum computation and ground state energy level determining method | |
| Guo et al. | Flight data visualization for simulation & evaluation: a general framework | |
| CN112419098B (en) | Power grid safety and stability simulation sample screening and expanding method based on safety information entropy | |
| CN116050300A (en) | Method and device for predicting aerodynamic thermal flow field of surface of high-speed moving body | |
| Hedayati et al. | Mitigating Data Scarcity for Satellite Reaction Wheel Fault Diagnosis with Wasserstein Generative Adversarial Networks | |
| CN109657795B (en) | Hard disk failure prediction method based on attribute selection | |
| Kong et al. | Activated parameter locating via causal intervention for model merging | |
| CN113821012B (en) | Fault diagnosis method for variable-working-condition satellite attitude control system | |
| CN116015787B (en) | Network intrusion detection method based on mixed continuous variable component sub-neural network | |
| Finkelstein et al. | Transfer learning promotes robust parametric mapping of diffusion encoded MR fingerprinting | |
| CA3077178A1 (en) | Apparatus and methods for improved subsurface data processing systems | |
| CN116822076A (en) | Method for monitoring moving bed separation process by using soft measurement modeling | |
| DENG et al. | A Few-Shot Learning Framework for Rotor Unbalance and Shaft Crack Fault Diagnostic Based on Physics-Informed Neural Network | |
| Zou et al. | Node Centrality Approximation For Large Networks Based On Inductive Graph Neural Networks |
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 |