CN118114020A - Equipment fault diagnosis method, system, computer equipment and storage medium - Google Patents
Equipment fault diagnosis method, system, computer equipment and storage medium Download PDFInfo
- Publication number
- CN118114020A CN118114020A CN202410227378.9A CN202410227378A CN118114020A CN 118114020 A CN118114020 A CN 118114020A CN 202410227378 A CN202410227378 A CN 202410227378A CN 118114020 A CN118114020 A CN 118114020A
- Authority
- CN
- China
- Prior art keywords
- fault diagnosis
- equipment
- model
- neural network
- layer
- 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
- 238000003745 diagnosis Methods 0.000 title claims abstract description 113
- 238000000034 method Methods 0.000 title claims abstract description 61
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 42
- 238000012549 training Methods 0.000 claims abstract description 34
- 238000013528 artificial neural network Methods 0.000 claims abstract description 32
- 238000000605 extraction Methods 0.000 claims abstract description 21
- 230000006870 function Effects 0.000 claims description 54
- 238000005096 rolling process Methods 0.000 claims description 39
- 238000004590 computer program Methods 0.000 claims description 19
- 238000012546 transfer Methods 0.000 claims description 13
- 230000015654 memory Effects 0.000 claims description 12
- 230000003044 adaptive effect Effects 0.000 claims description 10
- 230000000644 propagated effect Effects 0.000 claims description 6
- 238000010276 construction Methods 0.000 claims description 2
- 208000037408 Device failure Diseases 0.000 claims 2
- 230000007257 malfunction Effects 0.000 claims 1
- 238000009826 distribution Methods 0.000 abstract description 11
- 238000013508 migration Methods 0.000 description 12
- 230000005012 migration Effects 0.000 description 12
- 230000008569 process Effects 0.000 description 7
- 238000013526 transfer learning Methods 0.000 description 7
- 230000000694 effects Effects 0.000 description 6
- 238000013507 mapping Methods 0.000 description 6
- 238000010606 normalization Methods 0.000 description 6
- 238000012360 testing method Methods 0.000 description 6
- 235000014653 Carica parviflora Nutrition 0.000 description 5
- 241000243321 Cnidaria Species 0.000 description 5
- 238000002474 experimental method Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 210000002569 neuron Anatomy 0.000 description 4
- 238000012800 visualization Methods 0.000 description 4
- 230000004913 activation Effects 0.000 description 3
- 230000006978 adaptation Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 239000008186 active pharmaceutical agent Substances 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000009295 sperm incapacitation Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—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
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- 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
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- 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
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/096—Transfer learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention discloses a device fault diagnosis method, a system, computer equipment and a storage medium, and relates to the technical field of device fault diagnosis, wherein the method comprises the following steps: introducing a standardized layer between a feature extraction layer and a full connection layer of the one-dimensional convolutional neural network to construct a depth self-adaptive neural network; training the deep self-adaptive neural network according to historical operation data of the equipment to obtain a fault diagnosis model of the equipment; and carrying out fault diagnosis on the equipment to be diagnosed according to the fault diagnosis model. According to the invention, a standardized layer is introduced between the feature extraction layer and the full connection layer of the one-dimensional convolutional neural network, so that the distribution difference of the historical operation data of the equipment when the data enter the full connection layer through the feature extraction of the convolutional part is reduced, the data entering the full connection layer is approximately distributed, the generalization capability of a fault diagnosis model is improved, and more reliable fault diagnosis is realized.
Description
Technical Field
The present invention relates to the field of device fault diagnosis technologies, and in particular, to a device fault diagnosis method, a system, a computer device, and a storage medium.
Background
In the field of fault diagnosis of the rolling bearing of the oil transfer pump, technologies such as machine learning, deep learning and the like are generally adopted to diagnose faults. However, the existing method is only focused on the characteristics of the training data, and the generalization capability of the fault diagnosis model to other types of equipment data is ignored. This results in that in practical applications, the ability of the model to identify faults of the rolling bearing of the oil transfer pump under different working conditions is often not satisfactory for different models. Therefore, a generalization method for a fault diagnosis model is needed to solve the above-mentioned problem, so as to improve the generalization capability and efficiency of the fault diagnosis of the oil pump and realize more reliable diagnosis.
Disclosure of Invention
The invention aims to solve the technical problems of the prior art, and particularly provides a device fault diagnosis method, a system, computer equipment and a storage medium, wherein the method comprises the following steps:
1) In a first aspect, the present invention provides a method for diagnosing a device fault, which specifically includes the following steps:
introducing a standardized layer between a feature extraction layer and a full connection layer of the one-dimensional convolutional neural network to construct a depth self-adaptive neural network;
training the deep self-adaptive neural network according to historical operation data of the equipment to obtain a fault diagnosis model of the equipment;
and carrying out fault diagnosis on the equipment to be diagnosed according to the fault diagnosis model.
The equipment fault diagnosis method provided by the invention has the beneficial effects that:
And a standardized layer is introduced between the feature extraction layer and the full connection layer of the one-dimensional convolutional neural network, so that the distribution difference of the historical operation data of the equipment when the data enter the full connection layer through the feature extraction of the convolutional part is reduced, the data entering the full connection layer is approximately distributed, the generalization capability of a fault diagnosis model is improved, and more reliable fault diagnosis is realized.
On the basis of the scheme, the equipment fault diagnosis method can be improved as follows.
Further, training the deep adaptive neural network according to historical operation data of the device to obtain a fault diagnosis model of the device, including:
Based on the loss function, according to historical operation data of the equipment, the parameters of the depth self-adaptive neural network are propagated forward and updated in a reverse direction, and a fault diagnosis model of the equipment is obtained.
Further, the loss function is: a loss function obtained by combining covariance alignment with linear maximization of mean difference.
The beneficial effects of adopting the further scheme are as follows: the stability of accuracy in the migration process can be ensured.
Further, the device is a rolling bearing of an oil transfer pump.
2) In a second aspect, the present invention further provides an equipment fault diagnosis system, which has the following specific technical scheme:
the system comprises a model building module, a model training module and a fault diagnosis module;
The model building module is used for: introducing a standardized layer between a feature extraction layer and a full connection layer of the one-dimensional convolutional neural network to construct a depth self-adaptive neural network;
The model training module is used for: training the deep self-adaptive neural network according to historical operation data of the equipment to obtain a fault diagnosis model of the equipment;
The fault diagnosis module is used for: and carrying out fault diagnosis on the equipment to be diagnosed according to the fault diagnosis model.
On the basis of the scheme, the equipment fault diagnosis system can be improved as follows.
Further, the model training module is configured to:
Based on the loss function, according to historical operation data of the equipment, the parameters of the depth self-adaptive neural network are propagated forward and updated in a reverse direction, and a fault diagnosis model of the equipment is obtained.
Further, the loss function is: a loss function obtained by combining covariance alignment with linear maximization of mean difference.
Further, the device is a rolling bearing of an oil transfer pump.
3) In a third aspect, the present invention also provides a computer device, where the computer device includes a processor, and the processor is coupled to a memory, where at least one computer program is stored, where the at least one computer program is loaded and executed by the processor, so that the computer device implements any one of the above device fault diagnosis methods.
4) In a fourth aspect, the present invention also provides a computer readable storage medium, in which at least one computer program is stored, where the at least one computer program is loaded and executed by a processor, so that the computer implements any one of the above-mentioned fault diagnosis methods.
It should be noted that, the technical solutions of the second aspect to the fourth aspect and the corresponding possible implementation manners of the present invention may refer to the technical effects of the first aspect and the corresponding possible implementation manners of the first aspect, which are not described herein.
Drawings
Other features, objects and advantages of the present invention will become more apparent upon reading of the detailed description of non-limiting embodiments, made with reference to the following drawings in which:
FIG. 1 is a schematic flow chart of an apparatus fault diagnosis method according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a network architecture of a deep adaptive neural network;
FIG. 3 is a graph showing the variation of the loss when Relu functions are applied to a one-dimensional convolutional neural network (1D-CNN);
FIG. 4 is a load 0 feature visualization;
FIG. 5 is a load 1 feature visualization;
FIG. 6 is a load 2 feature visualization;
FIG. 7 is a load 3 feature visualization;
FIG. 8 is a normalized layer-based 1D-CNN model;
FIG. 9 is a block diagram of a deep domain adaptive network;
fig. 10 is a schematic structural diagram of an equipment fault diagnosis system according to an embodiment of the present invention;
Fig. 11 is a schematic structural diagram of a computer device according to an embodiment of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the present invention more apparent, the embodiments of the present invention will be described in further detail with reference to the accompanying drawings.
As shown in fig. 1, a device fault diagnosis method according to an embodiment of the present invention includes the following steps:
S1, introducing a standardized layer between a feature extraction layer and a full connection layer of a one-dimensional convolutional neural network, and constructing a 1D-CNN rolling bearing fault diagnosis model introduced with the standardized layer;
the rolling bearing fault diagnosis and migration learning method using the domain self-adaptive method combining linear MMD and core loss functions is used for the 1D-CNN in combination with the 1D-CNN and depth self-adaptive neural network structure introducing the standardized layer as shown in figure 2.
S2, training the deep self-adaptive neural network according to historical operation data of the equipment to obtain a fault diagnosis model of the equipment;
s3, performing fault diagnosis on the equipment to be diagnosed according to the fault diagnosis model.
Optionally, in S2, training the depth adaptive neural network according to the historical operation data of the device to obtain a fault diagnosis model of the device, including:
S20, based on the loss function, according to historical operation data of the equipment, the depth self-adaptive neural network is propagated forwards, and parameters of the depth self-adaptive neural network are updated in a back propagation mode, so that a fault diagnosis model of the equipment is obtained.
Optionally, in the above technical solution, the loss function is: a loss function obtained by combining covariance alignment with linear maximization of mean difference.
Optionally, in the above technical solution, the device is a rolling bearing of an oil transfer pump, and the device to be diagnosed for failure is a rolling bearing of an oil transfer pump to be diagnosed for failure.
Taking the rolling bearing of the oil transfer pump as equipment and the historical operation data of the equipment as 'the historical vibration data of the rolling bearing of the oil transfer pump', the following description is given to the invention:
S10, constructing a 1D-CNN rolling bearing fault diagnosis model introducing a standardized layer:
A standardized layer is introduced between a feature extraction layer and a full connection layer of a one-dimensional convolutional neural network (1D-CNN), a 1D-CNN rolling bearing fault diagnosis model introduced with the standardized layer is constructed, rolling bearing vibration data of an original oil transfer pump is used as input of the model, a fault diagnosis model for carrying out fault diagnosis on the rolling bearing of the oil transfer pump is obtained through training, the 1D-CNN rolling bearing fault diagnosis model introduced with the standardized layer can be obtained through training, steps of constructing feature engineering are reduced, and compared with the process of taking extracted signal features as input of the fault diagnosis model, the accuracy rate can reach 0.943 when the fault diagnosis of the rolling bearing is carried out, and the accuracy rate is higher than that of the rolling bearing fault diagnosis model constructed based on the feature engineering.
When Relu functions are used as the activation functions for training the 1D-CNN rolling bearing fault diagnosis model, the condition of up-down vibration is lost in the experimental process, and sometimes the condition of incapacitation occurs, as shown in fig. 3, wherein the horizontal axis represents training times, and the vertical axis represents corresponding loss. At this time, the problem of low accuracy of the trained fault diagnosis model occurs. According to analysis, the result is that the Relu function cuts off neurons with input data smaller than 0, so that the gradient becomes 0 in the diagnosis experiment process of the 1D-CNN rolling bearing fault diagnosis model. The waveform data vibrates up and down between positive and negative numbers, and when the waveform data is subjected to convolution and pooling operation, the values of some neurons become 0, so that the weight of the whole network is not updated, the whole loss of the network cannot be converged, and the ideal effect cannot be achieved.
For the above reasons, the invention introduces PRelu functions for the activation function of the 1D-CNN rolling bearing fault diagnosis model, and PRelu functions are specifically shown as formula 1 below.
Where x represents the input of the activation function, it can be seen from the above equation that the PRelu function does not directly set the data that is a neuron to 0 when the input data is less than 0, but continues to retain this information by multiplying it by a smaller weight. Therefore, the problem that when the input data is a non-positive number, the neuron is 0 and the network weight cannot be updated can be avoided. In PRelu, the parameter a is learnable, typically between 0 and 1.
The model generalization ability test was performed using a 1D-CNN rolling bearing fault diagnosis model without the introduction of a normalization layer, and the results are shown in table 1. Experiments are carried out on bearing data sets of the Western university by using data of different working conditions, and the generalization capability of the model is enhanced by adopting a Dropout mode. As can be seen from table 1, the data trained with the data set was significantly less effective at the other data sets than at the training data, which was only 0.75 at maximum.
Table 1:
| Data set | D-0 | D-0.1 | D-0.5 | D-0.8 |
| 1 | 1.0 | 1.0 | 1.0 | 1.0 |
| 2 | 0.667 | 0.638 | 0.75 | 0.667 |
| 3 | 0.372 | 0.56 | 0.432 | 0.45 |
| 4 | 0.505 | 0.509 | 0.672 | 0.506 |
| 5 | 0.202 | 0.332 | 0.23 | 0.192 |
| 6 | 0.491 | 0.332 | 0.33 | 0.254 |
| 7 | 0.195 | 0.321 | 0.17 | 0.178 |
| 8 | 0.257 | 0.217 | 0.45 | 0.447 |
| 9 | 0.235 | 0.17 | 0.24 | 0.237 |
| 10 | 0.316 | 0.16 | 0.24 | 0.237 |
| 11 | 0.202 | 0.148 | 0.3 | 0.314 |
| 12 | 0.15 | 0.237 | 0.15 | 0.15 |
| 13 | 0.16 | 0.181 | 0.15 | 0.15 |
| 14 | 0.171 | 0.194 | 0.20 | 0.216 |
In order to analyze the reason that the generalization capability of the model is not high, the method collects data after feature extraction in a 1D-CNN rolling bearing fault diagnosis model introduced into a standardized layer, and carries out T-sne visual analysis, and by taking fault data with the sampling frequency of 12K and the fault depth of 0.1778mm as an example, the obtained results are shown in fig. 4 to 7.
As can be seen from fig. 4 to 7, the clustered results after feature extraction show that at this time, the distribution of data is substantially the same and can be aggregated according to substantially different categories. This means that the data can be distinguished before entering the fully connected layer, whereas after entering the fully connected layer the generalization capability is not high due to the existence of the differences. Therefore, the invention introduces a standardized layer between the feature extraction layer and the full connection layer of the one-dimensional convolutional neural network, constructs a 1D-CNN rolling bearing fault diagnosis model introduced with the standardized layer, and converts between the feature extraction layer and the full connection layer of the 1D-CNN, so that the data sent into the full connection layer can be better close to training data, thereby improving the generalization capability of the model. The network structure of the 1D-CNN introduced into the normalization layer is shown in fig. 8, and the calculation formula when the normalization layer performs normalization is a conventional data normalization formula, namely formula 2:
in order to test the generalization capability of the proposed 1D-CNN rolling bearing fault diagnosis model with the introduction of the standardized layer, dropout techniques commonly used in the generalization of the current model are used as references, and Dropout rates of 0, 0.1, 0.5 and 0.8 are respectively set for comparison with the proposed model. After training the 1D-CNN model, the other data sets were tested, and the test results of the generalization ability of the 1D-CNN rolling bearing fault diagnosis model with the standardized layer introduced are shown in Table 2.
Table 2:
In Table 2, columns 2-5 are experimental results using different Dropout rates using the original 1D-CNN method, respectively, and column 6 is experimental results of the 1D-CNN failure diagnosis model incorporating the standardized layer as set forth herein. It can be seen that the present method shows higher accuracy than the original method on the data sets 1,2,3,4,5, 7, 10, 11, 12. This illustrates that a method of inserting a normalization layer between the feature extraction layer and the full connection layer of the 1D-CNN fault diagnosis model to improve the generalization ability of the model is possible.
S11, a rolling bearing fault diagnosis method introducing transfer learning is adopted:
When the 1D-CNN rolling bearing fault diagnosis model introduced with the standardized layer is used for carrying out model generalization capability test, the effect performance of the model under the same working condition is found to be far better than that of data in different working conditions. For example, using data from one load training, the accuracy is 100% when tested under the same load, but the highest accuracy is only 90.2% when tested under different loads. Therefore, the method further provides a rolling bearing fault diagnosis method introducing transfer learning to solve the model generalization capability of different loads and equipment.
Domain adaptation is a common method in transfer learning. In the migration learning, the domain with the existing label and data is called a source domain, and the domain with only the data and no label is called a target domain. Is generally defined as: given a source domain D s and a target domain D t, a learning task T s on the source domain and a learning task T t on the target domain are respectively carried out, and a model trained on the source domain still keeps a good effect on the target domain through a certain algorithm by utilizing the learning tasks T s of the source domain D s and the source domain.
The domain adaptive neural network (Domain Adaptive Neural Network, daNN) is the first to appear in domain adaptation, as shown in fig. 9, mainly two losses: the classification loss and the domain loss are used for reducing the distance between two features learned by the deep neural network, and the classification domain can ensure the performance of the source domain classifier, so that the representation with discriminant and domain invariance can be simultaneously learned. Compared with the traditional deep neural network, daNN introduces a layer of distance calculation for measuring the difference between the source domain and the target domain after the feature extraction layer, introduces a regularization term in the loss function, and takes the distance between the source domain and the target domain as the regularization term to form a new loss function for training the network. The rolling bearing plays a role in balancing force in the oil transfer pump, and different loads have influence on the accuracy of the fault diagnosis model. In FIGS. 4 to 7, the data after 1D-CNN feature extraction was clustered and found to be of a distinct class. Finally, when the full connection layer is entered, the final model generalization capability is not high due to the feature distribution difference of the training data and the test data. In order to improve model generalization, the most effective method is to perform feature distribution transformation when data enters the fully connected layer, so that the distribution of test data and training data is as similar as possible. DaNN by mapping the source domain and target domain data to Gao Weizi space, the distribution difference of the two domains is made as small as possible in the space, so that the generalization capability of the model on the target data set is improved. Therefore, in theory, it is possible to use DaNN for migration of the rolling bearing failure diagnosis model.
In the transfer learning, the Maximum mean difference (Maximum MEAN DISCREPANCY, MMD) is the most widely used function for measuring the difference between two distributions, and is mainly used for measuring the distance between two distributions. The measurement is performed by mapping the source domain data and the target domain data into the regenerated kernel hilbert space (Reproducing kernel Hilbert space, RKHS), and the calculation formula is shown in formula 3:
in the actual program implementation process, core skills similar to those in SVM are generally adopted, source domain data and target domain data are mapped to a high-dimensional space through a core function, the difference between the source domain data and the target domain data is calculated by utilizing a distance formula, and finally a matrix representing the difference between the source domain and the target domain is returned. The common kernel functions are shown in table 3.
Table 3:
The high latitude space to which different kernel function mapping data is mapped is different, so that the result of the final model changes with the change of the kernel function. Gaussian kernel functions are one of the most widespread types of kernel functions in machine learning and transfer learning applications. Therefore, the gaussian kernel function is widely used in the field of computer vision and natural language processing at present, and good effect is obtained.
Another approach used more in domain adaptation is statistical feature based associative alignment (Corelation Alignment, core). Its goal is to minimize the second order statistics of the source and target domains, the formula is equation 4:
Wherein:
When the MMD and the coral method are respectively applied to the migration of the rolling bearing fault diagnosis model introducing the 1D-CNN, the phenomenon of unstable accuracy exists, namely, the accuracy of MMD adopted on certain data sets is higher, and the accuracy of the coral method adopted on certain data sets is higher. Therefore, the method provides a method for combining MMD and coral, and constructs a loss function, specifically, a loss function formula obtained by combining covariance alignment and linear maximization mean difference is formula 7:
ξ＝ξC(DS,yS)+λ(MMD2(Ds,Dt)+coral(Ds,Dt)) (7)
Wherein D s is source domain data, y s is a corresponding source domain data tag, D t is target domain data, which is devoid of tags, ζ C is a network training loss function, MMD is a linear kernel function, core is a covariance versus function thereof, and λ is a hyper-parameter between 0 and 1. Firstly, source domain data and target domain data are subjected to linear kernel mapping, differences of distribution of the source domain data and the target domain data are calculated, then covariance alignment is adopted, distances of the source domain data and the target domain data under second-order statistical characteristics are calculated, the two differences are added to serve as final migration loss, the final migration loss is participated in a final model training process, and the complexity problem after Gaussian kernel function mapping and the unstable accuracy rate of the Gaussian kernel function mapping and the Gaussian kernel function are used independently are compensated.
The rolling bearing fault diagnosis migration experiment is carried out by selecting four data under different loads and carrying out four data with the numbers of 0,1,2 and 3. By constructing the depth adaptive neural network, the common migration learning loss functions such as linear kernel MMD (MMD), core, gaussian kernel MMD (rbf), and polynomial kernel MMD (Mmd) are respectively compared with DaNN neural networks. The overall experimental results are shown in tables 4 to 7. Tables 4 to 7 show the accuracy of the respective loss functions when the load numbers 0,1,2,3 were shifted from each other. Table 8 summarizes the final results for each experiment as well as the average results over the entire dataset.
Table 4:
Table 5:
Table 6:
table 7:
Table 8:
the experimental data shows that when the model migration is performed between data used for working conditions, compared with other loss functions, the accuracy is obviously improved, the accuracy of the model migration is 97.2% on the overall average effect, the accuracy is higher than that of the model migration adopting other loss functions, the average accuracy of the model on a data set is improved, and the effectiveness of the method is illustrated.
The invention discloses a fault diagnosis method, which is applied to fault diagnosis of an oil delivery pump by using a fault diagnosis model generalization method combining 1D-CNN and a depth self-adaptive neural network. Aiming at the characteristics of the training data, but neglecting the recognition capability of the fault diagnosis model on other types of equipment data, the method enhances the generalization capability of the fault diagnosis model, and can carry out more accurate fault diagnosis on equipment of different types under different working conditions. According to the method, a standardized layer is added between the characteristic extraction layer of the 1D-CNN and the full connection layer, so that the distribution difference of the rolling bearing data when the rolling bearing data enter the full connection through the characteristic extraction of the convolution part is reduced, and the 1D-CNN rolling bearing fault model can have better generalization capability; the field self-adaptive method introduced into the transfer learning provides a loss function combining linear MMD and coral for the rolling bearing fault diagnosis transfer learning of 1D-CNN. The effectiveness of the proposed method is demonstrated by experiments performed on the relevant dataset.
In the above embodiments, although steps S1, S2, etc. are numbered, only specific embodiments of the present invention are given, and those skilled in the art may adjust the execution sequence of S1, S2, etc. according to the actual situation, which is also within the scope of the present invention, and it is understood that some embodiments may include some or all of the above embodiments.
As shown in fig. 10, a device fault diagnosis system 200 according to an embodiment of the present invention includes a model construction module 201, a model training module 202, and a fault diagnosis module 203;
The model building module 201 is configured to: introducing a standardized layer between a feature extraction layer and a full connection layer of the one-dimensional convolutional neural network to construct a depth self-adaptive neural network;
model training module 202 is configured to: training the deep self-adaptive neural network according to historical operation data of the equipment to obtain a fault diagnosis model of the equipment;
the fault diagnosis module 203 is configured to: and carrying out fault diagnosis on the equipment to be diagnosed according to the fault diagnosis model.
Optionally, in the above technical solution, the model training module 202 is configured to:
Based on the loss function, according to historical operation data of the equipment, the parameters of the depth self-adaptive neural network are propagated forward and updated in a reverse direction, and a fault diagnosis model of the equipment is obtained.
Optionally, in the above technical solution, the loss function is: a loss function obtained by combining covariance alignment with linear maximization of mean difference.
Optionally, in the above technical solution, the device is a rolling bearing of an oil transfer pump.
It should be noted that, the beneficial effects of the device fault diagnosis system 200 provided in the above embodiment are the same as those of the device fault diagnosis method described above, and will not be described herein again. In addition, when the system provided in the above embodiment implements the functions thereof, only the division of the above functional modules is used as an example, in practical application, the above functional allocation may be implemented by different functional modules according to needs, that is, the system is divided into different functional modules according to practical situations, so as to implement all or part of the functions described above. In addition, the system and method embodiments provided in the foregoing embodiments belong to the same concept, and specific implementation processes thereof are detailed in the method embodiments and are not described herein again.
As shown in fig. 11, in a computer device 300 according to an embodiment of the present invention, the computer device 300 includes a processor 320, the processor 320 is coupled to a memory 310, at least one computer program 330 is stored in the memory 310, and the at least one computer program 330 is loaded and executed by the processor 320, so that the computer device 300 implements any one of the device fault diagnosis methods described above, specifically:
The computer device 300 may include one or more processors 320 (Central Processing Units, CPU) and one or more memories 310, where the one or more memories 310 store at least one computer program 330, where the at least one computer program 330 is loaded and executed by the one or more processors 320 to enable the computer device 300 to implement any of the device fault diagnosis methods provided by the embodiments above. Of course, the computer device 300 may also have a wired or wireless network interface, a keyboard, an input/output interface, and other components for implementing the functions of the device, which are not described herein.
The embodiment of the invention provides a computer readable storage medium, wherein at least one computer program is stored in the computer readable storage medium, and the at least one computer program is loaded and executed by a processor, so that the computer realizes any one of the equipment fault diagnosis methods.
Alternatively, the computer readable storage medium may be a Read-Only Memory (ROM), a random access Memory (Random Access Memory, RAM), a compact disc Read-Only Memory (CD-ROM), a magnetic tape, a floppy disk, an optical data storage device, and the like.
In an exemplary embodiment, a computer program product or a computer program is also provided, the computer program product or computer program comprising computer instructions stored in a computer readable storage medium. A processor of a computer device reads the computer instructions from a computer-readable storage medium, and the processor executes the computer instructions so that the computer device performs any of the above-described device fault diagnosis methods.
It should be noted that the terms "first," "second," and the like in the description and in the claims are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. The order of use of similar objects may be interchanged where appropriate such that embodiments of the application described herein may be implemented in other sequences than those illustrated or otherwise described.
Those skilled in the art will appreciate that the invention may be embodied as a system, method or computer program product, and that the invention may therefore be embodied in the form of: either entirely hardware, entirely software (including firmware, resident software, micro-code, etc.), or entirely software, or a combination of hardware and software, referred to herein generally as a "circuit," module "or" system. Furthermore, in some embodiments, the invention may also be embodied in the form of a computer program product in one or more computer-readable media, which contain computer-readable program code.
Any combination of one or more computer readable media may be employed. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. The computer readable storage medium can be, for example, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of the computer-readable storage medium would include the following: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this document, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.
While embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and not to be construed as limiting the invention, and that variations, modifications, alternatives and variations may be made to the above embodiments by one of ordinary skill in the art within the scope of the invention.
Claims (10)
1. A device failure diagnosis method, characterized by comprising:
introducing a standardized layer between a feature extraction layer and a full connection layer of the one-dimensional convolutional neural network to construct a depth self-adaptive neural network;
training the deep adaptive neural network according to historical operation data of equipment to obtain a fault diagnosis model of the equipment;
and carrying out fault diagnosis on the equipment to be diagnosed according to the fault diagnosis model.
2. The device fault diagnosis method according to claim 1, wherein training the deep adaptive neural network according to the historical operation data of the device to obtain a fault diagnosis model of the device comprises:
Based on the loss function, according to historical operation data of the equipment, the parameters of the depth self-adaptive neural network are propagated forward and updated in a reverse direction, and a fault diagnosis model of the equipment is obtained.
3. The apparatus fault diagnosis method according to claim 2, wherein the loss function is: a loss function obtained by combining covariance alignment with linear maximization of mean difference.
4. A method of diagnosing a malfunction of an apparatus according to any one of claims 1 to 3, wherein the apparatus is a rolling bearing of an oil delivery pump.
5. The equipment fault diagnosis system is characterized by comprising a model construction module, a model training module and a fault diagnosis module;
The model building module is used for: introducing a standardized layer between a feature extraction layer and a full connection layer of the one-dimensional convolutional neural network to construct a depth self-adaptive neural network;
The model training module is used for: training the deep adaptive neural network according to historical operation data of equipment to obtain a fault diagnosis model of the equipment;
The fault diagnosis module is used for: and carrying out fault diagnosis on the equipment to be diagnosed according to the fault diagnosis model.
6. The device fault diagnosis system of claim 5, wherein the model training module is configured to:
Based on the loss function, according to historical operation data of the equipment, the parameters of the depth self-adaptive neural network are propagated forward and updated in a reverse direction, and a fault diagnosis model of the equipment is obtained.
7. The equipment failure diagnosis system according to claim 6, wherein the loss function is: a loss function obtained by combining covariance alignment with linear maximization of mean difference.
8. An equipment failure diagnosis system according to any of claims 5 to 7, wherein the equipment is a rolling bearing of an oil transfer pump.
9. A computer device, characterized in that it comprises a processor coupled to a memory, in which at least one computer program is stored, which is loaded and executed by the processor, in order to make it implement a device failure diagnosis method according to any of claims 1 to 4.
10. A computer-readable storage medium, in which at least one computer program is stored, the at least one computer program being loaded and executed by a processor to cause the computer to implement a device fault diagnosis method as claimed in any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410227378.9A CN118114020A (en) | 2024-02-29 | 2024-02-29 | Equipment fault diagnosis method, system, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410227378.9A CN118114020A (en) | 2024-02-29 | 2024-02-29 | Equipment fault diagnosis method, system, computer equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN118114020A true CN118114020A (en) | 2024-05-31 |
Family
ID=91214961
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410227378.9A Pending CN118114020A (en) | 2024-02-29 | 2024-02-29 | Equipment fault diagnosis method, system, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118114020A (en) |
-
2024
- 2024-02-29 CN CN202410227378.9A patent/CN118114020A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111898095A (en) | Deep migration learning intelligent fault diagnosis method and device, storage medium and equipment | |
| CN110926782B (en) | Circuit breaker fault type judgment method and device, electronic equipment and storage medium | |
| CN111340238A (en) | Fault diagnosis method, device, equipment and storage medium of industrial system | |
| CN113923104B (en) | Network fault diagnosis method, equipment and storage medium based on wavelet neural network | |
| CN111562108A (en) | Rolling bearing intelligent fault diagnosis method based on CNN and FCMC | |
| CN113435509B (en) | Small sample scene classification and identification method and system based on meta-learning | |
| CN108919059A (en) | A kind of electric network failure diagnosis method, apparatus, equipment and readable storage medium storing program for executing | |
| CN112433896B (en) | Method, device, equipment and storage medium for predicting server disk faults | |
| CN113743528A (en) | Voltage source inverter fault diagnosis method and system based on convolutional neural network | |
| CN114609994B (en) | Fault diagnosis method and device based on multi-granularity regularized rebalancing increment learning | |
| CN117556369B (en) | Power theft detection method and system for dynamically generated residual error graph convolution neural network | |
| CN115791174B (en) | Rolling bearing abnormality diagnosis method, system, electronic equipment and storage medium | |
| CN113203914A (en) | Underground cable early fault detection and identification method based on DAE-CNN | |
| CN115112372A (en) | Bearing fault diagnosis method and device, electronic equipment and storage medium | |
| CN114638384A (en) | Fault diagnosis method and system based on machine learning | |
| CN116595465A (en) | High-dimensional sparse data outlier detection method and system based on self-encoder and data enhancement | |
| CN111027591A (en) | Node fault prediction method for large-scale cluster system | |
| CN114441173A (en) | Rolling bearing fault diagnosis method based on improved depth residual shrinkage network | |
| CN117741336A (en) | Fault line selection method and device for composite grounding system | |
| CN117591950A (en) | Rolling bearing variable working condition fault diagnosis method, device, terminal and storage medium | |
| CN118114020A (en) | Equipment fault diagnosis method, system, computer equipment and storage medium | |
| CN114202110B (en) | Service fault prediction method and device based on RF-XGBOOST | |
| CN115169740A (en) | Sequence prediction method and system of pooled echo state network based on compressed sensing | |
| CN113723593A (en) | Load shedding prediction method and system based on neural network | |
| CN113051809A (en) | Virtual health factor construction method based on improved restricted Boltzmann machine |
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 |