CN112836604A - Rolling bearing fault diagnosis and classification method, system and equipment based on VMD-SSAE and storage medium thereof - Google Patents

Rolling bearing fault diagnosis and classification method, system and equipment based on VMD-SSAE and storage medium thereof Download PDF

Info

Publication number
CN112836604A
CN112836604A CN202110087518.3A CN202110087518A CN112836604A CN 112836604 A CN112836604 A CN 112836604A CN 202110087518 A CN202110087518 A CN 202110087518A CN 112836604 A CN112836604 A CN 112836604A
Authority
CN
China
Prior art keywords
ssae
vmd
classification
wolf
classification model
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
Application number
CN202110087518.3A
Other languages
Chinese (zh)
Inventor
陈剑
张磊
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hefei University of Technology
Original Assignee
Hefei University of Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Hefei University of Technology filed Critical Hefei University of Technology
Priority to CN202110087518.3A priority Critical patent/CN112836604A/en
Publication of CN112836604A publication Critical patent/CN112836604A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/12—Classification; Matching
    • 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
    • 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
    • G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/08—Feature extraction

Abstract

The invention provides a rolling bearing fault diagnosis and classification method based on VMD-SSAE, which comprises the following steps: collecting vibration signals of rolling bearings with different fault types; based on variation modal decomposition, extracting time domain, frequency domain and time-frequency domain characteristics of the vibration signal; forming the features into a data set, and dividing the data set into a training set, a verification set and a test set; connecting a stack sparse self-encoder with a Softmax classifier, constructing a VMD-SSAE classification model, and training the VMD-SSAE classification model by using the training set; optimizing the VMD-SSAE classification model by adopting a wolf optimization algorithm and an error back propagation algorithm to obtain an ideal VMD-SSAE classification model; and inputting the data in the test set into an ideal VMD-SSAE classification model to obtain a diagnosis classification result. The rolling bearing fault diagnosis and classification method based on VMD-SSAE provided by the invention can accurately diagnose the fault type of the rolling bearing.

Description

Rolling bearing fault diagnosis and classification method, system and equipment based on VMD-SSAE and storage medium thereof
Technical Field
The invention relates to the technical field of rolling bearing fault diagnosis and classification, in particular to a rolling bearing fault diagnosis and classification method, system and equipment based on VMD-SSAE and a storage medium thereof.
Background
The rolling bearing is widely applied in the industrial field, is one of key parts which are extremely easy to damage in rotary machinery, is easy to generate fatigue and failure under the severe working environment with high speed and heavy load for a long time, thereby affecting the safe and stable operation of the whole equipment, and can cause huge social and economic losses and even casualties when serious failure occurs. The traditional equipment health monitoring needs to consume a large amount of manpower and material resources, and real-time monitoring is difficult to realize, so that the real-time monitoring and fault diagnosis of the rolling bearing are of great significance.
Vibration signals are important carriers of the operation state of mechanical equipment, and how to extract valuable characteristic information from complex vibration signals for evaluating the health condition of the equipment is always the focus of research in the field of fault diagnosis. With the rise of artificial intelligence, an intelligent fault diagnosis algorithm plays an increasingly important role in the field of fault diagnosis, although the existing machine learning method can identify the type of a bearing fault according to characteristics extracted manually, most of neural networks are of a shallow structure, the ability of learning higher-level and more abstract information from input is limited, and a large amount of test data is needed for training the networks. The limitation of the shallow structure can reduce the accuracy of fault diagnosis, and training a large amount of data on the network not only increases the workload of test data acquisition but also increases the time for network training. In addition, when the bearing normally works, the bearing is in a state that the bearing constantly changes, namely, the bearing is constantly in different load and variable rotating speed states. Therefore, a bearing fault diagnosis method with good diagnosis effect and robustness under different load and variable rotating speed states is needed.
As a deep learning method, the stack sparse autoencoder obtains the most abstract and essential characteristics of input data by extracting layer by layer, and has the advantages of strong self-adaption, strong robustness, good data fault tolerance and the like. In practical application, however, the sparse penalty factor of the relatively ideal SSAE network can be obtained only through tedious manual debugging and a large amount of comparative analysis, which increases the workload of debugging personnel to a great extent and also limits the fault diagnosis capability of the model. Therefore, a method is needed for adaptively selecting the sparse penalty factor of the SSAE network, so as to get rid of the tedious manual parameter adjustment work and achieve a good fault diagnosis effect.
Therefore, in order to solve the problem that the fault type of the rolling bearing is difficult to rapidly and accurately judge under time-varying non-stationary load and different bearing test data volume in the prior art, the invention designs a rolling bearing fault diagnosis and classification method, system, equipment and storage medium based on VMD-SSAE.
Disclosure of Invention
In view of the above drawbacks of the prior art, an object of the present invention is to provide a rolling bearing fault diagnosis and classification method, system, device and storage medium thereof based on VMD-SSAE, for solving the problem in the prior art that it is difficult to quickly and accurately determine the type of a bearing fault under time-varying non-stationary load and different bearing test data amounts.
In order to achieve the above objects and other related objects, the present invention provides a rolling bearing fault diagnosis and classification method based on VMD-SSAE, comprising the steps of:
collecting vibration signals of rolling bearings with different fault types;
based on variation modal decomposition, extracting time domain, frequency domain and time-frequency domain characteristics of the vibration signal;
forming the features into a data set, and dividing the data set into a training set, a verification set and a test set;
connecting a stack sparse self-encoder with a Softmax classifier, constructing a VMD-SSAE classification model, and training the VMD-SSAE classification model by using the training set;
optimizing the VMD-SSAE classification model by adopting a wolf optimization algorithm and an error back propagation algorithm to obtain an ideal VMD-SSAE classification model;
and inputting the data in the test set into an ideal VMD-SSAE classification model to obtain a diagnosis classification result.
In an embodiment of the present invention, the acquiring the vibration signals of the rolling bearings with different failure types includes: and acquiring the vibration signal of the rolling bearing under different loads and rotating speeds.
In an embodiment of the present invention, the extracting the time domain, the frequency domain, and the time-frequency domain features of the vibration signal based on the variational modal decomposition includes:
extracting time domain characteristics of the vibration signals to obtain time domain characteristic data;
extracting frequency domain characteristics of the vibration signals to obtain frequency domain characteristic data;
and adopting the variation modal decomposition to extract the time-frequency domain characteristics of the vibration signals to obtain time-frequency domain characteristic data.
In an embodiment of the present invention, the obtaining the time-frequency domain characteristic data includes:
adopting the variation modal decomposition to carry out self-adaptive decomposition on the vibration signal to obtain a plurality of inherent modal functions;
and extracting the time-frequency domain characteristics of each inherent mode function to obtain the time-frequency domain characteristic data.
In an embodiment of the present invention, the optimizing the VMD-SSAE classification model by using the grayish optimization algorithm and the error back propagation algorithm to obtain the ideal VMD-SSAE classification model includes the steps of:
adopting the gray wolf optimization algorithm to perform self-adaptive selection on the sparse penalty factor of the sparse self-encoder so as to obtain the optimized sparse penalty factor in the VMD-SSAE classification model;
and fine-tuning the optimized VMD-SSAE classification model by adopting the error back propagation algorithm to obtain an ideal VMD-SSAE classification model.
In an embodiment of the present invention, the adaptively selecting the sparse penalty factor of the sparse self-encoder by using the gray wolf optimization algorithm includes:
s511, initializing basic parameters of the gray wolf optimization algorithm, wherein the basic parameters comprise: the number of the grey wolves and the number of iterations, wherein the number of the grey wolves is more than three;
s512, calculating the fitness value of the gray wolf;
s513, sequentially determining the first three optimal wolfs according to the fitness value;
s514, updating the positions of other gray wolves according to the positions of the first three optimal wolves;
s515, adding one to the iteration times, and calculating the fitness value of the current wolf;
s516, comparing the fitness value of the current gray wolf with the fitness value of the previous generation gray wolf, and selecting the minimum value of the fitness values as the minimum fitness value;
s517, judging whether a termination condition is reached, if so, taking the position of the wolf corresponding to the minimum fitness value as an optimal sparse penalty factor of the sparse self-encoder, otherwise, returning to the step S513, and continuously updating the position of the wolf; wherein the termination condition comprises: the minimum fitness value is smaller than a preset fitness threshold value.
In an embodiment of the present invention, the step of sequentially determining the first three optimal wolfs according to the fitness value includes: the grey wolf with the minimum fitness value is used as the first optimal wolf, and the grey wolf with the second-order fitness value is used as the second optimal wolf and the third optimal wolf in sequence.
The invention also provides a rolling bearing fault diagnosis and classification system based on VMD-SSAE, which comprises:
the signal acquisition module is used for acquiring vibration signals of rolling bearings with different fault types;
the characteristic extraction module is used for extracting time domain, frequency domain and time-frequency domain characteristics of the vibration signal based on variational modal decomposition;
the data set construction module is used for forming the features into a data set and dividing the data set into a training set, a verification set and a test set;
the model building module is used for connecting the stack sparse self-encoder with a Softmax classifier to form a VMD-SSAE classification model and training the VMD-SSAE classification model by using the training set;
the model optimization module is used for optimizing the VMD-SSAE classification model by adopting a gray wolf optimization algorithm and an error back propagation algorithm so as to obtain an ideal VMD-SSAE classification model;
and the fault diagnosis module is used for inputting the data in the test set into the VMD-SSAE classification model to obtain a diagnosis classification result.
In an embodiment of the present invention, a rolling bearing fault diagnosis and classification apparatus based on VMD-SSAE includes: a processor coupled to a memory, the memory storing program instructions that, when executed by the processor, implement the rolling bearing fault diagnosis classification method as described above.
In an embodiment of the present invention, a computer-readable storage medium includes: comprises a program which, when run on a computer, causes the computer to execute the rolling bearing fault diagnosis classification method as described above.
As described above, the rolling bearing fault diagnosis and classification method, system, equipment and storage medium thereof based on VMD-SSAE provided by the invention can diagnose the fault type of the rolling bearing more accurately under time-varying non-stationary load and different bearing test data volume; secondly, the invention can adaptively perform optimal selection on the sparse penalty factor of the stack sparse self-encoder network, thereby not only reducing the complex workload, but also avoiding the subjective influence of artificially given parameters; the VMD-SSAE classification model also has better fault tolerance and robustness, so that better fault diagnosis effect can be achieved.
Drawings
FIG. 1 is a schematic flow chart of a rolling bearing fault diagnosis and classification method based on VMD-SSAE provided by the invention.
FIG. 2 is a time domain diagram of vibration signals of a normal bearing and an outer ring rolling element composite type fault bearing in the embodiment of the invention.
Fig. 3 is a flowchart illustrating step S2 according to an embodiment of the present invention.
Fig. 4 is a flowchart illustrating step S23 according to an embodiment of the present invention.
Fig. 5 is a VMD exploded view of a vibration signal of a normal bearing and outer ring rolling element compound type faulty bearing in the embodiment of the present invention.
FIG. 6 is a schematic structural diagram of an auto-encoder according to an embodiment of the present invention.
Fig. 7 is a schematic structural diagram of the connection between the SSAE network and the Softmax classifier in the embodiment of the present invention.
Fig. 8 is a flowchart illustrating step S51 according to an embodiment of the present invention.
Fig. 9 is a diagram of the result of the fault diagnosis classification of the model under the steady load in the embodiment of the present invention.
FIG. 10 is a line graph illustrating diagnostic classification accuracy for different models in an embodiment of the present invention.
FIG. 11 is a schematic structural diagram of a rolling bearing fault diagnosis and classification system based on VMD-SSAE provided by the invention.
Description of the element reference numerals
11 signal acquisition module 14 model construction module
12 feature extraction module 15 model optimization module
13 data set construction module 16 fault diagnosis module
Detailed Description
The embodiments of the present invention are described below with reference to specific embodiments, and other advantages and effects of the present invention will be easily understood by those skilled in the art from the disclosure of the present specification. The invention is capable of other and different embodiments and of being practiced or of being carried out in various ways, and its several details are capable of modification in various respects, all without departing from the spirit and scope of the present invention. It is to be noted that the features in the following embodiments and examples may be combined with each other without conflict.
It should be noted that the drawings provided in the following embodiments are only for illustrating the basic idea of the present invention, and the components related to the present invention are only shown in the drawings rather than drawn according to the number, shape and size of the components in actual implementation, and the type, quantity and proportion of the components in actual implementation may be changed freely, and the layout of the components may be more complicated.
As shown in FIG. 1, the invention provides a fault diagnosis and classification method for a rolling bearing based on VMD-SSAE, which comprises the following steps:
s1, collecting vibration signals of rolling bearings with different fault types;
s2, extracting time domain, frequency domain and time-frequency domain characteristics of the vibration signal based on the variation modal decomposition;
s3, forming the characteristic data into a data set, and dividing the data set into a training set, a verification set and a test set;
s4, connecting the stack sparse autoencoder with a Softmax classifier, constructing a VMD-SSAE classification model, and training the VMD-SSAE classification model by using a training set;
s5, optimizing the VMD-SSAE classification model by adopting a wolf optimization algorithm and an error back propagation algorithm to obtain an ideal VMD-SSAE classification model;
and S6, inputting the data in the test set into an ideal VMD-SSAE classification model to obtain a diagnosis classification result.
In an embodiment of the present invention, for step S1, for example, a testing machine is used to collect vibration signals of rolling bearings with different fault types, the testing machine uses a PCB three-way acceleration sensor, and the vibration signals of the faulty bearings are collected through a Signature Acquisition module in LMS test. In the present embodiment, the test bearings are single-row cylindrical rolling bearings of models NU1010EM and N1010EM, for example, and the fault types of the test bearings are now classified into nine types by performing damage point processing on the test bearings, and specific parameters of each type are shown in table 1 below.
TABLE 1
Figure BDA0002911427980000051
The operating conditions were divided into eight conditions according to different loads and speeds, and the description of each condition is shown in table 2 below.
TABLE 2
Figure BDA0002911427980000061
As shown in fig. 2, vibration signals of a normal bearing and a compound fault bearing with an outer ring and a rolling element are respectively collected under working conditions 3, 4, 7 and 8, wherein the sampling frequency is 20480 Hz. In the embodiment, the experimental working condition table is formed by a Latin hypercube experimental design method by taking the load, the rotating speed and the acceleration and deceleration change as experimental control factors, the service life test of the normal bearing and the outer ring rolling element composite type fault bearing is developed, and a time domain vibration signal is acquired.
As shown in fig. 3, step S2 further includes:
s21, extracting time domain characteristics of the vibration signals to obtain time domain characteristic data;
s22, extracting frequency domain characteristics of the vibration signals to obtain frequency domain characteristic data;
and S23, performing time-frequency domain feature extraction on the vibration signal by adopting variational modal decomposition to obtain time-frequency domain feature data.
In an embodiment of the present invention, for step S21, the time-domain characteristics of the vibration signal of the rolling bearing include a mean, a peak, a crest factor, a margin factor, a pulse coefficient, a shape coefficient, a mean square root, a skewness, a kurtosis, and a variance. The time domain features are sensitive to the fault response of the rolling bearing, have small correlation with the amplitude and the frequency of the vibration signal, have small correlation with the working condition of the rolling bearing, and only depend on the amplitude distribution function of the vibration signal, so the time domain features are used as time domain feature input data.
In an embodiment of the present invention, for step S22, the frequency domain information in the vibration signal is a very important component in signal analysis, and the type, severity and location of the fault of the rolling bearing can be determined by the frequency domain characteristic information. The frequency domain characteristics of the vibration signals of the rolling bearing comprise average frequency, center of gravity frequency, frequency root mean square and frequency standard deviation, and the frequency domain characteristics are used as frequency domain characteristic input data.
As shown in fig. 4, further, step S23 further includes:
s231, carrying out self-adaptive decomposition on the vibration signals by adopting variational modal decomposition to obtain a plurality of inherent modal functions;
and S232, extracting the time-frequency domain characteristics of each inherent mode function to obtain time-frequency domain characteristic data.
As shown in fig. 5, in an embodiment of the present invention, for step S231, the Variable Mode Decomposition (VMD) is a signal Decomposition estimation method, which determines a frequency center and a bandwidth of each component by iteratively searching for an optimal solution of the variable Mode Decomposition in the process of acquiring the Decomposition component, so that frequency domain division of the signal and effective separation of the components can be adaptively achieved, and the variable Mode Decomposition has better noise robustness. In this embodiment, for example, under the condition of operating condition 4, that is, the load is 3kN, and the rotation speed is 3000r/min, VMD decomposition is performed on the vibration signals of the normal bearing and the outer ring rolling element composite type faulty bearing, respectively, to obtain a plurality of Intrinsic Mode Functions (IMFs), where the number k of decomposition layers is, for example, 3.
In an embodiment of the present invention, after obtaining three layers of IMFs in step S232 by decomposition, the average value, peak value, crest factor, margin factor, pulse coefficient, shape coefficient, root mean square value, skewness, kurtosis, and variance of each IMF are counted and used as time-frequency domain feature data.
In an embodiment of the present invention, for step S3, the time domain, frequency domain, and time-frequency domain characteristic data of the vibration signal of each fault type under each working condition are respectively combined into a data set as input data of a Stack Sparse self-encoder (SSAE) neural network, and the data set is divided into a training set, a verification set, and a test set, where the training set is used to construct a model, the verification set is used to evaluate the model classification accuracy in the model construction process, so as to adaptively adjust a Sparse penalty factor, and the test set is used to evaluate the generalization capability of the final model; meanwhile, a one-hot code may be used to construct category labels for different fault types, and the fault types and category labels corresponding to the categories are shown in table 3 below.
TABLE 3
Figure BDA0002911427980000071
In an embodiment of the present invention, for step S4, the SSAE, as a typical deep learning model, has multiple non-linear hidden layers, and each hidden layer obtains advanced features by relearning from the original data information, so that the problem of feature redundancy and high-dimensional features can be effectively solved, and the SSAE also has the advantages of strong non-linear expression capability and strong generalization capability.
As shown in fig. 6, an Auto Encoder (AE) is a simple three-layer unsupervised learning neural network model, which is composed of an encoder and a decoder, and includes an input layer, a hidden layer and an output layer, wherein a training process is composed of two parts, namely an encoding process from the input layer to the hidden layer and a decoding process from the hidden layer to the output layer, the encoder converts input data from a high-dimensional space to a feature space with a lower dimension, and the decoder can reconstruct input data from the feature space to optimally solve a minimum error between the input data and the output data. First, given input data X ═ X1，x2，...，xNAnd mapping the input data to the hidden layer through a nonlinear activation function to obtain a coding result, wherein N represents the number of samples:
hw，b(x)＝s(W1X+b1) (1)
similarly, a nonlinear activation function is applied to encode the result hw，b(x) Mapping to an output layer to obtain an output
Figure BDA0002911427980000089
Figure BDA0002911427980000081
Wherein, W is weight, b is bias, s (X) is nonlinear activation function, sigmoid function is adopted, and is expressed as:
Figure BDA0002911427980000082
secondly, a mean square error function (MSE) is used as a loss function of the self-encoder, and then the reconstruction error is:
Figure BDA0002911427980000083
wherein the content of the first and second substances,
Figure BDA0002911427980000084
is the mean of the input data.
In order to prevent the fitting problem caused by excessive features, a regular optimization term is added into the loss function to improve the generalization performance of the self-encoder, and then the total loss function is:
Figure BDA0002911427980000085
wherein, lambda is a regular optimization coefficient, L is the number of network layers, and slThe number of the layer I neurons is shown as,
Figure BDA0002911427980000086
is the connection weight between layer l +1 neuron j and layer l neuron i.
Sparse auto-encoders (SAE) are based on the fact that an auto-encoder adds coefficient constraints to hidden layer neurons, i.e., adds a Sparse penalty term to the objective function of the auto-encoder. SAE can learn more abstract and representative compression features, which improve the performance of conventional auto-encoders and show more practical application value. If the input sample is X ═ X1，x2，...，xN}, then the mean activation value for hidden layer neurons is:
Figure BDA0002911427980000087
the relative entropy can measure the distance between two random distributions, and the value of the relative entropy increases with the difference between the two random distributions, and when the two random distributions are the same, the relative entropy is zero. After introducing relative entropy, the penalty factor is expressed as:
Figure BDA0002911427980000088
wherein D is the number of hidden layer neurons; ρ is a sparsity coefficient, which generally takes a value approximately equal to zero but not zero, e.g., 0.05.
Thus, the overall loss function for SAE is:
JSAE(W，b)＝J(W，b)+βKL(ρ||ρk) (8)
wherein, beta is a sparse penalty factor used for controlling relativity between the reconstruction term and the penalty term.
A Stack Sparse Auto Encoder (SSAE) is formed by stacking a plurality of Sparse auto encoders in series. The purpose of the SSAE is to extract high-order features of input data layer by layer, in the process, the dimensionality of the input data is reduced layer by layer, a complex input data is converted into a simple high-order feature, then the high-order feature is input into a classifier, and the classification of the feature is achieved through the classifier.
As shown in FIG. 7, in one embodiment of the invention, for example, three layers of sparse self-encoders SAE1, SAE2 and SAE3 are stacked to form an SSAE neural network, and combined with a Softmax classifier to construct a VMD-SSAE neural network classification model. The SSAE neural network takes the characteristics output by the SAE1 hidden layer as the input data of SAE2, takes the characteristics output by the SAE2 hidden layer as the input data of SAE3, trains the SSAE neural network layer by using a gradient descent method, updates the weight and the offset of each layer of network through multiple iterations, and finally extracts the input original data characteristics; the fault diagnosis classification is then achieved by a Softmax classifier connected to the SSAE output. Wherein the output vector of the SSAE network is represented as:
Y＝f(X，β1，2，3) (9)
wherein X is an input vector; beta is a1，2，3Is a sparse penalty factor for each SAE.
Then taking the output of the SSAE network as the input of the Softmax classifier, the output vector P of the VMD-SSAE classification modeliComprises the following steps:
Figure BDA0002911427980000091
where p (i) represents the probability that the input vector x belongs to the class i (i ═ 1, 2.. K), where K is the total number of classes, and in this embodiment, K ═ 9.
In this embodiment, the sample data of the verification set may also be input into the VMD-SSAE classification model, and the mean square error between the prediction output of the VMD-SSAE classification model on the verification set and the class label of the verification set is used as an index for evaluating the classification accuracy of the VMD-SSAE classification model, which is expressed as:
Figure BDA0002911427980000092
wherein l is a category label; n is the number of the sample data in the verification set; k is the total number of categories; p (. beta.) of1，2，3) Representing a given parameter beta1，2，3The value of (c) is a prediction output of the VMD-SSAE model. The smaller the verification error, the closer the prediction output is to the class label, indicating that the VMD-SSAE model has higher classification accuracy on the verification set.
Thus, the mathematical model of the VMD-SSAE classification model may be expressed as:
Figure BDA0002911427980000101
wherein, P (. beta.) is1,2,3) Is an independent variable; mesIs an objective function; the constraint condition is that the value of the independent variable is greater than zero; the mathematics ofThe essence of the model is to find the value of the argument corresponding to the minimum objective function under the constraint conditions.
Further, step S5 further includes:
s51, self-adaptively selecting the sparse penalty factor of the sparse self-encoder by adopting a wolf optimization algorithm to obtain the sparse penalty factor in the optimized VMD-SSAE classification model;
and S52, fine tuning the optimized VMD-SSAE classification model by adopting an error back propagation algorithm to obtain an ideal VMD-SSAE classification model.
As shown in fig. 8, step S51 further includes:
s511, initializing basic parameters of the gray wolf optimization algorithm, wherein the basic parameters comprise: the number of the grey wolves and the iteration times, wherein the number of the grey wolves is more than three;
s512, calculating the fitness value of the gray wolf;
s513, sequentially determining the first three optimal wolfs according to the fitness value;
s514, updating the positions of other gray wolves according to the positions of the first three optimal wolves;
s515, adding one to the iteration times, and calculating the fitness value of the current wolf;
s516, comparing the fitness value of the current gray wolf with the fitness value of the previous generation gray wolf, and selecting the minimum value of the fitness values as the minimum fitness value;
s517, judging whether a termination condition is reached, if so, taking the position of the wolf corresponding to the minimum fitness value as an optimal sparse penalty factor of the sparse self-encoder, otherwise, returning to the step S513, and continuously updating the position of the wolf; wherein the termination condition comprises: the minimum fitness value is smaller than a preset fitness threshold value, or the iteration number reaches a preset iteration number upper limit.
The Grey Wolf optimization algorithm (GWOlf Optimizer, GWOO) is a group intelligent optimization algorithm, achieves the optimization purpose by simulating the predation behavior of Grey Wolf groups and based on a Wolf group cooperation mechanism, and has the advantages of strong convergence performance, few parameters, easiness in implementation and the like. In the GWO algorithm, the first three wolf individuals closest to the prey are defined as α, β, δ, which have the best fitness values in the population, so that α, β, δ wolf leads the remaining wolf population to search for the prey to a promising space in the search area.
In an embodiment of the invention, for step S511, for example, the number of the wolf populations is 50, and the upper limit of the preset number of iterations is 150.
In one embodiment of the present invention, for step S512, the fitness value of each gray wolf is calculated using formula (13):
fitness＝Mes(p(β1，2，3)，l) (13)
in this embodiment, for step S513, the principle of sequentially determining the first three best wolfs according to the fitness value of each wolf is as follows: the gray wolf corresponding to the minimum fitness value is used as the first optimal wolf, namely alpha wolf, and then the gray wolf corresponding to the next fitness value is used as the second optimal wolf and the third optimal wolf, namely beta wolf and delta wolf in sequence.
In the present embodiment, for step S514, in the GWO algorithm, the distance between the wolf and the prey, i.e. the optimal sparse penalty factor, is represented as:
Figure BDA0002911427980000111
wherein the content of the first and second substances,
Figure BDA0002911427980000112
is a coefficient vector with a value of | C | ═ 2 × r1，r1Is [0, 1 ]]A random number in between;
Figure BDA0002911427980000113
is the position vector of the prey;
Figure BDA0002911427980000114
is the location vector of the gray wolf; t represents the current number of iterations.
Thus, the location update of the gray wolf is represented as:
Figure BDA0002911427980000115
wherein the content of the first and second substances,
Figure BDA0002911427980000116
is a coefficient vector having a value of | A | ═ 2 × a × r2A, a is a convergence factor, whose value decreases gradually from 2 to 0, r2Is [0, 1 ]]A random number in between.
In this embodiment, the first three optimal wolf positions are respectively substituted for the initial position of the prey, so that the other wolf individuals can perform position update according to the positions of the three optimal wolf individuals, i.e. the positions of α, β, δ wolf
Figure BDA0002911427980000117
Respectively replace
Figure BDA0002911427980000118
Sequentially calculate out
Figure BDA0002911427980000119
And then updating the positions of other gray wolves according to the formula (15).
Figure BDA00029114279800001110
The steps of the above method are divided for clarity of description, and may be combined into one step or split into some steps, and the steps are decomposed into multiple steps, so long as the steps contain the same logical relationship, which is within the protection scope of the present invention; it is within the scope of the present invention to add insignificant modifications to the algorithms or processes or to introduce insignificant design changes to the core design without changing the algorithms or processes.
In an embodiment of the present invention, for example, under different working conditions, there are 300 sample data for each fault type, and 70% of the sample data for each fault type in working conditions 1, 3, 5, and 7, that is, 210 sample data in each fault type, are randomly extracted as a training set; randomly extracting 20% of sample data of each fault type in working conditions 1, 3, 5 and 7, namely 60 sample data in each fault type as a verification set; randomly extracting 10% of sample data of each fault type in the working conditions 1, 3, 5 and 7, namely, 30 sample data in each fault type as a test set 0; and the data in the training set, the verification set and the test set 0 are not repeatedly extracted. 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100% of sample data of each fault type in the working conditions 2, 4, 6, 8 are respectively randomly extracted and are respectively marked as a test set 1, a test set 2, a test set 3, a test set 4, a test set 5, a test set 6, a test set 7, a test set 8, a test set 9 and a test set 10, and the sample data are used for testing the fault diagnosis capability of the model under different loads and different test sample quantities. In the embodiment, a VMD-SSAE classification model is constructed by adopting a three-layer sparse self-encoder and a Softmax classifier, and sparse penalty factors of SAE1, SAE2 and SAE3 are adaptively selected by an algorithm GWO. In this embodiment, the diagnostic classification result of the model is represented in the form of a confusion matrix, which is an index for judging the classification model result, rows of the confusion matrix represent the real categories to which the data belongs, columns represent the predicted categories of the data by the classification model, so that the diagonal elements indicate the probability of correct classification of the model, and other elements indicate the probability of confusion of the classification.
As shown in fig. 9, in the present embodiment, the test set 0 is used to perform fault diagnosis classification on the model obtained by adaptively selecting the sparse penalty factor of the sparse autoencoder using the GWO algorithm, and the result is: the prediction accuracy of the model is 99.5%, and only a small amount of fifth-class fault data are divided into eighth-class fault types, because the fifth-class fault and the eighth-class fault contain outer-ring fault point data, a small amount of error classification phenomena can occur in the characteristic classification process. Therefore, under the condition of stable load, the deep neural network model has good fault diagnosis and classification capability.
In the present embodiment, the test set 1 to the test set 10 are used to test the fault diagnosis capability of the model under different loads and test sample sizes, and the diagnosis classification accuracy is shown in table 4 below.
TABLE 4
Figure BDA0002911427980000121
Under different loads, the classification accuracy of the test sets with different test sample amounts is higher than 98.0%, wherein the classification accuracy of the test set 3 and the test set 5 is as high as 99.5%. Therefore, the VMD-SSAE model optimized by the GWO algorithm has good fault diagnosis capability under different loads and different test sample sizes.
As shown in fig. 10, in an embodiment of the present invention, in order to verify the effectiveness of the VMD-SSAE classification model optimized by using GWO algorithm, the SSAE model optimized by using GWO algorithm is compared with the SSAE model with random parameters, SAE model, KNN (K-Nearest Neighbor classification algorithm), and SVM (Support Vector Machines), respectively, to compare the diagnostic classification accuracy. In this embodiment, the training sets are used to train the models respectively, and then the trained models are tested by using the test sets 1 to 10, wherein the sparse penalty factor of the SSAE model with random parameters is a random parameter and is different from the sparse penalty factor in the present invention; the parameters of the SAE model are: the number of neurons in an implicit layer is set to be 9, the maximum iteration number is set to be 150, the regularization coefficient is set to be 0.002, the sparsity coefficient is set to be 0.3, the sparse penalty factor is a random parameter, and the maximum iteration number of the Softmax classifier is set to be 100; the SVM uses a Radial Basis Function (RBF) as its kernel Function, with the kernel parameter set to 0.3 and the penalty factor set to 1.2. The average classification accuracy of the different models is described in table 5 below.
TABLE 5
Figure BDA0002911427980000131
The comparison results are as follows: the classification accuracy of the five models is over 90 percent; the highest accuracy is 99.5%, which corresponds to an optimized SSAE model, and the lowest accuracy is 90.2%, which corresponds to an SAE model, which shows the effectiveness of feature extraction in time domain, frequency domain and time-frequency domain, i.e. the features of different fault types can be well extracted by the feature extraction. By comparing the average classification accuracy of the five models, the fault classification accuracy of the SSAE model optimized by using the GWO algorithm and the SSAE model with random parameters is higher than that of other models, but the classification accuracy of the two models under different test sets fluctuates due to the fact that random parameters are adopted by the SSAE model with random parameters and the sparse penalty factors of the SAE model. Therefore, the VMD-SSAE model optimized by the GWO algorithm has good fault diagnosis classification capability under different loads and different test data volumes.
As shown in fig. 11, the present invention further provides a rolling bearing fault diagnosis and classification system based on VMD-SSAE, comprising: the system comprises a signal acquisition module 11, a feature extraction module 12, a data set construction module 13, a model construction module 14, a model optimization module 15 and a fault diagnosis module 16. The signal acquisition module 11 is used for acquiring vibration signals of rolling bearings with different fault types; the feature extraction module 12 is configured to extract time domain, frequency domain, and time-frequency domain features of the vibration signal based on the variational modal decomposition; the data set construction module 13 is configured to form a data set from the features, and divide the data set into a training set, a verification set, and a test set; the model construction module 14 is used for connecting the stack sparse self-encoder with a Softmax classifier to form a VMD-SSAE classification model, and training the VMD-SSAE classification model by using the training set; the model optimization module 15 is configured to optimize the VMD-SSAE classification model by using a grey wolf optimization algorithm and an error back propagation algorithm to obtain an ideal VMD-SSAE classification model; the fault diagnosis module 16 is used for inputting the data in the test set into the VMD-SSAE classification model to obtain a diagnosis classification result.
It should be noted that, in order to highlight the innovative part of the present invention, a module which is not so closely related to solve the technical problem proposed by the present invention is not introduced in the present embodiment, but this does not indicate that no other module exists in the present embodiment.
In addition, it is clear to those skilled in the art that, for convenience and brevity of description, the specific working process of the system described above may refer to the corresponding process in the foregoing method embodiment, and is not described herein again. In the embodiments provided in the present invention, it should be understood that the disclosed system, apparatus and method may be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the modules is merely a logical division, and in actual implementation, there may be other divisions, for example, multiple modules or components may be combined or integrated into another system, or some features may be omitted, or not implemented. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.
Modules described as separate parts may or may not be physically separate, and parts displayed as modules may or may not be physical modules, may be located in one place, or may be distributed on a plurality of network modules. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, functional modules in the embodiments of the present invention may be integrated into one processing module, or each of the modules may exist alone physically, or two or more modules are integrated into one module. The integrated module can be realized in a form of hardware or a form of a software functional unit.
The integrated module, if implemented in the form of a software functional module and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a read-only memory (ROM), a Random Access Memory (RAM), a magnetic disk, and an optical disk.
In an embodiment of the invention, the rolling bearing fault diagnosis and classification device based on the VMD-SSAE comprises the following components: a processor coupled to a memory, the memory storing program instructions that, when executed by the processor, implement the method of rolling bearing fault diagnostic classification.
As described above, the rolling bearing fault diagnosis and classification method, system, equipment and storage medium thereof based on VMD-SSAE provided by the invention can diagnose the fault type of the rolling bearing more accurately under time-varying non-stationary load and different bearing test data volume; secondly, the invention can adaptively perform optimal selection on the sparse penalty factor of the stack sparse self-encoder network, thereby not only reducing the complex workload, but also avoiding the subjective influence of artificially given parameters; the VMD-SSAE classification model also has better fault tolerance and robustness, so that better fault diagnosis effect can be achieved. Therefore, the invention effectively overcomes various defects in the prior art and has high industrial utilization value.
The foregoing embodiments are merely illustrative of the principles and utilities of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or change the above-mentioned embodiments without departing from the spirit and scope of the present invention. Accordingly, it is intended that all equivalent modifications or changes which can be made by those skilled in the art without departing from the spirit and technical spirit of the present invention be covered by the claims of the present invention.

Claims (10)

1. A rolling bearing fault diagnosis and classification method based on VMD-SSAE is characterized by comprising the following steps:
collecting vibration signals of rolling bearings with different fault types;
based on variation modal decomposition, extracting time domain, frequency domain and time-frequency domain characteristics of the vibration signal;
forming the features into a data set, and dividing the data set into a training set, a verification set and a test set;
connecting a stack sparse self-encoder with a Softmax classifier, constructing a VMD-SSAE classification model, and training the VMD-SSAE classification model by using the training set;
optimizing the VMD-SSAE classification model by adopting a wolf optimization algorithm and an error back propagation algorithm to obtain an ideal VMD-SSAE classification model;
and inputting the data in the test set into an ideal VMD-SSAE classification model to obtain a diagnosis classification result.
2. The VMD-SSAE-based rolling bearing fault diagnosis and classification method according to claim 1, wherein the collecting vibration signals of rolling bearings with different fault types comprises: and acquiring the vibration signal of the rolling bearing under different loads and rotating speeds.
3. The VMD-SSAE based rolling bearing fault diagnosis and classification method according to claim 1, wherein the time domain, frequency domain and time-frequency domain feature extraction is performed on the vibration signal based on the variational modal decomposition, and the method comprises the following steps:
extracting time domain characteristics of the vibration signals to obtain time domain characteristic data;
extracting frequency domain characteristics of the vibration signals to obtain frequency domain characteristic data;
and adopting the variation modal decomposition to extract the time-frequency domain characteristics of the vibration signals to obtain time-frequency domain characteristic data.
4. The VMD-SSAE based rolling bearing fault diagnosis and classification method according to claim 3, wherein the obtaining of the time-frequency domain feature data comprises the steps of:
adopting the variation modal decomposition to carry out self-adaptive decomposition on the vibration signal to obtain a plurality of inherent modal functions;
and extracting the time-frequency domain characteristics of each inherent mode function to obtain the time-frequency domain characteristic data.
5. The VMD-SSAE based rolling bearing fault diagnosis and classification method according to claim 1, wherein the VMD-SSAE classification model is optimized by a wolf optimization algorithm and an error back propagation algorithm to obtain an ideal VMD-SSAE classification model, comprising the steps of:
adopting the gray wolf optimization algorithm to perform self-adaptive selection on the sparse penalty factor of the sparse self-encoder so as to obtain the optimized sparse penalty factor in the VMD-SSAE classification model;
and fine-tuning the optimized VMD-SSAE classification model by adopting the error back propagation algorithm to obtain an ideal VMD-SSAE classification model.
6. The VMD-SSAE based rolling bearing fault diagnosis and classification method according to claim 5, wherein the self-adaptive selection of the sparse penalty factor of the sparse self-encoder by adopting the wolf optimization algorithm comprises the steps of:
s511, initializing basic parameters of the gray wolf optimization algorithm, wherein the basic parameters comprise: the number of the grey wolves and the number of iterations, wherein the number of the grey wolves is more than three;
s512, calculating the fitness value of the gray wolf;
s513, sequentially determining the first three optimal wolfs according to the fitness value;
s514, updating the positions of other gray wolves according to the positions of the first three optimal wolves;
s515, adding one to the iteration times, and calculating the fitness value of the current wolf;
s516, comparing the fitness value of the current gray wolf with the fitness value of the previous generation gray wolf, and selecting the minimum value of the fitness values as the minimum fitness value;
s517, judging whether a termination condition is reached, if so, taking the position of the wolf corresponding to the minimum fitness value as an optimal sparse penalty factor of the sparse self-encoder, otherwise, returning to the step S513, and continuously updating the position of the wolf; wherein the termination condition comprises: the minimum fitness value is smaller than a preset fitness threshold value.
7. The VMD-SSAE based rolling bearing fault diagnosis and classification method according to claim 6, wherein the step of sequentially determining the first three optimal wolfs according to the fitness value comprises: the grey wolf with the minimum fitness value is used as the first optimal wolf, and the grey wolf with the second-order fitness value is used as the second optimal wolf and the third optimal wolf in sequence.
8. A rolling bearing fault diagnosis and classification system based on VMD-SSAE is characterized by at least comprising:
the signal acquisition module is used for acquiring vibration signals of rolling bearings with different fault types;
the characteristic extraction module is used for extracting time domain, frequency domain and time-frequency domain characteristics of the vibration signal based on variational modal decomposition;
the data set construction module is used for forming the features into a data set and dividing the data set into a training set, a verification set and a test set;
the model building module is used for connecting the stack sparse self-encoder with a Softmax classifier to form a VMD-SSAE classification model and training the VMD-SSAE classification model by using the training set;
the model optimization module is used for optimizing the VMD-SSAE classification model by adopting a gray wolf optimization algorithm and an error back propagation algorithm so as to obtain an ideal VMD-SSAE classification model;
and the fault diagnosis module is used for inputting the data in the test set into the VMD-SSAE classification model to obtain a diagnosis classification result.
9. The utility model provides a rolling bearing fault diagnosis sorting equipment based on VMD-SSAE which characterized in that: comprising a processor coupled with a memory, the memory storing program instructions that, when executed by the processor, implement the method of any of claims 1 to 7.
10. A computer-readable storage medium, characterized in that: comprising a program which, when run on a computer, causes the computer to perform the method of any one of claims 1 to 7.
CN202110087518.3A 2021-01-22 2021-01-22 Rolling bearing fault diagnosis and classification method, system and equipment based on VMD-SSAE and storage medium thereof Pending CN112836604A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110087518.3A CN112836604A (en) 2021-01-22 2021-01-22 Rolling bearing fault diagnosis and classification method, system and equipment based on VMD-SSAE and storage medium thereof

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110087518.3A CN112836604A (en) 2021-01-22 2021-01-22 Rolling bearing fault diagnosis and classification method, system and equipment based on VMD-SSAE and storage medium thereof

Publications (1)

Publication Number Publication Date
CN112836604A true CN112836604A (en) 2021-05-25

Family

ID=75930841

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110087518.3A Pending CN112836604A (en) 2021-01-22 2021-01-22 Rolling bearing fault diagnosis and classification method, system and equipment based on VMD-SSAE and storage medium thereof

Country Status (1)

Country Link
CN (1) CN112836604A (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113177537A (en) * 2021-06-29 2021-07-27 湖北博华自动化系统工程有限公司 Fault diagnosis method and system for rotary mechanical equipment
CN113204849A (en) * 2021-05-26 2021-08-03 西安工业大学 Gear peeling fault detection method for gear box
CN113608018A (en) * 2021-06-30 2021-11-05 中冶南方都市环保工程技术股份有限公司 Adaptive VMD detection method and device for improving harmonic detection precision and storage medium
CN113670609A (en) * 2021-07-21 2021-11-19 广州大学 Fault detection method, system, device and medium based on wolf optimization algorithm
CN114714146A (en) * 2022-04-08 2022-07-08 北京理工大学 Method for simultaneously predicting surface roughness and cutter abrasion

Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107560851A (en) * 2017-08-28 2018-01-09 合肥工业大学 Rolling bearing Weak fault feature early stage extracting method
CN110132596A (en) * 2019-04-24 2019-08-16 昆明理工大学 A method of the rolling bearing fault diagnosis based on wavelet packet and GWO-SVM
CN110455537A (en) * 2019-09-18 2019-11-15 合肥工业大学 A kind of Method for Bearing Fault Diagnosis and system
CN110470477A (en) * 2019-09-19 2019-11-19 福州大学 A kind of Fault Diagnosis of Roller Bearings based on SSAE and BA-ELM
CN110595765A (en) * 2019-08-26 2019-12-20 西安理工大学 Wind turbine generator gearbox fault diagnosis method based on VMD and FA _ PNN
CN111289250A (en) * 2020-02-24 2020-06-16 湖南大学 Method for predicting residual service life of rolling bearing of servo motor
CN111428418A (en) * 2020-02-28 2020-07-17 贵州大学 Bearing fault diagnosis method and device, computer equipment and storage medium
CN111476263A (en) * 2019-12-27 2020-07-31 江苏科技大学 Bearing defect identification method based on SDAE and improved GWO-SVM
CN111797567A (en) * 2020-06-09 2020-10-20 合肥工业大学 Deep learning network-based bearing fault classification method and system
CN112163472A (en) * 2020-09-15 2021-01-01 东南大学 Rolling bearing diagnosis method based on multi-view feature fusion

Patent Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107560851A (en) * 2017-08-28 2018-01-09 合肥工业大学 Rolling bearing Weak fault feature early stage extracting method
CN110132596A (en) * 2019-04-24 2019-08-16 昆明理工大学 A method of the rolling bearing fault diagnosis based on wavelet packet and GWO-SVM
CN110595765A (en) * 2019-08-26 2019-12-20 西安理工大学 Wind turbine generator gearbox fault diagnosis method based on VMD and FA _ PNN
CN110455537A (en) * 2019-09-18 2019-11-15 合肥工业大学 A kind of Method for Bearing Fault Diagnosis and system
CN110470477A (en) * 2019-09-19 2019-11-19 福州大学 A kind of Fault Diagnosis of Roller Bearings based on SSAE and BA-ELM
CN111476263A (en) * 2019-12-27 2020-07-31 江苏科技大学 Bearing defect identification method based on SDAE and improved GWO-SVM
CN111289250A (en) * 2020-02-24 2020-06-16 湖南大学 Method for predicting residual service life of rolling bearing of servo motor
CN111428418A (en) * 2020-02-28 2020-07-17 贵州大学 Bearing fault diagnosis method and device, computer equipment and storage medium
CN111797567A (en) * 2020-06-09 2020-10-20 合肥工业大学 Deep learning network-based bearing fault classification method and system
CN112163472A (en) * 2020-09-15 2021-01-01 东南大学 Rolling bearing diagnosis method based on multi-view feature fusion

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
周建中 等著, 武汉：华中科技大学出版社 *
徐飞 等: "基于VMD-样本熵和SSAE的齿轮故障诊断", 《组合机床与自动化加工技术》 *
王奉涛 等: "基于EMD和SSAE的滚动轴承故障诊断方法", 《振动工程学报》 *
袁宪锋 等: "SSAE和IGWO-SVM的滚动轴承故障诊断", 《振动、测试与诊断》 *

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113204849A (en) * 2021-05-26 2021-08-03 西安工业大学 Gear peeling fault detection method for gear box
CN113177537A (en) * 2021-06-29 2021-07-27 湖北博华自动化系统工程有限公司 Fault diagnosis method and system for rotary mechanical equipment
CN113177537B (en) * 2021-06-29 2021-09-17 湖北博华自动化系统工程有限公司 Fault diagnosis method and system for rotary mechanical equipment
CN113608018A (en) * 2021-06-30 2021-11-05 中冶南方都市环保工程技术股份有限公司 Adaptive VMD detection method and device for improving harmonic detection precision and storage medium
CN113670609A (en) * 2021-07-21 2021-11-19 广州大学 Fault detection method, system, device and medium based on wolf optimization algorithm
CN114714146A (en) * 2022-04-08 2022-07-08 北京理工大学 Method for simultaneously predicting surface roughness and cutter abrasion

Similar Documents

Publication Publication Date Title
CN112836604A (en) Rolling bearing fault diagnosis and classification method, system and equipment based on VMD-SSAE and storage medium thereof
CN105224872B (en) A kind of user's anomaly detection method based on neural network clustering
CN111680446B (en) Rolling bearing residual life prediction method based on improved multi-granularity cascade forest
CN115018021B (en) Machine room abnormity detection method and device based on graph structure and abnormity attention mechanism
Xia et al. Multi-stage fault diagnosis framework for rolling bearing based on OHF Elman AdaBoost-Bagging algorithm
CN110751108B (en) Subway distributed vibration signal similarity determination method
CN109389171B (en) Medical image classification method based on multi-granularity convolution noise reduction automatic encoder technology
CN112039903B (en) Network security situation assessment method based on deep self-coding neural network model
CN113743016B (en) Engine residual life prediction method based on self-encoder and echo state network
CN110987436B (en) Bearing fault diagnosis method based on excitation mechanism
CN111397902A (en) Rolling bearing fault diagnosis method based on feature alignment convolutional neural network
CN112529638B (en) Service demand dynamic prediction method and system based on user classification and deep learning
CN115270986A (en) Data anomaly detection method and device and computer equipment
CN114819315A (en) Bearing degradation trend prediction method based on multi-parameter fusion health factor and time convolution neural network
CN112085062A (en) Wavelet neural network-based abnormal energy consumption positioning method
Saufi et al. Machinery fault diagnosis based on a modified hybrid deep sparse autoencoder using a raw vibration time-series signal
CN117117859B (en) Photovoltaic power generation power prediction method and system based on neural network
CN116738868B (en) Rolling bearing residual life prediction method
CN116881841A (en) Hybrid model fault diagnosis method based on F1-score multistage decision analysis
CN116720743A (en) Carbon emission measuring and calculating method based on data clustering and machine learning
CN115290326A (en) Rolling bearing fault intelligent diagnosis method
CN115275977A (en) Power load prediction method and device
CN113935413A (en) Distribution network wave recording file waveform identification method based on convolutional neural network
CN113435321A (en) Method, system and equipment for evaluating state of main shaft bearing and readable storage medium
CN114238852A (en) Operation data analysis method and device, storage medium and electronic equipment

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