CN114201830A - Rolling bearing fault diagnosis model establishing method and system - Google Patents
Rolling bearing fault diagnosis model establishing method and system Download PDFInfo
- Publication number
- CN114201830A CN114201830A CN202111498873.6A CN202111498873A CN114201830A CN 114201830 A CN114201830 A CN 114201830A CN 202111498873 A CN202111498873 A CN 202111498873A CN 114201830 A CN114201830 A CN 114201830A
- Authority
- CN
- China
- Prior art keywords
- fault
- dbn
- rolling bearing
- normal
- vibration
- 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
- 238000005096 rolling process Methods 0.000 title claims abstract description 96
- 238000003745 diagnosis Methods 0.000 title claims abstract description 47
- 238000000034 method Methods 0.000 title claims abstract description 44
- 238000005457 optimization Methods 0.000 claims abstract description 38
- 239000011159 matrix material Substances 0.000 claims abstract description 35
- 238000000605 extraction Methods 0.000 claims abstract description 12
- 238000001228 spectrum Methods 0.000 claims abstract description 10
- 241000287127 Passeridae Species 0.000 claims description 47
- 238000012549 training Methods 0.000 claims description 30
- 238000012360 testing method Methods 0.000 claims description 23
- 230000008859 change Effects 0.000 claims description 13
- 230000000694 effects Effects 0.000 claims description 9
- 230000006870 function Effects 0.000 claims description 7
- 230000015654 memory Effects 0.000 claims description 6
- 238000003860 storage Methods 0.000 claims description 3
- 238000000354 decomposition reaction Methods 0.000 abstract description 8
- 230000008569 process Effects 0.000 description 10
- 244000062645 predators Species 0.000 description 9
- 230000006872 improvement Effects 0.000 description 6
- 238000004422 calculation algorithm Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 238000004590 computer program Methods 0.000 description 4
- 238000013135 deep learning Methods 0.000 description 3
- 230000002431 foraging effect Effects 0.000 description 3
- 238000010845 search algorithm Methods 0.000 description 3
- 230000001133 acceleration Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000003909 pattern recognition Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 238000005299 abrasion Methods 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000003190 augmentative effect Effects 0.000 description 1
- 238000005311 autocorrelation function Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 238000005315 distribution function Methods 0.000 description 1
- 238000004870 electrical engineering Methods 0.000 description 1
- 235000003642 hunger Nutrition 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000004393 prognosis Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/17—Mechanical parametric or variational design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/02—Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Geometry (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Computer Hardware Design (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Analysis (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Computational Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention relates to a rolling bearing fault diagnosis model based on VMD-WVD and SSA parameter optimization DBN. The model establishment method is as follows: the method comprises the steps of acquiring a plurality of vibration signals of the rolling bearing in different vibration states, and firstly, decomposing the vibration signals into a series of BIMF with different frequencies through variation modal decomposition. Variance contribution rates for each BIMF are compared. And converting the BIMF signal with the maximum contribution rate from a one-dimensional time domain signal into a two-dimensional characteristic spectrum through the WVD, thereby obtaining a two-dimensional characteristic matrix of the fault state. And optimizing DBN parameter combinations by using the global optimization capability of the SSA to obtain an optimized DBN, and optimizing the input value of the two-dimensional characteristic matrix to obtain the optimal network structure of the DBN, namely obtaining the rolling bearing fault. The result shows that the model established by the invention can improve the signal feature extraction capability, solve the problem under the global optimization condition and improve the accuracy of fault diagnosis.
Description
Technical Field
The invention relates to the technical field of fault diagnosis of industrial equipment, in particular to a rolling bearing fault diagnosis model building method and building system based on VMD-WVD and SSA parameter optimization DBN.
Background
The rolling bearing is used as an important component of a transmission instrument and is widely applied to various transmission systems in the fields of electric power, wind power, military, industrial machinery production and the like. The rolling bearing is often accompanied with faults such as abrasion, noise, load imbalance and the like due to the characteristics of long working time, high working frequency, severe working environment and the like.
The fault diagnosis of the rolling bearing is divided into three parts: data acquisition, feature extraction and pattern recognition. Among them, feature extraction is the most important part, and it greatly affects the upper and lower limits of pattern recognition accuracy. Various one-dimensional or two-dimensional signal feature extraction techniques are applied to vibration signal feature extraction, such as fourier transform, empirical mode decomposition, synthetic empirical mode decomposition, continuous wavelet transform, hilbert envelope spectrum, and variational mode decomposition.
As deep learning evolves, deep learning models are often used for optimization, diagnosis, and prognosis. Common network models include Deep Belief Networks (DBNs), Convolutional Neural Networks (CNNs), long-term memory neural networks (LSTM), Support Vector Machines (SVMs), and the like. After signal characteristic processing, the classification and identification precision of the deep learning network model is higher than that of other machine learning algorithms.
Therefore, it is desirable to provide a method for establishing a fault diagnosis model of a rolling bearing based on deep learning, so as to improve the accuracy of fault diagnosis.
Disclosure of Invention
The invention aims to solve the technical problem of providing a rolling bearing fault diagnosis model establishing method and system based on VMD-WVD and SSA parameter optimization DBN.
In order to solve the problems, the technical scheme adopted by the invention is as follows:
in one aspect, a rolling bearing fault diagnosis model establishing method is provided, and the method includes:
acquiring a plurality of normal vibration signals of a rolling bearing in a normal vibration state, a plurality of outer ring fault vibration signals in an outer ring fault vibration state, a plurality of inner ring fault vibration signals in an inner ring fault vibration state and a plurality of rolling body fault vibration signals in a rolling body fault vibration state;
extracting the characteristics of each vibration signal to obtain a data set containing a plurality of normal matrixes, a plurality of outer ring fault matrixes, a plurality of inner ring fault matrixes and a plurality of rolling element fault matrixes;
optimizing the parameters of the DBN by using the global optimization capability of the SSA to obtain an optimized DBN;
and inputting the data set into the optimized DBN for training to obtain the fault diagnosis model of the rolling bearing.
As a further improvement of the present invention, the extracting the characteristics of each vibration signal to obtain a data set including a plurality of normal matrices, a plurality of outer ring fault matrices, a plurality of inner ring fault matrices, and a plurality of rolling element fault matrices includes:
for each normal vibration signal, performing time-frequency change sequentially through the VMD and the WVD to obtain a normal matrix;
for each inner ring fault vibration signal, performing time-frequency change sequentially through the VMD and the WVD to obtain an outer ring fault matrix;
for each inner ring fault vibration signal, carrying out time-frequency change sequentially through the VMD and the WVD to obtain an inner ring fault matrix;
for each rolling element fault vibration signal, performing time-frequency change sequentially through VMD and WVD to obtain a rolling element fault matrix;
and storing all the normal matrix, the outer ring fault matrix, the inner ring fault matrix and the rolling body fault matrix into a data set.
As a further improvement of the present invention, the obtaining of the normal matrix after performing time-frequency change sequentially by VMD and WVD for each normal vibration signal includes:
for each normal vibration signal, decomposing the normal vibration signal into a plurality of normal BIMFs with different frequencies through VMD, comparing the variance contribution rate of each normal BIMF, and converting the normal BIMF with the maximum contribution rate from a one-dimensional time domain signal into a two-dimensional characteristic spectrum through WVD to obtain the normal matrix.
As a further improvement of the present invention, the variance contribution rate of each normal BIMF is compared using the following formula:
wherein u isk(j) The original signal is decomposed by VMD to obtain j IMF components, mseb (i) is IMFiN is the signal length.
As a further improvement of the present invention, the optimizing the parameters of the DBN by using the global optimization capability of the SSA to obtain the optimized DBN includes:
s301, setting all preset parameters of SSA, initializing the maximum iteration times and searching a preset range, wherein the preset parameters comprise a preset sparrow number, a preset danger sensing sparrow number and a preset warning value;
s302, initializing a sparrow position and calculating the fitness, and taking the RMSE of DBN training data to be optimized as a fitness function of the SSA;
s303, comparing the estimated fitness of each sparrow individual to obtain the optimal position and the worst position of each sparrow;
s304, when the current optimization iteration times reach the maximum iteration times, ending the iteration and outputting the optimal position, and training the DBN to be optimized by using the optimal position to obtain the optimized DBN with the optimal learning rate and the optimal batch size.
As a further improvement of the present invention, the steps further comprise:
s305, when the current optimization iteration frequency is judged not to reach the maximum iteration frequency, updating the position and the fitness of each sparrow individual, re-initializing, and setting a new search range, wherein the new search range is out of the search preset range;
s306, taking the current optimal position as a parameter value of a DBN training part to obtain an initial optimized DBN;
s307, comparing the estimated fitness of each sparrow individual to obtain the optimal position of each sparrow;
and S308, judging whether the current optimization iteration number reaches the maximum iteration number, and executing the step S304 or the step S305 to obtain the optimized DBN.
As a further improvement of the present invention, the inputting the data set into the optimized DBN for training to obtain the rolling bearing fault diagnosis model includes:
dividing the data set into a plurality of test sets and a plurality of test sets according to the vibration state, wherein each training set and each test set carry a label for indicating the vibration state;
inputting a plurality of test sets into the optimized DBN for training to obtain a trained DBN;
inputting a plurality of test sets into the trained DBN to test the classification effect of the trained DBN;
and when the classification effect is determined to reach the preset effect, taking the trained DBN as a fault diagnosis model of the rolling bearing.
In a second aspect, there is provided a rolling bearing fault diagnosis model building system, including:
the device comprises an acquisition module, a control module and a control module, wherein the acquisition module is used for acquiring a plurality of normal vibration signals of the rolling bearing in a normal vibration state, a plurality of outer ring fault vibration signals in an outer ring fault vibration state, a plurality of inner ring fault vibration signals in an inner ring fault vibration state and a plurality of rolling body fault vibration signals in a rolling body fault vibration state;
the data set generating module is used for extracting the characteristics of each vibration signal to obtain a data set comprising a plurality of normal matrixes, a plurality of outer ring fault matrixes, a plurality of inner ring fault matrixes and a plurality of rolling element fault matrixes;
the optimization module is used for optimizing the parameters of the DBN by utilizing the global optimization capability of the SSA to obtain an optimized DBN;
and the model generation module is used for inputting the data set into the optimized DBN for training to obtain the rolling bearing fault diagnosis model.
In a third aspect, a computer-readable storage medium is provided for storing computer instructions, and the computer instructions, when executed by a processor, implement the rolling bearing fault diagnosis model building method according to the first aspect.
In a fourth aspect, an electronic device is provided, which includes a memory and a processor, and computer instructions stored in the memory and executed on the processor, wherein the computer instructions, when executed by the processor, implement the rolling bearing fault diagnosis model building method according to the first aspect.
Adopt the produced beneficial effect of above-mentioned technical scheme to lie in:
the embodiment of the invention provides a method and a system for establishing a fault diagnosis model of a rolling bearing, wherein the model establishing method comprises the following steps: the method comprises the steps of acquiring a plurality of vibration signals of the rolling bearing in different vibration states, and firstly, decomposing the vibration signals into a series of BIMF with different frequencies through variation modal decomposition. Variance contribution rates for each BIMF are compared. And converting the BIMF signal with the maximum contribution rate from a one-dimensional time domain signal into a two-dimensional characteristic spectrum through the WVD, thereby obtaining a two-dimensional characteristic matrix of the fault state. And optimizing DBN parameter combinations by using the global optimization capability of the SSA to obtain an optimized DBN, and optimizing the input value of the two-dimensional characteristic matrix to obtain the optimal network structure of the DBN, namely obtaining the rolling bearing fault. By the model establishing method, the signal feature extraction capability can be improved, the problem is solved under the global optimization condition, and the accuracy of fault diagnosis is improved by the established model.
Drawings
Fig. 1 is a time domain waveform diagram of a vibration signal.
Fig. 2 is a time domain diagram of the BIMF component of the outer ring fault vibration signal through the VMD.
Fig. 3 is an FFT spectrum of the outer-loop fault signal.
FIG. 4 is a three-dimensional view of a WVD resolved BIMF.
FIG. 5 is a graph of RMSE as a function of iteration number.
Fig. 6 is a graph of the error variation of the training data.
Fig. 7 is an accuracy chart of a rolling bearing failure diagnosis model.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in detail and fully with reference to the following embodiments.
Example 1
The embodiment of the invention provides a rolling bearing fault diagnosis model establishing method, wherein VMD refers to variational modal decomposition, WVD refers to Wigner-Ville distribution, SSA refers to a sparrow search algorithm, and DBN refers to a deep confidence network.
The method comprises the following steps:
and step S1, acquiring a plurality of normal vibration signals of the rolling bearing in a normal vibration state, a plurality of outer ring fault vibration signals in an outer ring fault vibration state, a plurality of inner ring fault vibration signals in an inner ring fault vibration state and a plurality of rolling body fault vibration signals in a rolling body fault vibration state.
The normal vibration signal, the outer ring fault vibration signal, the inner ring fault vibration signal and the rolling body fault vibration signal can be input by a tester or obtained in real time when the rolling bearing vibrates.
And step S2, extracting the characteristics of each vibration signal to obtain a data set containing a plurality of normal matrixes, a plurality of outer ring fault matrixes, a plurality of inner ring fault matrixes and a plurality of rolling element fault matrixes.
The characteristic extraction of each vibration signal comprises a VMD decomposition process and a process of obtaining a characteristic matrix through WVD.
The VMD is a vibration signal processing method, in an actual test, each vibration signal carries a noise signal, and the noise can be removed through the VDM to obtain a vibration signal meeting the requirement. The VMD may decompose each vibration signal into Band limited intrinsic mode functions (BIMFs) of a plurality of different frequencies around a fixed center frequency and a limited bandwidth.
The method comprises the steps of a construction process and a solution process of the variational problem.
The construction process of the variation problem is as follows:
firstly, a unilateral spectrum of a vibration signal is obtained by Hilbert transform by adopting the following formula:
(δ(t)+j/πt)*uk(t)；
estimating the center frequency of each analysis signal by adopting the following formula, and then modulating the frequency spectrum of each mode to a corresponding estimation base frequency band:
and thirdly, estimating the bandwidth of each BIMF.
The expression of the variational process is as follows:
the constraint condition of the variation process is K modal functions uk(t) the sum of the bandwidths is minimal, and the mathematical expression is as follows:
in the formula, ωkRepresents the center frequency of the modal component, δ (t) is the dirac distribution function, t is time, and represents the convolution.
Wherein, { uk}＝{u1,u2,…,ukIs a set of k bimf. { omega [ [ omega ] ]k}＝ {ω1,ω2,…,ω3Is the center frequency set of each BIMF. Is a convolution symbol.
Is the gradient operator symbol and j is an imaginary unit.
The solution process of the variational problem is to convert the constraint problem into an unconstrained problem. The method introduces quadratic multiplication parameters and Lagrangian operators. The augmented lagrange function is as follows:
in the formula, alpha is a secondary multiplication parameter; λ (t) is the Lagrangian operator; and < > is an internal product operation. And solving the optimal solution of the constraint variation model by using an alternating direction multiplier algorithm.
After obtaining a plurality of BIMFs, the variance contribution ratio of each BIMF is obtained using the following formula:
wherein u isk(j) The original signal is decomposed by VMD to obtain j IMF components, mseb (i) is IMFiN is the signal length.
The variance contribution rate represents the importance of the BIMF component, and the greater the variance contribution rate is, the more important the corresponding BIMF component is. After the variance contribution rate of each BIMF is obtained, the BIMF signal with the maximum contribution rate is converted into a two-dimensional characteristic spectrum from a one-dimensional time domain signal through the WVD, and therefore a characteristic matrix corresponding to the input vibration signal is obtained.
WVD is a typical quadratic time-frequency transform technique, and the Wigner-Ville distribution of one-dimensional bimfx (t) is the fourier transform of its instantaneous autocorrelation function, i.e.:
wherein: x is the number of*(t) is the complex conjugate of x (t), ω represents the center frequency of the modal component, and t is time.
Specifically, since the vibration signal includes four types of normal vibration signal, outer ring fault vibration signal, inner ring fault vibration signal, and rolling element fault vibration signal, this step may include the following steps S201 to S205.
Step S201, for each normal vibration signal, performing time-frequency change sequentially through the VMD and the WVD to obtain a normal matrix.
The method comprises the following steps: for each normal vibration signal, decomposing the normal vibration signal into a plurality of normal BIMFs with different frequencies through VMD, comparing the variance contribution rate of each normal BIMF, and converting the normal BIMF with the maximum contribution rate from a one-dimensional time domain signal into a two-dimensional characteristic spectrum through WVD to obtain the normal matrix.
And S202, sequentially carrying out time-frequency change on each inner ring fault vibration signal through the VMD and the WVD to obtain an outer ring fault matrix.
And S203, sequentially carrying out time-frequency change on each rolling element fault vibration signal through VMD and WVD to obtain a rolling element fault matrix.
Step S202 and step S203 are similar to step S201, and only differ in the type of signal processed by the VMD, which is not specifically limited in the embodiment of the present invention.
And S204, storing all the normal matrix, the outer ring fault matrix, the inner ring fault matrix and the rolling body fault matrix into a data set.
And step S3, optimizing the parameters of the DBN by using the global optimization capability of the SSA, and obtaining the optimized DBN by using the learning rate and the batch size as the optimization target of the SSA.
It includes:
step S301, setting all preset parameters of SSA, initializing the maximum iteration times and searching the preset range, wherein the preset parameters comprise the preset sparrow number, the preset danger sensing sparrow number and the preset warning value.
The Sparrow Search Algorithm (SSA) is mathematically described as follows: there is a d-dimensional search space in which there are n sparrows, divided into producers and predators throughout the sparrow population, which are responsible for different tasks in the population.
Producers are responsible for foraging and directing the entire colony to more places on the food. Predators constantly monitor producers and borrow food by machine. In the population, the identity of producers and predators is dynamically changing, and each sparrow has the opportunity to become a producer, as long as it finds more abundant food. However, the ratio of producers to predators is constant, with high-yielding predators becoming producers and with laggard producers becoming predators.
In addition, natural enemies exist in biological groups, sparrows occupying 10% -20% of the total number of the sparrow groups have warning consciousness, enemies can be found out and alarms can be given out, and when the warning value exceeds a safety threshold value, the whole group can give up food to a safer place in time.
In the invention, when the sparrow algorithm is used, all preset parameters of the SSA need to be set, and the maximum iteration times and the preset search range need to be initialized.
And S302, initializing a sparrow position and calculating the fitness, and taking the Root Mean Square Error (RMSE) of the network of the DBN training data to be optimized as the fitness function of the SSA.
For example: there is a d-dimensional search space in which there are n sparrows, divided into producers and predators, in the entire sparrow population, and the positional information of the sparrow population can be represented by the array X:
for any sparrow, the position information can be represented by vector X [ alpha, beta ]]Is expressed and based on a vector X [ alpha, beta ]]Obtaining the initial fitness of each sparrow, and storing the fitness of all the sparrows into a vector FxAnd (3) lining:
the learning rate of the DBN and the value range of the batch size determine the search space of sparrows. The root mean square error minimization of the DBN serves as an optimization indicator for SSA. Each iteration process can generate an optimal combined solution, and the combined solution is put into network training to obtain the network root mean square error of the iteration. According to the SSA optimization mechanism, the network root mean square error is reduced along with the increase of the iteration times, and finally, the optimal combination of the network learning rate and the batch size under the minimum root mean square error is obtained.
Step S303, comparing the estimated fitness of each sparrow individual to obtain the optimal position and the worst position of each sparrow, wherein the position of the sparrow with the minimum fitness is the current global optimal position XbestAnd the position of the sparrow with the largest fitness is the global worst position Xworst；
And S304, when the current optimization iteration times reach the maximum iteration times, ending the iteration and outputting the optimal position, and training the DBN to be optimized by using the optimal position to obtain the optimized DBN with the optimal learning rate and the optimal batch size.
Wherein the final position is the global optimum position X obtained in step S303bestCan pass through its corresponding vector Xbest[α*,β*]And training the DBN to be optimized to obtain the optimized DBN with the optimal learning rate and the optimal batch size.
Further, the steps further include:
and S305, when the current optimization iteration frequency is judged not to reach the maximum iteration frequency, updating the position and the fitness of each sparrow individual, re-initializing, and setting a new search range, wherein the new search range is out of the search preset range.
As for the way of updating the position of each sparrow individual, the following method may be employed:
in the population, producers occupy 10% -20% of the total population, and have higher adaptability and stronger foraging capacity. Thus, a greater extent of foraging is obtained. During each iteration, the location update of the producer refers to the following equation.
Where t (t ═ 1,2, …, Itera) is the number of iterations of the optimization, itermaxIs a fixed value representing the maximum number of iterations. Both α and Q are random numbers (α ∈ (0, 1)]、Q～N(μ,σ2)). L represents a matrix of size 1 × D and having elements all 1. R2Indicates a warning value, ST indicates a safety threshold value whenR2And when ST is more than or equal to ST, the sparrow group needs to be immediately migrated to a safe area to avoid an intruder.
Thus, the predator performs a position update according to the following formula:
in the formula, XPIndicating the optimal location in the producer. XworstIndicating the worst position of the current range. When i > n/2, it indicates that the ith predator has poor fitness and is likely to be starving.
And the sparrows with warning ability update the positions according to the following formula:
when f isi＝fgIt is shown that a sparrow finds an enemy and is close to other sparrows to avoid attack by the enemy.
S306, taking the current optimal position as a parameter value of a DBN training part to obtain an initial optimized DBN;
i.e. the global optimum position X obtained by step S303bestCorresponding vector Xbest[α*,β*]As parameter values for the DBN training portion.
S307, comparing the estimated fitness of each sparrow individual to obtain the optimal position of each sparrow;
and step S308, judging whether the current optimization iteration number reaches the maximum iteration number, and executing step S304 or step S305, thereby obtaining the optimized DBN.
In other words, in the process, the optimized DBN is obtained by taking the learning rate and the batch size as the optimization targets of the sparrow search algorithm.
And S4, inputting the data set into the optimized DBN for training to obtain the rolling bearing fault diagnosis model.
The method comprises the following steps:
step S401, dividing the data set into a plurality of test sets and a plurality of test sets according to the vibration state, wherein each training set and each test set carry a label for indicating the vibration state; the vibration state comprises a normal vibration state, an outer ring fault vibration state, an inner ring fault vibration state and a rolling body fault vibration state;
s402, inputting a plurality of test sets into the optimized DBN for training to obtain a trained DBN;
step S403, inputting a plurality of test sets into the trained DBN to test the classification effect of the trained DBN;
and S404, when the classification effect is determined to reach the preset effect, taking the trained DBN as a fault diagnosis model of the rolling bearing.
Hereinafter, the feature extraction section of step S2 and the generation of optimized DBN in step S3 in the above rolling bearing failure diagnosis model will be described with a specific embodiment.
A feature extraction section:
(1) experimental data show that:
vibration signals from the bearing center of the electrical engineering laboratory at the university of kesy reservoir were used as data to verify the validity of the proposed model in the diagnosis of rolling bearing faults.
The bearing test bed comprises a driving motor, a load motor, an acceleration sensor and a torque sensor. The acceleration sensor collected data at four different loads (load 0-0 HP/1797rpm, load 1-1 HP/1772 rpm, load 2-2 HP/1750rpm, and load 3-3 HP/1730 rpm). The sampling frequency was 12 kHz. The failure types include inner ring failure, outer ring failure, and rolling body failure. The rolling element failure is inside the bearing, and the failure point is not visible. Each fault type contains four fault diameters, "0.007", "0.014" and "0.021". The fault diameter represents the depth of the fault.
The method adopts a plurality of outer ring fault vibration signals with the fault diameter of 0.014, the motor load is 3HP, the motor rotating speed is 1730rpm, and a vibration signal time domain waveform diagram is shown in figure 1.
(2) Vibration signal feature extraction
The outer ring fault vibration signals are subjected to feature extraction, fig. 2 is a plurality of BIMF time domain diagrams obtained by performing variation modal decomposition on the plurality of outer ring fault signals, and fig. 3 is a frequency domain diagram obtained by performing fourier transform on each BIMF.
To select the key BIMF, the variance contribution rate of each BIMF to the vibration signal at each fault condition is calculated. The mean value of the variance contribution of each BIMF component was obtained as shown in table 1. The mean variance contribution of each BIMF represents the contribution of the BIMF to the original vibration signal characteristics. The larger the mean variance contribution rate is, the closer the features of the BIMF are to those of the noise-free original signal, and it is found through calculation that the influence of the fourth BIMF on the vibration signal is large, so that the WVD is performed on the fourth BIMF, and a three-dimensional spectrogram of the fourth BIMF after the WVD transformation is shown in fig. 4.
TABLE 1 variance contribution ratio of BIMF
Optimizing the generation of the DBN:
the DBN is adopted in the application, and the network structure is 100-100. The SSA is used to search for a globally optimal solution of both the learning rate and the batch size of the DBN. The search range of the learning rate is: [ 0.011 ], the search range for batch size is: [190]. The parameter settings for SSA are shown in table 2. FIG. 5 shows the RMSE of the test data as the number of iterations increases. After 5 iterations, the RMSE starts to converge, indicating that the algorithm has a faster convergence rate. After the 10 th iteration, the RMSE of the test data converged to 0.095. After the 15 th iteration, the RMSE is stably converged to 0.085, which shows that the optimization algorithm for optimizing the DBN by adopting the SSA has good global optimization capability and strong robustness. Finally, the best combination of learning rate and batch size is: [0.1673,4].
TABLE 2 SSA parameter settings
The structural parameters of the DBN are shown in table 3. The number of nodes of the input layer and the output layer is determined by the input sample and the fault category respectively. The DBN parameter combination settings obtained after global optimization by SSA are shown in table 4.
TABLE 3 parameters of DBN Structure
TABLE 4 setting of DBN parameter combinations
Further, the accuracy and precision of the obtained fault diagnosis model are tested, as shown in fig. 6 and fig. 7, as can be seen from fig. 6, at the 200 th iteration, the fault diagnosis model starts to converge stably in the region, indicating that the model is well trained. After 500 iterations, the error converged to 0.0201, so the accuracy of the training data was 98% and the accuracy of the test data was 91.5%.
As can be seen from fig. 7, after 10 repetitions, the recognition rates of 10 diagnoses are 98%, 97.7%, 98.2%, 97.8%, 97.67%, 98%, 97.7%, 97.98%, 98% and 98%, respectively, and the average accuracy is 98%, which indicates that the model has strong robustness and high accuracy for classifying rolling bearing faults.
Example 2
The invention provides a rolling bearing fault diagnosis model establishing system, which comprises:
the device comprises an acquisition module, a control module and a control module, wherein the acquisition module is used for acquiring a plurality of normal vibration signals of the rolling bearing in a normal vibration state, a plurality of outer ring fault vibration signals in an outer ring fault vibration state, a plurality of inner ring fault vibration signals in an inner ring fault vibration state and a plurality of rolling body fault vibration signals in a rolling body fault vibration state;
the data set generating module is used for extracting the characteristics of each vibration signal to obtain a data set comprising a plurality of normal matrixes, a plurality of outer ring fault matrixes, a plurality of inner ring fault matrixes and a plurality of rolling element fault matrixes;
the optimization module is used for optimizing the parameters of the DBN by utilizing the global optimization capability of the SSA to obtain an optimized DBN;
and the model generation module is used for inputting the data set into the optimized DBN for training to obtain the rolling bearing fault diagnosis model.
Example 3
An embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the following steps:
acquiring a plurality of normal vibration signals of a rolling bearing in a normal vibration state, a plurality of outer ring fault vibration signals in an outer ring fault vibration state, a plurality of inner ring fault vibration signals in an inner ring fault vibration state and a plurality of rolling body fault vibration signals in a rolling body fault vibration state;
extracting the characteristics of each vibration signal to obtain a data set containing a plurality of normal matrixes, a plurality of outer ring fault matrixes, a plurality of inner ring fault matrixes and a plurality of rolling element fault matrixes;
optimizing the parameters of the DBN by using the global optimization capability of the SSA to obtain an optimized DBN;
and inputting the data set into the optimized DBN for training to obtain the fault diagnosis model of the rolling bearing.
Example 4
The present embodiment provides a processing apparatus, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the processor implements the following steps:
acquiring a plurality of normal vibration signals of a rolling bearing in a normal vibration state, a plurality of outer ring fault vibration signals in an outer ring fault vibration state, a plurality of inner ring fault vibration signals in an inner ring fault vibration state and a plurality of rolling body fault vibration signals in a rolling body fault vibration state;
extracting the characteristics of each vibration signal to obtain a data set containing a plurality of normal matrixes, a plurality of outer ring fault matrixes, a plurality of inner ring fault matrixes and a plurality of rolling element fault matrixes;
optimizing the parameters of the DBN by using the global optimization capability of the SSA to obtain an optimized DBN;
and inputting the data set into the optimized DBN for training to obtain the fault diagnosis model of the rolling bearing.
Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art may still make modifications to the technical solutions described in the foregoing embodiments, or may substitute some technical features of the embodiments; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. A method for establishing a fault diagnosis model of a rolling bearing, which is characterized by comprising the following steps:
acquiring a plurality of normal vibration signals of a rolling bearing in a normal vibration state, a plurality of outer ring fault vibration signals in an outer ring fault vibration state, a plurality of inner ring fault vibration signals in an inner ring fault vibration state and a plurality of rolling body fault vibration signals in a rolling body fault vibration state;
extracting the characteristics of each vibration signal to obtain a data set containing a plurality of normal matrixes, a plurality of outer ring fault matrixes, a plurality of inner ring fault matrixes and a plurality of rolling element fault matrixes;
optimizing the parameters of the DBN by using the global optimization capability of the SSA to obtain an optimized DBN;
and inputting the data set into the optimized DBN for training to obtain the fault diagnosis model of the rolling bearing.
2. The method for establishing the fault diagnosis model of the rolling bearing according to claim 1, wherein the step of performing feature extraction on each vibration signal to obtain a data set comprising a plurality of normal matrixes, a plurality of outer ring fault matrixes, a plurality of inner ring fault matrixes and a plurality of rolling element fault matrixes comprises the following steps:
for each normal vibration signal, performing time-frequency change sequentially through the VMD and the WVD to obtain a normal matrix;
for each inner ring fault vibration signal, performing time-frequency change sequentially through the VMD and the WVD to obtain an outer ring fault matrix;
for each inner ring fault vibration signal, carrying out time-frequency change sequentially through the VMD and the WVD to obtain an inner ring fault matrix;
for each rolling element fault vibration signal, performing time-frequency change sequentially through VMD and WVD to obtain a rolling element fault matrix;
and storing all the normal matrix, the outer ring fault matrix, the inner ring fault matrix and the rolling body fault matrix into a data set.
3. The method for establishing the rolling bearing fault diagnosis model according to claim 2, wherein the obtaining of the normal matrix after time-frequency change sequentially by the VMD and the WVD for each normal vibration signal comprises:
for each normal vibration signal, decomposing the normal vibration signal into a plurality of normal BIMFs with different frequencies through VMD, comparing the variance contribution rate of each normal BIMF, and converting the normal BIMF with the maximum contribution rate from a one-dimensional time domain signal into a two-dimensional characteristic spectrum through WVD to obtain the normal matrix.
4. The method for establishing the fault diagnosis model of the rolling bearing according to claim 3, wherein the variance contribution rate of each normal BIMF is compared by adopting the following formula:
wherein u isk(j) The original signal is decomposed by VMD to obtain j IMF components, mseb (i) is IMFiN is the signal length.
5. The method for building the rolling bearing fault diagnosis model according to claim 1, wherein the optimizing the DBN parameters by using the global optimization capability of the SSA comprises:
s301, setting all preset parameters of SSA, initializing the maximum iteration times and searching a preset range, wherein the preset parameters comprise a preset sparrow number, a preset danger sensing sparrow number and a preset warning value;
s302, initializing a sparrow position and calculating the fitness, and taking the RMSE of DBN training data to be optimized as a fitness function of the SSA;
s303, comparing the estimated fitness of each sparrow individual to obtain the optimal position and the worst position of each sparrow;
s304, when the current optimization iteration times reach the maximum iteration times, ending the iteration and outputting the optimal position, and training the DBN to be optimized by using the optimal position to obtain the optimized DBN with the optimal learning rate and the optimal batch size.
6. The rolling bearing fault diagnosis model building method according to claim 5, characterized by further comprising the steps of:
s305, when the current optimization iteration frequency is judged not to reach the maximum iteration frequency, updating the position and the fitness of each sparrow individual, re-initializing, and setting a new search range, wherein the new search range is out of the search preset range;
s306, taking the current optimal position as a parameter value of a DBN training part to obtain an initial optimized DBN;
s307, comparing the estimated fitness of each sparrow individual to obtain the optimal position of each sparrow;
and S308, judging whether the current optimization iteration number reaches the maximum iteration number, and executing the step S304 or the step S305 to obtain the optimized DBN.
7. The method for establishing the rolling bearing fault diagnosis model according to claim 5, wherein the inputting the data set into the optimized DBN for training to obtain the rolling bearing fault diagnosis model comprises:
dividing the data set into a plurality of test sets and a plurality of test sets according to the vibration state, wherein each training set and each test set carry a label for indicating the vibration state;
inputting a plurality of test sets into the optimized DBN for training to obtain a trained DBN;
inputting a plurality of test sets into the trained DBN to test the classification effect of the trained DBN;
and when the classification effect is determined to reach the preset effect, taking the trained DBN as a fault diagnosis model of the rolling bearing.
8. A rolling bearing fault diagnosis model establishment system is characterized by comprising:
the device comprises an acquisition module, a control module and a control module, wherein the acquisition module is used for acquiring a plurality of normal vibration signals of the rolling bearing in a normal vibration state, a plurality of outer ring fault vibration signals in an outer ring fault vibration state, a plurality of inner ring fault vibration signals in an inner ring fault vibration state and a plurality of rolling body fault vibration signals in a rolling body fault vibration state;
the data set generating module is used for extracting the characteristics of each vibration signal to obtain a data set comprising a plurality of normal matrixes, a plurality of outer ring fault matrixes, a plurality of inner ring fault matrixes and a plurality of rolling element fault matrixes;
the optimization module is used for optimizing the parameters of the DBN by utilizing the global optimization capability of the SSA to obtain an optimized DBN;
and the model generation module is used for inputting the data set into the optimized DBN for training to obtain the rolling bearing fault diagnosis model.
9. A computer-readable storage medium for storing computer instructions, wherein the computer instructions, when executed by a processor, perform a rolling bearing fault diagnosis model building method according to any one of claims 1 to 7.
10. An electronic device comprising a memory and a processor and computer instructions stored on the memory and executed on the processor, wherein the computer instructions, when executed by the processor, perform the rolling bearing fault diagnosis model building method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111498873.6A CN114201830A (en) | 2021-12-09 | 2021-12-09 | Rolling bearing fault diagnosis model establishing method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111498873.6A CN114201830A (en) | 2021-12-09 | 2021-12-09 | Rolling bearing fault diagnosis model establishing method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114201830A true CN114201830A (en) | 2022-03-18 |
Family
ID=80651600
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111498873.6A Pending CN114201830A (en) | 2021-12-09 | 2021-12-09 | Rolling bearing fault diagnosis model establishing method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114201830A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115901248A (en) * | 2022-10-18 | 2023-04-04 | 江苏新道格自控科技有限公司 | Gearbox fault feature extraction method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200210826A1 (en) * | 2018-12-29 | 2020-07-02 | Northeastern University | Intelligent analysis system using magnetic flux leakage data in pipeline inner inspection |
| CN112733603A (en) * | 2020-12-11 | 2021-04-30 | 江苏大学 | Frequency conversion scroll compressor fault diagnosis method based on improved VMD and SVM |
| CN114964778A (en) * | 2022-05-13 | 2022-08-30 | 重庆理工大学 | Bearing fault diagnosis method based on wavelet time-frequency graph and deep learning |
-
2021
- 2021-12-09 CN CN202111498873.6A patent/CN114201830A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200210826A1 (en) * | 2018-12-29 | 2020-07-02 | Northeastern University | Intelligent analysis system using magnetic flux leakage data in pipeline inner inspection |
| CN112733603A (en) * | 2020-12-11 | 2021-04-30 | 江苏大学 | Frequency conversion scroll compressor fault diagnosis method based on improved VMD and SVM |
| CN114964778A (en) * | 2022-05-13 | 2022-08-30 | 重庆理工大学 | Bearing fault diagnosis method based on wavelet time-frequency graph and deep learning |
Non-Patent Citations (2)
| Title |
|---|
| 盛肖炜等: "基于参数优化VMD 和改进DBN 的滚动轴承故障诊断方法研究", 机电工程, vol. 38, no. 9, 20 September 2021 (2021-09-20), pages 1107 - 1116 * |
| 高佳程等: "基于VMD-WVD分布与堆栈稀疏自编码网络的局放类型识别", 中国电机工程学报, vol. 39, no. 14, 20 July 2019 (2019-07-20), pages 4118 - 4129 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115901248A (en) * | 2022-10-18 | 2023-04-04 | 江苏新道格自控科技有限公司 | Gearbox fault feature extraction method |
| CN115901248B (en) * | 2022-10-18 | 2023-09-19 | 江苏新道格自控科技有限公司 | Gear box fault feature extraction method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110276416B (en) | Rolling bearing fault prediction method | |
| CN112183581A (en) | Semi-supervised mechanical fault diagnosis method based on self-adaptive migration neural network | |
| CN109186964B (en) | rotary machine fault diagnosis method based on angle resampling and ROC-SVM | |
| Pandya et al. | Fault diagnosis of rolling element bearing by using multinomial logistic regression and wavelet packet transform | |
| Sun et al. | Fast bearing fault diagnosis of rolling element using Lévy Moth-Flame optimization algorithm and Naive Bayes | |
| Liao et al. | Hydroelectric generating unit fault diagnosis using 1-D convolutional neural network and gated recurrent unit in small hydro | |
| CN111428418A (en) | Bearing fault diagnosis method and device, computer equipment and storage medium | |
| CN111238807A (en) | Fault diagnosis method for planetary gear box | |
| CN114964778A (en) | Bearing fault diagnosis method based on wavelet time-frequency graph and deep learning | |
| Ye et al. | Multiscale weighted morphological network based feature learning of vibration signals for machinery fault diagnosis | |
| CN111387975A (en) | Electroencephalogram signal identification method based on machine learning | |
| CN111060316A (en) | Rolling bearing state monitoring method and system based on convolutional neural network model | |
| CN114201830A (en) | Rolling bearing fault diagnosis model establishing method and system | |
| Zhao et al. | A novel deep fuzzy clustering neural network model and its application in rolling bearing fault recognition | |
| CN112945557A (en) | Slewing bearing fault diagnosis method and device and storage medium | |
| Ye et al. | Deep negative correlation multisource domains adaptation network for machinery fault diagnosis under different working conditions | |
| Kavianpour et al. | An intelligent gearbox fault diagnosis under different operating conditions using adversarial domain adaptation | |
| CN112069918A (en) | Fault diagnosis method and device for planetary gearbox | |
| Du et al. | Fault diagnosis of plunger pump in truck crane based on relevance vector machine with particle swarm optimization algorithm | |
| CN116881841A (en) | Hybrid model fault diagnosis method based on F1-score multistage decision analysis | |
| Hu et al. | On the use of EEMD and SVM based approach for bearing fault diagnosis of wind turbine gearbox | |
| Zhang et al. | Rolling Bearings Fault Diagnosis under Variable Conditions Using RCMFE and Improved Support Vector Machine. | |
| Zhang et al. | Gearbox health condition identification by neuro-fuzzy ensemble | |
| CN116481811A (en) | Fault feature extraction method based on GWO optimized SVMD | |
| Jia et al. | Intelligent Fault Severity Detection of Rotating Machines Based on VMD‐WVD and Parameter‐Optimized DBN |
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 |