CN116735206A - Bearing friction state identification method based on attractor and convolutional neural network - Google Patents
Bearing friction state identification method based on attractor and convolutional neural network Download PDFInfo
- Publication number
- CN116735206A CN116735206A CN202310771744.2A CN202310771744A CN116735206A CN 116735206 A CN116735206 A CN 116735206A CN 202310771744 A CN202310771744 A CN 202310771744A CN 116735206 A CN116735206 A CN 116735206A
- Authority
- CN
- China
- Prior art keywords
- friction state
- attractor
- neural network
- convolutional neural
- bearing
- 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
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 35
- 238000000034 method Methods 0.000 title claims abstract description 32
- 238000012549 training Methods 0.000 claims abstract description 26
- 230000001133 acceleration Effects 0.000 claims abstract description 21
- 238000012360 testing method Methods 0.000 claims abstract description 21
- 239000007788 liquid Substances 0.000 claims abstract description 10
- 230000009467 reduction Effects 0.000 claims abstract description 10
- 239000011159 matrix material Substances 0.000 claims description 5
- 238000005070 sampling Methods 0.000 claims description 4
- 239000013598 vector Substances 0.000 claims description 4
- 230000003111 delayed effect Effects 0.000 claims description 3
- 238000011176 pooling Methods 0.000 description 9
- 238000010586 diagram Methods 0.000 description 4
- 239000003921 oil Substances 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 238000003745 diagnosis Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000007789 sealing Methods 0.000 description 2
- 101150013553 CD40 gene Proteins 0.000 description 1
- 102100040245 Tumor necrosis factor receptor superfamily member 5 Human genes 0.000 description 1
- 238000005299 abrasion Methods 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000001066 destructive effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000036039 immunity Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 239000010687 lubricating oil Substances 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/10—Pre-processing; Data cleansing
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M13/00—Testing of machine parts
- G01M13/04—Bearings
- G01M13/045—Acoustic or vibration analysis
-
- 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
- G06F18/211—Selection of the most significant subset of features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/771—Feature selection, e.g. selecting representative features from a multi-dimensional feature space
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2123/00—Data types
- G06F2123/02—Data types in the time domain, e.g. time-series data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
- G06F2218/06—Denoising by applying a scale-space analysis, e.g. using wavelet analysis
-
- 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
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Databases & Information Systems (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Multimedia (AREA)
- Acoustics & Sound (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a bearing friction state identification method based on an attractor and a convolutional neural network, which comprises the following specific steps of S1: collecting acceleration vibration signals of a bearing seat of the sliding bearing test bed in a liquid friction state, a mixed friction state and a dry friction state respectively; s2: noise reduction is carried out on the acquired acceleration vibration signals, and then a three-dimensional attractor graph is generated through a phase space reconstruction method; s3: projecting the three-dimensional attractor graph into a two-dimensional attractor graph, and normalizing the two-dimensional attractor graph to form an original image set; s4: building a convolutional neural network model and training based on an original image set; s5: and carrying out bearing friction state identification on the vibration signal to be detected based on the trained convolutional neural network model so as to determine the corresponding friction state. By applying the attractor correlation method to the bearing friction state identification, the nonlinear characteristic difference of the bearing in different friction states can be revealed, so that a basis is provided for the identification of the bearing friction state.
Description
Technical Field
The invention relates to the technical field of image recognition, in particular to a bearing friction state recognition method based on attractors and a convolutional neural network.
Background
The ship plays an extremely important role as the most important transportation means at sea and in both the military and civil applications. Under the condition that the marine ship is in a vertically and horizontally inclined swinging environment, the friction and abrasion among all parts of a rotor system in a propulsion shaft system are increased along with the change of the ship power plant equipment in the working process, so that destructive friction is easier to occur, the operation efficiency of the ship power plant equipment is reduced once the friction occurs, and the safety and the reliability of the ship power plant equipment are greatly influenced. Therefore, the friction state of the ship propulsion shafting bearing is accurately identified, which has important significance for ensuring the safe operation of the ship,
the running state of the shafting system is usually judged by manual reasoning according to professional experience by experts in each field, and the occurrence of shallow intelligent models such as a neural network, a support vector machine and the like replaces the traditional manual diagnosis process, so that the intelligent age of fault diagnosis is started. However, as the propulsion shafting system has serious nonlinear vibration sources such as oil film force, sealing force, airflow exciting force and the like, a plurality of nonlinear problems exist, and the problem to be solved is how to accurately identify the running state of the shafting system under the condition that the shafting system has the nonlinear problems.
Disclosure of Invention
The invention provides a bearing friction state identification method based on an attractor and a convolutional neural network, which aims to solve the technical problem of accurately identifying the running state of a shafting system under the condition that the shafting system has nonlinear problems.
In order to achieve the above object, the technical scheme of the present invention is as follows:
a bearing friction state identification method based on attractors and convolutional neural networks comprises the following specific steps:
s1: collecting acceleration vibration signals of a bearing seat of the sliding bearing test bed in a liquid friction state, a mixed friction state and a dry friction state respectively;
s2: noise reduction is carried out on the acquired acceleration vibration signals, and then a three-dimensional attractor graph is generated through a phase space reconstruction method;
s3: projecting the three-dimensional attractor graph into a two-dimensional attractor graph, and normalizing the two-dimensional attractor graph to form an original image set;
s4: building a convolutional neural network model and training based on an original image set;
s5: and carrying out bearing friction state identification on the vibration signal to be detected based on the trained convolutional neural network model so as to determine the corresponding friction state.
Specifically, in the step S2,
s21: decomposing the acquired acceleration vibration signals by a harmonic wavelet packet method and carrying out noise reduction treatment;
s22: determining parameters of phase space reconstruction: embedding dimension m and delay timeτ, defining a time sequence x= { x of acceleration vibration signals i I=1, 2, …, N }, constructing a batch of phase space matrices X, x= [ X ] based on the embedding dimension m and the delay time τ i ,X i+τ ,X i+2τ ,…,X i+(m-1)τ ](i=1, 2, …, M), N being the time series length; x is X i Is the ith phase space matrix; m=n- (M-1) τ is the number of phase space matrices, and M-dimensional vectors are all points X of reconstructed phase space 1 ,X 2 ,…,X M The method comprises the steps of carrying out a first treatment on the surface of the And reconstructing the acceleration vibration signal attractor after noise reduction treatment according to the selected embedding dimension m and the delay time tau to obtain a three-dimensional attractor.
Specifically, in the step S3, the root mean square of each two coordinate axes in the three-dimensional attractor graph is calculated, and the two coordinate axes with the largest root mean square are taken as the two-dimensional attractor graph coordinate axes, and the two-dimensional attractor graph is generated by projection.
Specifically, in the step S4,
s41: classifying the original image set according to the liquid friction state, the mixed friction state and the dry friction state, setting corresponding labels, constructing and forming a training set, and randomly extracting a part of images from the original image set to construct a test set;
s42: building a convolutional neural network model, and training the convolutional neural network model by using a training set;
s43: and (3) evaluating the accuracy of the test set by using the trained convolutional neural network model, and if the difference between the test result of the test set and the training result of the training set exceeds a threshold value, retraining.
Specifically, in S22, the embedding dimension m and the delay time τ are selected by the C-C method, the associated integral C (m, N, r, t) is defined first, the statistic S (m, N, r, t) is reconstructed, r represents the distance between the points in the phase space, and t is the local maximum time;
definition τ s For time series sampling interval, τ is delay time, then time series is delayed by τ d =ττ s Delay time window τ w =(m-1)τ d Calculating time series delay tau from statistics d Further determining the delay time tau and thenFrom τ w 、τ d The embedding dimension m is determined.
The beneficial effects are that: according to the invention, the collected acceleration vibration signals are subjected to phase space reconstruction to generate the attractor two-dimensional projection diagram, the attractor two-dimensional projection diagram forms a training set, the training set is trained by utilizing the convolutional neural network model, and the successfully trained convolutional neural network model can identify the friction state of the bearing according to the two-dimensional attractor diagram generated by the bearing motion in practice, so that the problem of friction state identification of a shafting system when the nonlinear problem exists is solved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions of the prior art, the drawings that are needed in the embodiments or the description of the prior art will be briefly described below, it will be obvious that the drawings in the following description are some embodiments of the present invention, and that other drawings can be obtained according to these drawings without inventive effort to a person skilled in the art.
FIG. 1 is a step diagram of a method for identifying the friction state of a bearing based on an attractor and a convolutional neural network in the invention;
FIG. 2 is a three-dimensional attractor graph in the state of mixed friction, dry friction and liquid friction after phase space reconstruction in the present invention;
FIG. 3 is a two-dimensional attractor graph of the present invention in a state of mixed friction, dry friction and liquid friction generated by projection.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The propulsion shafting system has serious nonlinear vibration sources such as oil film force, sealing force, airflow exciting force and the like, so that the system has a plurality of nonlinear problems, and particularly the system is more prominent when faults such as misalignment, friction, looseness and the like occur. Therefore, nonlinear theory and method are needed to be adopted, so that the research result meets the actual requirement of engineering, and the attractor related theory and method are applied to the research of monitoring and identifying the bearing system, so that the nonlinear characteristic difference of the bearing under different states can be revealed, and a basis is provided for state identification. Applicants' studies found that: the intelligent recognition of the friction state of the sliding bearing can be realized through the convolutional neural network self-learning technology, the defects of poor noise immunity, low precision and the like of the conventional recognition algorithm are overcome, the defect of low recognition efficiency due to manual experience can be overcome, and the recognition efficiency of different friction states of the sliding bearing can be effectively improved.
The embodiment provides a bearing friction state identification method based on an attractor and a convolutional neural network, which takes a shafting test bed as an example, as shown in fig. 1, and comprises the following specific steps:
s1: collecting acceleration vibration signals of a bearing seat of the sliding bearing test bed in a liquid friction state, a mixed friction state and a dry friction state respectively;
s2: noise reduction is carried out on the acquired acceleration vibration signals, and then a three-dimensional attractor graph is generated through a phase space reconstruction method;
s3: projecting the three-dimensional attractor graph into a two-dimensional attractor graph, and normalizing the two-dimensional attractor graph to form an original image set;
s4: building a convolutional neural network model and training based on an original image set;
s5: and carrying out bearing friction state identification on the vibration signal to be detected based on the trained convolutional neural network model so as to determine the corresponding friction state.
The nonlinear characteristic difference of the bearing under different friction states can be revealed by applying the related theory and method of the attractors to the research of monitoring and identifying the bearing system, thereby providing a basis for identifying the friction state of the bearing.
In a specific embodiment, the step S1: the invention uses an acceleration sensor arranged on a bearing seat to collect vibration signals under different running states of a shafting, adjusts the rotating speed of a main shaft through a stepless speed regulator in a control box, thereby changing the thickness of an oil film between a main journal of a sliding bearing and a bearing bush, selects CD40 lubricating oil to lubricate the main journal and the bearing bush, gradually increases the rotating speed of the main shaft from 15r/min to 240r/min by using the stepless speed regulator under the condition of ensuring that the load is unchanged under the condition of ensuring that the positive pressure applied by a test is 700N, runs for 10min under each rotating speed, and collects acceleration vibration signals of the bearing seat of a sliding bearing test bed under the liquid friction state, the mixed friction state and the dry friction state respectively;
in a specific embodiment, the step S2: the acquired acceleration vibration signals are subjected to noise reduction, and then a three-dimensional attractor graph shown in figure 2 is generated through a phase space reconstruction method;
s21: decomposing the acquired acceleration vibration signals into 7 layers by a harmonic wavelet packet method, and carrying out noise reduction treatment on 64 frequency bands;
s22: determining parameters of phase space reconstruction: the embedding dimension m and the delay time tau,
specifically, an embedding dimension m and a delay time tau are selected through a C-C method, an associated integral C (m, N, r, t) is defined, statistics S (m, N, r, t) are reconstructed, r represents the distance between points in a phase space, and t is the local maximum time;
definition τ s For time series sampling interval, τ is delay time, then time series is delayed by τ d =ττ s Delay time window τ w =(m-1)τ d Calculating time series delay tau from statistics d Further determining the delay time τ, and then determining the delay time τ w 、τ d Determining an embedding dimension m;
specifically, the correlation integral C (m, N, r, t) of the time series is
Where θ (·) is the Heaviside unit function,
X i is the i-th phase space momentArray, X j For the j-th phase space matrix, if x is less than 0, θ (= 0); if x is greater than or equal to 0, θ (·) =1, the associated integral function may represent the probability that any two points in the phase space are less than r apart;
time sequence { x } i Dividing into t disjoint sub-sequences, defining each sub-sequence statistic
Wherein: c (C) s For the associated integration of the s-th sub-sequence,the associated integral to the m th power of the s-th subsequence;
let N → ≡ be:
the difference is defined according to r:
ΔS(m,t)=max{S(m,r,t)}-min{S(m,r,t)}(4)
the local maximum time t corresponds to the zero point of S (m, r, t) or the minimum value of DeltaS (m, t), the time series delay τ d Corresponding to the first local time, at the moment, the reconstructed space points are closest to be uniformly distributed, and the reconstructed attractor orbit is completely unfolded in the phase space; determining a time series delay tau from a first local time d And then according to tau d =ττ s Obtaining a delay time tau;
the delay time window selection requirements S (m, r, t) and DeltaS (m, t) both approach 0, thus defining an index S cor (t) is:
wherein:for the average value of Δs (m, t) corresponding to the different m and r, +.>An average value of S (m, r, t) corresponding to different m and r;
finding the global minimum point to obtain the delay time window tau w Is then determined by τ w =(m-1)τ d Obtaining the value range of m and r to be more than or equal to 2 and less than or equal to 5,and sigma is the standard deviation of the time series, and m and t can be obtained according to the formulas (3) to (5).
Specifically, a batch of phase space matrices X, x= [ X ] is constructed from the embedding dimension m and the delay time τ i ,X i+τ ,X i+2τ ,…,X i+(m-1)τ ],(i=1,2,…,M),
Defining a time sequence x= { x of acceleration vibration signals i I=1, 2, …, N }, N being the length of the time series; x is X i Is the ith phase space matrix; m=n- (M-1) τ is the number of phase space matrices, and M-dimensional vectors are all points X of reconstructed phase space 1 ,X 2 ,…,X M And reconstructing a phase space of a time sequence data set X (t) of the acceleration vibration signal according to the selected embedding dimension m and delay time tau, and constructing a three-dimensional attractor graph by taking X (t) in the reconstructed phase space matrix X as an abscissa and X (t+tau) as an ordinate.
In a specific embodiment, the step S3: projecting the three-dimensional attractor graph into a two-dimensional attractor graph shown in fig. 3, and normalizing the two-dimensional attractor graph to form an original image set;
specifically, root mean square of every two coordinate axes in the three-dimensional attractor graph is calculated, two coordinate axes with the largest root mean square are taken as two-dimensional attractor graph coordinate axes, the two-dimensional attractor graph is generated through projection, and the horizontal and vertical coordinates of the two-dimensional attractor graph are (-6- +6) and are saved as an original image set.
In a specific embodiment, the step S4: building a convolutional neural network model and training based on an original image set;
s41: classifying the original image set according to the liquid friction state, the mixed friction state and the dry friction state, setting corresponding labels, constructing and forming a training set, and randomly extracting 30% of the original images from the original image set to construct a test set.
S42: building a convolutional neural network model in a MATLAB deep learning editor, and training the convolutional neural network model by using a training set;
s43: and performing accuracy rate assessment on the trained convolutional neural network model by using the test set, and if the difference between the test result of the test set and the training result of the training set exceeds a threshold value, retraining.
The convolutional neural network model consists of a convolutional layer 1, an active layer 1, a pooling layer 1, a convolutional layer 2, an active layer 2, a convolutional layer 3, an active layer 3, a pooling layer 2, a depth serial layer 1, a convolutional layer 4, an active layer 4, a pooling layer 3, a depth serial layer 2, a convolutional layer 5, an active layer 5, a pooling layer 4, a depth serial layer 3, a convolutional layer 6, an active layer 6, a pooling layer 5, a depth serial layer 4, a convolutional layer 7, an active layer 7, a pooling layer 6, a depth serial layer 5, a full connection layer and a classified output layer.
Wherein each layer of convolution layer: extracting features of the concentrated images to obtain feature images;
each layer of activation layer: nonlinear mapping is carried out on the results of the corresponding convolution layers;
each layer of pooling layer: selecting the extracted features of the corresponding convolution layers;
each depth tandem layer: the results after the corresponding pooling layers are processed are connected in series;
full tie layer: all feature matrixes of the pooling layer are converted into one-dimensional feature big vectors;
classification output layer: and outputting the result.
If the test result of the test set is more than 1% different from the training result of the training set in threshold value, the learning rate is changed, and the training is performed again by sampling number and training round number.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some or all of the technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit of the invention.
Claims (5)
1. The bearing friction state identification method based on the attractor and the convolutional neural network is characterized by comprising the following specific steps of:
s1: collecting acceleration vibration signals of a bearing seat of the sliding bearing test bed in a liquid friction state, a mixed friction state and a dry friction state respectively;
s2: noise reduction is carried out on the acquired acceleration vibration signals, and then a three-dimensional attractor graph is generated through a phase space reconstruction method;
s3: projecting the three-dimensional attractor graph into a two-dimensional attractor graph, and normalizing the two-dimensional attractor graph to form an original image set;
s4: building a convolutional neural network model and training based on an original image set;
s5: and carrying out bearing friction state identification on the vibration signal to be detected based on the trained convolutional neural network model so as to determine the corresponding friction state.
2. The method for recognizing a friction state of a bearing based on an attractor and a convolutional neural network according to claim 1, wherein in S2,
s21: decomposing the acquired acceleration vibration signals by a harmonic wavelet packet method and carrying out noise reduction treatment;
s22: determining parameters of phase space reconstruction: embedding dimension m and delay time tau, defining acceleration vibrationTime series x= { x of dynamic signals i I=1, 2, …, N }, constructing a batch of phase space matrices X, x= [ X ] based on the embedding dimension m and the delay time τ i ,X i+τ ,X i+2τ ,…,X i+(m-1)τ ](i=1, 2, …, M), N being the time series length; x is X i Is the ith phase space matrix; m=n- (M-1) τ is the number of phase space matrices, and M-dimensional vectors are all points X of reconstructed phase space 1 ,X 2 ,…,X M The method comprises the steps of carrying out a first treatment on the surface of the And reconstructing the acceleration vibration signal attractor after noise reduction treatment according to the selected embedding dimension m and the delay time tau to obtain a three-dimensional attractor.
3. The method for recognizing the friction state of the bearing based on the attractor and the convolutional neural network according to claim 1, wherein in the step S3, root mean square of every two coordinate axes in the three-dimensional attractor graph is calculated, two coordinate axes with the largest root mean square are taken as two-dimensional attractor graph coordinate axes, and the two-dimensional attractor graph is generated by projection.
4. The method for recognizing a friction state of a bearing based on an attractor and a convolutional neural network according to claim 1, wherein in S4,
s41: classifying the original image set according to the liquid friction state, the mixed friction state and the dry friction state, setting corresponding labels, constructing and forming a training set, and randomly extracting a part of images from the original image set to construct a test set;
s42: building a convolutional neural network model, and training the convolutional neural network model by using a training set;
s43: and (3) evaluating the accuracy of the test set by using the trained convolutional neural network model, and if the difference between the test result of the test set and the training result of the training set exceeds a threshold value, retraining.
5. The method for identifying the friction state of the bearing based on the attractor and the convolutional neural network according to claim 2, wherein in the S22, the embedding dimension m and the delay time tau are selected by a C-C method, the correlation integral C (m, N, r, t) is defined, the statistic S (m, N, r, t) is reconstructed, r represents the distance between the points in the phase space, and t is the local maximum time;
definition τ s For time series sampling interval, τ is delay time, then time series is delayed by τ d =ττ s Delay time window τ w =(m-1)τ d Calculating time series delay tau from statistics d Further determining the delay time τ, and then determining the delay time τ w 、τ d The embedding dimension m is determined.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310771744.2A CN116735206A (en) | 2023-06-27 | 2023-06-27 | Bearing friction state identification method based on attractor and convolutional neural network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310771744.2A CN116735206A (en) | 2023-06-27 | 2023-06-27 | Bearing friction state identification method based on attractor and convolutional neural network |
Publications (1)
Publication Number | Publication Date |
---|---|
CN116735206A true CN116735206A (en) | 2023-09-12 |
Family
ID=87900974
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310771744.2A Pending CN116735206A (en) | 2023-06-27 | 2023-06-27 | Bearing friction state identification method based on attractor and convolutional neural network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116735206A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117574114A (en) * | 2024-01-15 | 2024-02-20 | 安徽农业大学 | Remote reconstruction and jump disturbance detection method for running data of rotary machine |
-
2023
- 2023-06-27 CN CN202310771744.2A patent/CN116735206A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117574114A (en) * | 2024-01-15 | 2024-02-20 | 安徽农业大学 | Remote reconstruction and jump disturbance detection method for running data of rotary machine |
CN117574114B (en) * | 2024-01-15 | 2024-04-19 | 安徽农业大学 | Remote reconstruction and jump disturbance detection method for running data of rotary machine |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107941537B (en) | A kind of mechanical equipment health state evaluation method | |
Zhao et al. | Improved generative adversarial network for vibration-based fault diagnosis with imbalanced data | |
CN109299705B (en) | Rotary machine fault diagnosis method based on one-dimensional depth residual convolution neural network | |
Tao et al. | Bearing fault diagnosis method based on stacked autoencoder and softmax regression | |
CN111582392B (en) | Multi-working-condition health state online monitoring method for key components of wind turbine generator | |
CN111931625B (en) | Product key part residual life prediction method based on asymmetric loss neural network | |
CN111458142A (en) | Sliding bearing fault diagnosis method based on generation of countermeasure network and convolutional neural network | |
CN116735206A (en) | Bearing friction state identification method based on attractor and convolutional neural network | |
CN112633098A (en) | Fault diagnosis method and system for rotary machine and storage medium | |
CN114548190B (en) | Wind turbine fault diagnosis method and device based on self-adaptive residual error neural network | |
CN111709190A (en) | Wind turbine generator operation data image identification method and device | |
CN107180140A (en) | Shafting fault recognition method based on dual-tree complex wavelet and AdaBoost | |
CN115290326A (en) | Rolling bearing fault intelligent diagnosis method | |
Wang et al. | Construction of the efficient attention prototypical net based on the time–frequency characterization of vibration signals under noisy small sample | |
CN116416497A (en) | Bearing fault diagnosis system and method | |
CN112418175A (en) | Rolling bearing fault diagnosis method and system based on domain migration and storage medium | |
CN111428772B (en) | Photovoltaic system depth anomaly detection method based on k-nearest neighbor adaptive voting | |
Sun et al. | Curvature enhanced bearing fault diagnosis method using 2D vibration signal | |
CN114459760B (en) | Rolling bearing fault diagnosis method and system in strong noise environment | |
CN114595730A (en) | Rolling bearing fault diagnosis method based on GAF-DRSN | |
Yeh et al. | Using convolutional neural network for vibration fault diagnosis monitoring in machinery | |
Shaowu et al. | Study on the health condition monitoring method of hydraulic pump based on convolutional neural network | |
Chung et al. | A binary weight convolutional neural network hardware accelerator for analysis faults of the CNC machinery on FPGA | |
Lu et al. | An automated bearing fault diagnosis using a self-normalizing convolutional neural network | |
CN116522237A (en) | Rolling bearing fault diagnosis method based on parallel convolution neural network |
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 |