CN114548186A - Fault diagnosis method for rotor of gearbox of air compressor high-speed machine - Google Patents

Fault diagnosis method for rotor of gearbox of air compressor high-speed machine Download PDF

Info

Publication number
CN114548186A
CN114548186A CN202210200427.0A CN202210200427A CN114548186A CN 114548186 A CN114548186 A CN 114548186A CN 202210200427 A CN202210200427 A CN 202210200427A CN 114548186 A CN114548186 A CN 114548186A
Authority
CN
China
Prior art keywords
rotor
air compressor
signal
fault
data
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202210200427.0A
Other languages
Chinese (zh)
Inventor
吴孝炯
罗海华
张小根
董益华
方昌勇
王天兴
蒋月红
张曦
王俊伟
赵申轶
叶飞宇
徐明阳
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang Energy Group Research Institute Co Ltd
Zhejiang Zheneng Lanxi Power Generation Co Ltd
Original Assignee
Zhejiang Energy Group Research Institute Co Ltd
Zhejiang Zheneng Lanxi Power Generation Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Zhejiang Energy Group Research Institute Co Ltd, Zhejiang Zheneng Lanxi Power Generation Co Ltd filed Critical Zhejiang Energy Group Research Institute Co Ltd
Priority to CN202210200427.0A priority Critical patent/CN114548186A/en
Publication of CN114548186A publication Critical patent/CN114548186A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/12—Classification; Matching
    • G—PHYSICS
    • 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
    • G01M13/02—Gearings; Transmission mechanisms
    • G01M13/021—Gearings
    • 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
    • G01M13/02—Gearings; Transmission mechanisms
    • G01M13/028—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
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • 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
    • 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)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Artificial Intelligence (AREA)
  • Signal Processing (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Acoustics & Sound (AREA)
  • Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)

Abstract

The invention relates to a fault diagnosis method for a rotor of a high-speed gear box of an air compressor, which comprises the following steps: collecting a vibration signal of a rotor of a high-speed gear box of a centrifugal air compressor; carrying out VMD signal decomposition on the rotor vibration signal; acquiring feature vector data of a rotor vibration signal; establishing a KELM model, importing the characteristic vector data of the rotor vibration signal into the KELM model for training and prediction, and establishing a fault diagnosis model based on the rotor vibration signal of the high-speed gearbox; and applying the trained KELM model to a centrifugal air compressor operation system, acquiring operation data in real time, and comparing the change trend of the real-time data with the change trend of model prediction data. The invention has the beneficial effects that: according to the method, the vibration signal of the high-speed gearbox rotor of the centrifugal air compressor is collected, after VMD decomposition, eigenvector calculation and KPCA dimension reduction are carried out on the vibration signal, a KELM model can be established for training to obtain a fault diagnosis model, and machine set shutdown caused by sudden faults is avoided.

Description

Fault diagnosis method for rotor of gearbox of air compressor high-speed machine
Technical Field
The invention relates to the technical field of fault diagnosis, in particular to a fault diagnosis method for a rotor of a gearbox of a high-speed machine of an air compressor.
Background
The centrifugal air compressor has high rotating speed, so that the high-speed gear box is widely applied to various centrifugal air compressors, a main shaft rotor of the high-speed gear box is a key mechanical part, the maintenance cost is high, and the high-speed gear box can be failed due to untimely treatment when the high-speed gear box fails in a non-maintenance period. During the operation of the centrifugal air compressor, the gear box is in a sealing state during the operation, so that the operation state inside the gear cannot be monitored in real time, regular maintenance is required, and labor and time are consumed. The existing neural network is easy to fall into the problem of local minimum, and the prediction effect is poor.
Disclosure of Invention
The invention aims to overcome the defects in the prior art and provides a fault diagnosis method for a rotor of a gearbox of a high-speed machine of an air compressor, which adopts the following technical scheme:
step1, collecting vibration signals of a rotor of a high-speed gear box of a centrifugal air compressor, wherein the vibration signals of the rotor comprise fault signals and normal signals.
And 2, performing VMD signal decomposition on the rotor vibration signal.
And 3, acquiring the characteristic vector data of the fault signal and the characteristic vector data of the normal signal.
And 4, establishing a Kernel Extreme Learning Machine (KELM) model, importing the characteristic vector data of the fault signal and the characteristic vector data of the normal signal into the KELM model for training and predicting, and establishing a fault diagnosis model based on the rotor vibration signal of the high-speed gearbox.
And 5, applying the trained KELM model to a centrifugal air compressor operation system, collecting operation data in real time, and comparing the change trend of the real-time data with the change trend of model prediction data.
Preferably, in step2, the VMD signal is decomposed to obtain a plurality of relatively stable subsequences with different frequencies of the rotor vibration signal, so as to obtain the natural mode functions.
Preferably, in step3, the feature vector data of the fault signal includes: fault signal sample entropy and fault signal characteristic energy; the feature vector data of the normal signal includes: normal signal sample entropy and normal signal characteristic energy.
Preferably, in step3, the feature vector data of the fault signal includes: a Kernel Principal Component Analysis (KPCA) dimension reduction result of the entropy of the fault signal sample and a KPCA dimension reduction result of the characteristic energy of the fault signal; the feature vector data of the normal signal includes: and the KPCA dimension reduction result of the normal signal sample entropy and the KPCA dimension reduction result of the normal signal characteristic energy.
Preferably, the KPCA dimensionality reduction specifically comprises the following steps:
step1, data standardization processing;
step2, acquiring a kernel matrix K, and mapping the original data from the data space to the feature space by using a Gaussian radial basis kernel function, wherein the formula is as follows:
Figure BDA0003527207980000021
where xi, xj represent vectors, exp is an exponential function with a natural constant e as the base, K (x)i,xj) Representing a Gaussian radial basis kernel function, wherein gamma is a kernel function parameter;
step3, acquiring a centralized kernel matrix Kc for correcting the kernel matrix K, wherein the formula is as follows:
KC＝K-lNK-KlN+lNKlN
wherein lNThe matrix is N multiplied by N, and each element is 1/N;
step4, calculating the eigenvalue of the centering kernel matrix Kc, wherein the eigenvector corresponding to the eigenvalue of Kc is lambda1,…,λnSorting the eigenvalues in a descending order, and correspondingly adjusting the eigenvectors;
step5, orthogonalizing and unitizing the feature vectors in Step4 by the Schmidt orthogonalization method to obtain a1,…,an；
Step6, calculating the accumulated contribution rate r of the characteristic value1,…,rnGiven a contribution ratio requirement p, if rm>And p, selecting the first m principal components of the characteristic value as the data after dimension reduction.
Preferably, in step4, when the KELM model is established, a genetic algorithm is adopted to optimize the kernel parameters and the regularization coefficients.
Preferably, in step5, when the difference between the real-time data and the model prediction data is greater than a difference threshold value, the centrifugal air compressor operation system gives an alarm.
The invention has the beneficial effects that: according to the method, vibration signals of the high-speed gearbox rotor of the centrifugal air compressor are collected, after VMD decomposition, eigenvector calculation and KPCA dimension reduction are carried out on the vibration signals, a KELM model can be established for training to obtain a fault diagnosis model, early warning monitoring is carried out based on the fault diagnosis model, maintenance personnel can be guaranteed to maintain equipment in time, and machine set shutdown caused by sudden faults is avoided.
Drawings
FIG. 1 is a flow chart of a fault diagnosis method for a rotor of a gearbox of an air compressor high-speed machine provided by the application;
FIG. 2 is a diagram illustrating the decomposition result of the VMD in the embodiment.
Detailed Description
The present invention will be further described with reference to the following examples. The following examples are set forth merely to aid in the understanding of the invention. It should be noted that, for a person skilled in the art, several modifications can be made to the invention without departing from the principle of the invention, and these modifications and modifications also fall within the protection scope of the claims of the present invention.
Example 1:
a fault diagnosis method for a gearbox rotor of an air compressor high-speed machine is shown in figure 1 and comprises the following steps:
step1, collecting vibration signals of a rotor of a high-speed gear box of a centrifugal air compressor, wherein the vibration signals of the rotor comprise fault signals and normal signals.
In step1, the vibration sensing device can acquire the vibration signal of the rotor of the high-speed gearbox of the air compressor. In addition, the rotor vibration signal in step1 may also be referred to as a raw signal.
And 2, carrying out VMD signal decomposition on the rotor vibration signal.
As shown in fig. 2, the VMD decomposition method is used to decompose the original signal acquired in step1, so as to obtain a plurality of relatively stable subsequences with different frequencies and different scales of the rotor vibration signal, thereby obtaining each natural mode function. In fig. 2, the ordinate represents a modal component, and the abscissa represents a sampling time.
And 3, acquiring the characteristic vector data of the fault signal and the characteristic vector data of the normal signal.
In an alternative implementation, the feature vector data of the fault signal includes: fault signal sample entropy and fault signal characteristic energy; the feature vector data of the normal signal includes: normal signal sample entropy and normal signal characteristic energy.
It should be noted that the decomposed signal has characteristic energy, and the change rule thereof can reflect the fault state of the rotor to some extent, and the parameter calculation method is as follows:
Figure BDA0003527207980000031
where ci (t) represents the magnitude of each scale coefficient or natural mode function at time t.
In addition, the sample entropy is specifically calculated as follows:
Figure BDA0003527207980000032
wherein N is the original data length, m is the embedding dimension, r is the similar tolerance, and the average value is marked as Bm(r), increasing the dimension by 1, then marking as Bm+1(r)。
For example, sample entropy calculations are performed on 40 groups and 4 types of high-speed gearbox faults, and the obtained sample entropy distribution corresponding to each intrinsic mode function is shown in the following table.
Figure BDA0003527207980000041
This application carries out signal acquisition when every trouble operating mode takes place, numbers various fault types, has had corresponding various trouble operating mode signals before sample entropy analysis, carries out VMD to the fault signal again and decomposes in order to obtain the sample entropy.
In another alternative implementation, the feature vector data of the fault signal includes: a KPCA dimension reduction result of the fault signal sample entropy and a KPCA dimension reduction result of the fault signal characteristic energy; the feature vector data of the normal signal includes: and the KPCA dimension reduction result of the normal signal sample entropy and the KPCA dimension reduction result of the normal signal characteristic energy.
The KPCA dimension reduction can map most quantity features of the obtained samples to a high-dimensional space, thereby realizing the extraction of nonlinear features. And the KPCA algorithm is introduced to realize the dimension reduction of data, so that the calculation time can be effectively reduced, and the efficiency and the model prediction precision are improved.
The KPCA dimension reduction specifically comprises the following steps:
step1, data standardization processing;
step2, acquiring a kernel matrix K, and mapping the original data from the data space to the feature space by using a Gaussian radial basis kernel function, wherein the formula is as follows:
Figure BDA0003527207980000051
where xi, xj represent vectors, exp is an exponential function with a natural constant e as the base, K (x)i,xj) Representing a gaussian radial basis kernel function, and gamma is a kernel function parameter.
Step3, acquiring a centralized kernel matrix Kc for correcting the kernel matrix K, wherein the formula is as follows:
KC＝K-lNK-KlN+lNKlN
wherein lNIs an N multiplied by N matrix, each element is 1/N;
step4, calculating the eigenvalue of the centering kernel matrix Kc, wherein the eigenvector corresponding to the eigenvalue of Kc is lambda1,…,λnThe eigenvalue determines the magnitude v of the variance1,…,νnThat is, the larger the eigenvalue is, the more useful information is contained, so that the eigenvalues are sorted in a descending order, and the eigenvector is adjusted correspondingly;
step5, orthogonalizing and unitizing the feature vectors in Step4 by the Schmidt orthogonalization method to obtain a1,…,an；
Step6, calculating the accumulated contribution rate r of the characteristic value1,…,rnGiven a contribution ratio requirement p, if rm>And p, selecting the first m principal components of the characteristic value as the data after dimension reduction.
And 4, establishing a KELM model, importing the characteristic vector data of the fault signal and the characteristic vector data of the normal signal into the KELM model for training and predicting, and establishing a fault diagnosis model based on the vibration signal of the rotor of the high-speed gearbox.
Optionally, in building the KELM model, a genetic algorithm is used to optimize the kernel parameters and the regularization coefficients.
And 5, applying the trained KELM model to a centrifugal air compressor operation system, collecting operation data in real time, and comparing the change trend of the real-time data with the change trend of model prediction data.
The KELM is adopted for state recognition of various fault signals, and the diagnosis time and the diagnosis precision spent on various fault signals are shown in the following table. The diagnosis result of the diagnosis model for each fault type shows that the model prediction precision is higher.
Type of failure Time required for diagnosis(s) Diagnostic accuracy (%)
Broken tooth 11.36 100
Wear of tooth surface 12.56 97.6
Scratch mark of tooth surface 12.78 95.3
Fatigue of tooth surface 11.67 98.7
If the vibration signal of the high-speed gearbox is continuously in an abnormal state, namely when a prediction vibration signal and an actual vibration signal of the model have large errors, the model judges that the inside of the rotor of the high-speed gearbox breaks down, early warning can be sent out in advance, maintenance personnel can be guaranteed to timely maintain the equipment, and the accident shutdown of the air compressor unit caused by sudden faults is avoided.
Optionally, in step5, when the difference between the real-time data and the model prediction data is larger than the difference threshold value, the centrifugal air compressor running system gives an alarm.

Claims (7)

1. A fault diagnosis method for a rotor of a high-speed gearbox of an air compressor is characterized by comprising the following steps:
step1, collecting vibration signals of a rotor of a high-speed gearbox of a centrifugal air compressor, wherein the vibration signals of the rotor comprise fault signals and normal signals;
step2, performing VMD signal decomposition on the rotor vibration signal;
step3, acquiring feature vector data of the fault signal and feature vector data of the normal signal;
step4, establishing a KELM model, importing the characteristic vector data of the fault signal and the characteristic vector data of the normal signal into the KELM model for training and predicting, and establishing a fault diagnosis model based on a rotor vibration signal of a high-speed gearbox;
and 5, applying the trained KELM model to a centrifugal air compressor operation system, collecting operation data in real time, and comparing the change trend of the real-time data with the change trend of model prediction data.
2. The method for diagnosing the faults of the rotor of the high-speed gearbox of the air compressor as claimed in claim 1, wherein in the step2, the VMD signal is decomposed to obtain a plurality of relatively stable subsequences with different frequencies and different scales of the vibration signal of the rotor, so that each natural mode function is obtained.
3. The method for diagnosing the fault of the rotor of the high-speed gearbox of the air compressor as claimed in claim 1, wherein in the step3, the characteristic vector data of the fault signal comprises: fault signal sample entropy and fault signal characteristic energy; the feature vector data of the normal signal includes: normal signal sample entropy and normal signal characteristic energy.
4. The method for diagnosing the fault of the rotor of the high-speed gearbox of the air compressor as claimed in claim 1, wherein in the step3, the characteristic vector data of the fault signal comprises: KPCA dimension reduction results of the entropy of the fault signal sample and KPCA dimension reduction results of the characteristic energy of the fault signal; the feature vector data of the normal signal includes: and the KPCA dimension reduction result of the normal signal sample entropy and the KPCA dimension reduction result of the normal signal characteristic energy.
5. The air compressor high-speed gearbox rotor fault diagnosis method as claimed in claim 4, wherein the KPCA dimension reduction specifically comprises the following steps:
step1, data standardization processing;
step2, acquiring a kernel matrix K, and mapping the original data from the data space to the feature space by using a Gaussian radial basis kernel function, wherein the formula is as follows:
Figure FDA0003527207970000011
where xi, xj represent vectors, exp is an exponential function with a natural constant e as the base, K (x)i,xj) Representing a gaussian radial basis kernel function, and gamma is a kernel function parameter.
Step3, acquiring a centralized kernel matrix Kc for correcting the kernel matrix K, wherein the formula is as follows:
KC＝K-lNK-KlN+lNKlN
wherein lNThe matrix is N multiplied by N, and each element is 1/N;
step4, calculating the eigenvalue of the centering kernel matrix Kc, wherein the eigenvector corresponding to the eigenvalue of Kc is lambda1,…,λnSorting the eigenvalues in a descending order, and correspondingly adjusting the eigenvectors;
step5, orthogonalizing and unitizing the feature vectors in Step4 by the Schmidt orthogonalization method to obtain a1,…,an；
Step6, calculating the accumulated contribution rate r of the characteristic value1,…,rnGiven a contribution ratio requirement p, if rm>And p, selecting the first m principal components of the characteristic value as the data after dimension reduction.
6. The air compressor high-speed gearbox rotor fault diagnosis method as claimed in claim 1, wherein in step4, a genetic algorithm is adopted to optimize nuclear parameters and regularization coefficients when the KELM model is established.
7. The method for diagnosing the rotor fault of the high-speed gearbox of the air compressor as claimed in claim 1, wherein in step5, the centrifugal air compressor operation system gives an alarm when the difference between the real-time data and the model prediction data is greater than a difference threshold value.
CN202210200427.0A 2022-03-02 2022-03-02 Fault diagnosis method for rotor of gearbox of air compressor high-speed machine Pending CN114548186A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210200427.0A CN114548186A (en) 2022-03-02 2022-03-02 Fault diagnosis method for rotor of gearbox of air compressor high-speed machine

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210200427.0A CN114548186A (en) 2022-03-02 2022-03-02 Fault diagnosis method for rotor of gearbox of air compressor high-speed machine

Publications (1)

Publication Number Publication Date
CN114548186A true CN114548186A (en) 2022-05-27

Family

ID=81661769

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210200427.0A Pending CN114548186A (en) 2022-03-02 2022-03-02 Fault diagnosis method for rotor of gearbox of air compressor high-speed machine

Country Status (1)

Country Link
CN (1) CN114548186A (en)

Similar Documents

Publication Publication Date Title
CN110276416B (en) Rolling bearing fault prediction method
CN111353482B (en) LSTM-based fatigue factor recessive anomaly detection and fault diagnosis method
CN107247230B (en) Rotary motor state monitoring method based on support vector machine and data driving
CN113834657B (en) Bearing fault early warning and diagnosis method based on improved MSET and frequency spectrum characteristics
CN109858104B (en) Rolling bearing health assessment and fault diagnosis method and monitoring system
CN113469060A (en) Multi-sensor fusion convolution neural network aeroengine bearing fault diagnosis method
CN110926809A (en) Big data analysis-based wind turbine generator transmission chain fault early warning method
CN113657454B (en) Nuclear power rotating machinery state monitoring method based on autoregressive BiGRU
CN109523171B (en) SVDD-based gas turbine air inlet system health degree evaluation method
CN114861827B (en) Coal mining machine predictive diagnosis and health management method based on multi-source data fusion
CN110046409B (en) ResNet-based steam turbine component health state evaluation method
CN111103137A (en) Wind turbine gearbox fault diagnosis method based on deep neural network
CN111723850A (en) Automatic verification equipment state evaluation method based on intelligent inspection system
CN112207631B (en) Method for generating tool detection model, method, system, device and medium for detecting tool detection model
CN113579851A (en) Non-stationary drilling process monitoring method based on adaptive segmented PCA
CN114326486B (en) Process monitoring method based on probability slow feature analysis and elastic weight consolidation
CN110007660B (en) Online soft measurement method for transient equivalent thermal stress of steam turbine set of thermal power plant
KR20210006832A (en) Method and apparatus for machine fault diagnosis
CN113313365A (en) Degradation early warning method and device for primary air fan
CN115165326A (en) Fan fault diagnosis method through mechanical transmission chain lubricating oil (grease) impurity analysis
CN114548186A (en) Fault diagnosis method for rotor of gearbox of air compressor high-speed machine
CN113435228A (en) Motor bearing service life prediction and analysis method based on vibration signal modeling
US11339763B2 (en) Method for windmill farm monitoring
CN113468818B (en) Winding running state monitoring method and device for wind generating set
CN115326393A (en) Wind turbine generator bearing pair fault diagnosis method based on temperature information

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