CN111241632A - Suspension system fault pre-diagnosis method based on random matrix theory and Euclidean distance - Google Patents

Suspension system fault pre-diagnosis method based on random matrix theory and Euclidean distance Download PDF

Info

Publication number
CN111241632A
CN111241632A CN202010067591.XA CN202010067591A CN111241632A CN 111241632 A CN111241632 A CN 111241632A CN 202010067591 A CN202010067591 A CN 202010067591A CN 111241632 A CN111241632 A CN 111241632A
Authority
CN
China
Prior art keywords
matrix
suspension system
fault
msr
theory
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.)
Granted
Application number
CN202010067591.XA
Other languages
Chinese (zh)
Other versions
CN111241632B (en
Inventor
龙志强
王平
高明
窦峰山
胡海林
钟成
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.)
National University of Defense Technology
CRRC Tangshan Co Ltd
Original Assignee
National University of Defense Technology
CRRC Tangshan 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 National University of Defense Technology, CRRC Tangshan Co Ltd filed Critical National University of Defense Technology
Priority to CN202010067591.XA priority Critical patent/CN111241632B/en
Publication of CN111241632A publication Critical patent/CN111241632A/en
Application granted granted Critical
Publication of CN111241632B publication Critical patent/CN111241632B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • 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

  • Vehicle Body Suspensions (AREA)

Abstract

The invention discloses a suspension system fault pre-diagnosis method based on a random matrix theory and Euclidean distance, which comprises the following steps: s1, constructing an initial matrix based on the random matrix theory and big data; s2, converting the initial matrix into a batch random matrix through a preset mobile window; s3, converting the batch random matrix into a non-Hermite matrix according to a random matrix theory single-loop theorem; s4, calculating a characteristic value of the non-Hermite matrix, obtaining the MSR of the non-Hermite matrix according to the characteristic value, and taking the MSR as a health state value of the suspension system; s5, when the health state value MSR of the suspension system reaches an early warning value, calculating the contribution degree of each parameter; s6, calculating the distance between the contribution degree of each parameter in the step S5 and the contribution degree of each parameter in the existing various fault data, and taking the fault with the minimum distance with the current state as the fault to be generated by the suspension system, namely the predictive fault diagnosis result. The invention realizes predictive fault diagnosis by using random matrix theory and Euclidean distance, and has simple detection and high reliability.

Description

Suspension system fault pre-diagnosis method based on random matrix theory and Euclidean distance
Technical Field
The invention relates to the technical field of magnetic suspension trains, in particular to a suspension system fault pre-diagnosis method based on a random matrix theory and Euclidean distance.
Background
With the popularization of magnetic levitation trains, the safety and reliability of levitation systems are receiving more and more attention. In the running process of the magnetic suspension train, once the suspension system breaks down, the train cannot run. This is largely avoided if predictive diagnostics can be performed on the system before a failure of the levitation system occurs. Therefore, how to accurately implement the predictive fault diagnosis of the suspension system is a problem which needs to be solved urgently at present.
Disclosure of Invention
In view of the above, the invention provides a suspension system fault pre-diagnosis method based on a random matrix theory and an Euclidean distance, and the method realizes predictive fault diagnosis by using the random matrix theory and the Euclidean distance, and has the advantages of simple detection and high reliability.
On one hand, the invention provides a suspension system fault pre-diagnosis method based on a random matrix theory and Euclidean distance, which comprises the following steps:
s1, constructing an initial matrix based on the random matrix theory and big data;
s2, converting the initial matrix into a batch random matrix through a preset mobile window;
s3, converting the batch random matrix into a non-Hermite matrix according to a random matrix theory single-loop theorem;
s4, calculating a characteristic value of the non-Hermite matrix, obtaining an MSR of the characteristic value according to the characteristic value, and taking the MSR as a health state value of the suspension system;
s5, when the health state value MSR of the suspension system reaches an early warning value, calculating the contribution degree of each parameter;
s6, calculating the distance between the contribution degree of each parameter in the step S5 and the contribution degree of each parameter in the existing various fault data, and taking the fault with the minimum distance with the current state as the fault to be generated by the suspension system, namely the predictive fault diagnosis result.
As a further improvement, the initial matrix is described by the expression:
Figure BDA0002376425770000021
wherein A represents an initial matrix, T represents a matrix transposition, R represents the number of variables, NmRepresenting the number of samples in the mth time series,
Figure BDA0002376425770000022
represents R line NmSet of column matrices, X1、X2、Xk、XRThe elements of the initial matrix a are respectively expressed as:
Figure BDA0002376425770000023
in the formula, XkOne-dimensional random matrix, x, representing the k-th variable in the m-th time seriesk1、xk2、
Figure BDA0002376425770000024
Are respectively a one-dimensional random matrix XkOf (2) is used.
As a further improvement, the expression of the batch random matrix is:
Aj＝(Yj1,Yj2,...,Yjk,...,YjR)T∈CR×d(3)
in the formula, Yj1、Yj2、Yjk、YjRRespectively, a batch random matrix AjVector of length d, d representing the width of the moving window, CR×dRepresenting a matrix set of R rows and d columns, j ∈ [1, Nm-d+1]Is a batch random matrix AjWherein:
Yjk＝(yk1,yk2,...,ykd)∈C1×d(4)
in the formula, yk1、yk2、ykdIs YjkElement (2) of (C)1×dRepresenting a matrix set of l rows and d columns.
As a further improvement, the width size of the moving window is obtained by the following formula:
d＝R*c (6)
wherein c represents the ratio of determinant and is obtained by the following steps:
1) setting the minimum value of c as:
Figure BDA0002376425770000025
wherein epsilon is a proportional factor for preventing d from being too large;
2) taking the selected time sequence as an object, and taking 0.1 as a step length from the minimum value of c to 1 by c to obtain different MSR curves;
3) comparing the obtained different MSR curves, and taking the optimal c to make the curve monotonically decrease;
4) other time series are used to determine if c is appropriate.
As a further improvement, the non-Hermite matrix in step S3 is obtained specifically by: after c is obtained, the random matrix A is applied to the batchjIs normalized to obtain a standard non-Hermite matrix
Figure BDA0002376425770000031
The expression of (a) is:
Figure BDA0002376425770000032
in the formula, Bj1、Bj2、Bjk、BjRRespectively, non-Hermite matrix
Figure BDA0002376425770000033
Vector of medium length d.
As a further improvement, the non-Hermite matrix
Figure BDA0002376425770000034
The characteristic value of (a) is obtained by the following steps:
is provided with
Figure BDA0002376425770000035
Has a characteristic value of
Figure BDA0002376425770000036
Wherein i 1.., n, each in increasing order;
according to the following formula, a non-Hermite matrix is obtained through calculation
Figure BDA0002376425770000037
Characteristic value of
Figure BDA0002376425770000038
Figure BDA0002376425770000039
In the formula (I), the compound is shown in the specification,
Figure BDA00023764257700000310
is composed of
Figure BDA00023764257700000311
The matrix of equivalent singular values of (a),
Figure BDA00023764257700000312
is composed of
Figure BDA00023764257700000313
The transpose matrix of (a) is,
Figure BDA00023764257700000314
is composed of
Figure BDA00023764257700000315
The transposed matrix of (1), wherein,
Figure BDA00023764257700000316
the following formula is used to solve:
Figure BDA00023764257700000317
in the formula, U is a Harley matrix.
As a further improvement, a non-Hermite matrix is defined
Figure BDA00023764257700000318
The MSR of the characteristic value of (1) is MSRj,qFor reflecting the matrix
Figure BDA00023764257700000319
The distribution of characteristic values of (a), which is obtained by the following formula:
Figure BDA00023764257700000320
in the formula, MSRj,qRepresents the state of health value of the suspension system at the q-th time,
Figure BDA00023764257700000321
is a characteristic value λ
Figure BDA00023764257700000322
Of (c) is used.
As a further improvement, MSRj,qThe method is used for early warning and judgment of the suspension system, and obtains the fault early warning occurrence time of the suspension system through the following processes:
setting a threshold value theta when MSRj,q>θ, and MSRj,q-1<And when theta is reached, the q-1 moment is the moment of the fault early warning of the suspension system.
As a further improvement, a group of historical data matrixes obtained according to the fault early warning time are defined as H, wherein H belongs to EZ×M，EZ×MThe characteristic value delta of H is diag (delta) representing a matrix set of Z rows and M columns1,...,δZ) Then, the expression of the contribution of each parameter in H is specifically as follows:
Figure BDA0002376425770000041
in the formula, #rIs the contribution, δ, of the r-th eigenvalue in the current staterIs the r-th eigenvalue of H, and Z is the number of H eigenvalues.
As a further improvement, #rThe method is used for suspension system predictive fault diagnosis, and obtains a suspension system predictive fault diagnosis result through the following formula:
D＝minDt(13)
wherein D represents the predictive failure diagnosis result of the suspension system, DtAnd representing the distance between the contribution degree of each parameter and the contribution degree of each parameter in the existing various fault data, wherein:
Figure BDA0002376425770000042
in the formula, #r,f,hContribution degree of the r characteristic value of the f group in the h fault, VhThe number of sets of eigenvalues in the h-th fault.
The invention provides a suspension system fault pre-diagnosis method based on a random matrix theory and Euclidean distance, which mainly realizes early fault detection through two steps: firstly, obtaining a health state value of a suspension system through a random matrix theory; secondly, calculating the contribution degree of each parameter through principal component analysis; and thirdly, selecting the minimum distance as a result by utilizing the Euclidean distance to realize a fault pre-diagnosis result, and the method has the advantages of simple diagnosis and high reliability.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate an embodiment of the invention and, together with the description, serve to explain the invention and not to limit the invention. In the drawings:
fig. 1 is a flowchart of a suspension system fault pre-diagnosis method based on a random matrix theory and euclidean distance according to an embodiment of the present invention.
Detailed Description
It should be noted that the embodiments and features of the embodiments may be combined with each other without conflict. The present invention will be described in detail below with reference to the embodiments with reference to the attached drawings.
For a better understanding of the invention, the following explanations are made in advance:
fig. 1 is a flowchart of a suspension system fault pre-diagnosis method based on a random matrix theory and euclidean distance according to an embodiment of the present invention. As shown in fig. 1, the method for predicting the suspension system fault based on the random matrix theory and the euclidean distance includes the following steps:
s1, constructing an initial matrix based on the random matrix theory and big data;
s2, converting the initial matrix into a batch random matrix through a preset mobile window;
s3, converting the batch random matrix into a non-Hermite matrix (non-Hermite matrix) according to a random matrix theory single-loop theorem;
s4, calculating a characteristic value of the non-Hermite matrix, obtaining an MSR (mean spectral radius) of the characteristic value according to the characteristic value, and taking the MSR as a health state value of the suspension system;
s5, when the health state value MSR of the suspension system reaches an early warning value, calculating the contribution degree of each parameter;
s6, calculating the distance between the contribution degree of each parameter in the step S5 and the contribution degree of each parameter in the existing various fault data, and taking the fault with the minimum distance with the current state as the fault to be generated by the suspension system, namely the predictive fault diagnosis result.
From the above discussion, it is clear that the present invention achieves early fault detection primarily in two steps: firstly, analyzing a characteristic value and an energy spectrum of a complex system through a mathematical statistical method by using a Random Matrix Theory (RMT) to obtain an internal attribute of the complex system, wherein the internal attribute can be used for analyzing a relevant state of the system to reflect an operation state of the system) to obtain a health state value of a suspension system; secondly, calculating the contribution degree of each parameter through principal component analysis; and thirdly, selecting the minimum distance as a result by utilizing the Euclidean distance to realize a fault pre-diagnosis result, and the method has the advantages of simple diagnosis and high reliability.
Specifically, the steps are realized through the following processes:
assuming that the number of variables is R, the expression of the one-dimensional random matrix of the kth variable in the mth time series is as follows:
Figure BDA0002376425770000061
in the formula, XkOne-dimensional random matrix, x, representing the k-th variable in the m-th time seriesk1、xk2、
Figure BDA0002376425770000062
Are respectively a one-dimensional random matrix XkElement (b) of (A), NmRepresents the number of samples in the mth time series.
The initial matrix is described as:
Figure BDA0002376425770000063
wherein A denotes an initial matrix, T denotes a matrix transpose, R denotes the number of variables,
Figure BDA0002376425770000064
represents R line NmSet of column matrices, X1、X2、Xk、XRRespectively, the elements of the initial matrix a.
Setting the width of a moving window as d, and converting the initial matrix A into a batch random matrix A through the moving window with the width of dj：
Aj＝(Yj1,Yj2,...,Yjk,...,YjR)T∈CR×d(3)
In the formula, Yj1、Yj2、Yjk、YjRRespectively, a batch random matrix AjVector of medium length d, CR×dRepresenting a matrix set of R rows and d columns, j ∈ [1, Nm-d+1]Is a batch random matrix AjWherein:
Yjk＝(yk1,yk2,...,ykd)∈C1×d(4)
in the formula, yk1、yk2、ykdIs YjkElement (2) of (C)1×dRepresenting a matrix set of l rows and d columns.
In fact, the larger d, the larger the amount of data, the higher the accuracy of SOH (State of Health) evaluation, but the larger the amount of calculation. Therefore, it is necessary to make a reasonable choice to ensure the accuracy and efficiency of the method. According to the random matrix theory single-ring theorem, the following can be obtained:
Figure BDA0002376425770000065
when the appropriate c is selected, d can also be determined from equation (6):
d＝R*c (6)
although d can be determined by equation (6), c is not a fixed value, but a range. Specifically, c can be obtained by the following steps:
first, the minimum value of c is set as:
Figure BDA0002376425770000066
wherein epsilon is a proportional factor for preventing d from being too large;
secondly, taking the selected time sequence as an object, and taking 0.1 as a step length from the minimum value of c to 1 by c to obtain different MSR curves;
thirdly, comparing the obtained different MSR curves, and taking the best c to make the curve monotonically decrease;
finally, other time series are used to determine if c is appropriate.
Note that, after c is obtained, the random matrix A is applied to the batchjIs normalized to obtain a standard non-Hermite matrix
Figure BDA0002376425770000071
The expression of (a) is:
Figure BDA0002376425770000072
in the formula, Bj1、Bj2、Bjk、BjRRespectively, non-Hermite matrix
Figure BDA0002376425770000073
Vector of medium length d.
Figure BDA0002376425770000074
The singular value equivalence matrix of (a) is obtained by equation (9):
Figure BDA0002376425770000075
in the formula (I), the compound is shown in the specification,
Figure BDA0002376425770000076
is composed of
Figure BDA0002376425770000077
The matrix of equivalent singular values of (a),
Figure BDA0002376425770000078
is composed of
Figure BDA0002376425770000079
Is a transposed matrix of (1), U is a Harley momentAnd (5) arraying.
Suppose that
Figure BDA00023764257700000710
Has a characteristic value of
Figure BDA00023764257700000711
Wherein i 1.., n, each in increasing order;
according to the following formula, a non-Hermite matrix is obtained through calculation
Figure BDA00023764257700000712
Characteristic value of
Figure BDA00023764257700000713
Figure BDA00023764257700000714
In the formula (I), the compound is shown in the specification,
Figure BDA00023764257700000715
is composed of
Figure BDA00023764257700000716
The transposed matrix of (2).
Then, a non-Hermite matrix is defined
Figure BDA00023764257700000717
The MSR of the characteristic value of (1) is MSRj,qFor reflecting non-Hermite matrix
Figure BDA00023764257700000718
The distribution of characteristic values of (a), which is obtained by the following formula:
Figure BDA00023764257700000719
in the formula, MSRj,qRepresents the state of health value of the suspension system at the q-th time,
Figure BDA00023764257700000720
is a characteristic value
Figure BDA00023764257700000721
Of (c) is used.
Up to this point, the health value of the suspension system used as an arbitrary time can be obtained by the above equation (11).
As a further preferred embodiment, the MSR of the inventionj,qThe method is used for early warning and judgment of the suspension system, and obtains the fault early warning occurrence time of the suspension system through the following processes:
setting a threshold value theta when MSRj,q>θ, and MSRj,q-1<And when theta is reached, the q-1 moment is the moment of the fault early warning of the suspension system.
Furthermore, it is worth mentioning that a set of historical data matrices obtained according to the fault pre-warning time is defined as H, wherein H ∈ EZ×M，EZ×MThe characteristic value delta of H is diag (delta) representing a matrix set of Z rows and M columns1,...,δZ) Then, the expression of the contribution of each parameter in H is specifically as follows:
Figure BDA0002376425770000081
in the formula, #rIs the contribution, δ, of the r-th eigenvalue in the current staterIs the r-th eigenvalue of H, and Z is the number of H eigenvalues.
As a further improvement, #rThe method is used for suspension system fault pre-diagnosis, and the suspension system fault pre-diagnosis result is obtained through the following processes:
Figure BDA0002376425770000082
in the formula, DtIndicating the distance, psi, between the contribution of each parameter and the contribution of each parameter in the existing fault datar,f,hContribution degree of the r characteristic value of the f group in the h fault, VhIs the number of sets of eigenvalues in the h fault, where:
taking the fault with the minimum distance from the current state as the fault to be generated by the system, namely the fault pre-diagnosis result of the suspension system:
D＝minDt(13)
in the formula, D represents a suspension system predictive failure diagnosis result.
In a word, the invention realizes the predictive fault diagnosis of the suspension system by using the random matrix theory and the Euclidean distance, and has the advantages of simple detection and high reliability.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.

Claims (10)

1. The suspension system fault pre-diagnosis method based on the random matrix theory and the Euclidean distance is characterized by comprising the following steps of:
s1, constructing an initial matrix based on the random matrix theory and big data;
s2, converting the initial matrix into a batch random matrix through a preset mobile window;
s3, converting the batch random matrix into a non-Hermite matrix according to a random matrix theory single-loop theorem;
s4, calculating a characteristic value of the non-Hermite matrix, obtaining an MSR of the characteristic value according to the characteristic value, and taking the MSR as a health state value of the suspension system;
s5, when the health state value MSR of the suspension system reaches an early warning value, calculating the contribution degree of each parameter;
s6, calculating the distance between the contribution degree of each parameter in the step S5 and the contribution degree of each parameter in the existing various fault data, and taking the fault with the minimum distance with the current state as the fault to be generated by the suspension system, namely the predictive fault diagnosis result.
2. The method for suspension system fault pre-diagnosis based on stochastic matrix theory and Euclidean distance according to claim 1, wherein the initial matrix is described by the following expression:
Figure FDA0002376425760000011
wherein A represents an initial matrix, T represents a matrix transposition, R represents the number of variables, NmRepresents the number of samples in the mth time series,
Figure FDA0002376425760000012
represents R line NmSet of column matrices, X1、X2、Xk、XRThe elements of the initial matrix a are respectively expressed as:
Figure FDA0002376425760000013
in the formula, XkOne-dimensional random matrix, x, representing the k-th variable in the m-th time seriesk1、xk2、
Figure FDA0002376425760000014
Are respectively a one-dimensional random matrix XkOf (2) is used.
3. The method for predicting the suspension system fault based on the random matrix theory and the Euclidean distance according to claim 2, wherein the expression of the batch random matrix is as follows:
Aj＝(Yj1,Yj2,...,Yjk,...,YjR)T∈CR×d(3)
in the formula, Yj1、Yj2、Yjk、YjRRespectively, a batch random matrix AjVector of length d, d representing the width of the moving window, CR×dRepresenting a matrix set of R rows and d columns, j ∈ [1, Nm-d+1]Is a batch random matrix AjWherein:
Yjk＝(yk1,yk2,...,ykd)∈Cl×d(4)
in the formula, yk1、yk2、ykdIs YjkElement (2) of (C)1×dRepresenting a matrix set of l rows and d columns.
4. The method for predicting the suspension system fault based on the stochastic matrix theory and the Euclidean distance according to claim 3, wherein the width of the moving window is obtained by the following formula:
d＝R*c (6)
wherein c represents the ratio of determinant and is obtained by the following steps:
1) setting the minimum value of c as:
Figure FDA0002376425760000021
wherein epsilon is a proportional factor for preventing d from being too large;
2) taking the selected time sequence as an object, and taking 0.1 as a step length from the minimum value of c to 1 by c to obtain different MSR curves;
3) comparing the obtained different MSR curves, and taking the optimal c to make the curve monotonically decrease;
4) other time series are used to determine if c is appropriate.
5. The method for predicting the suspension system fault based on the stochastic matrix theory and the Euclidean distance according to claim 4, wherein the non-Hermite matrix in the step S3 is obtained by the following means: after c is obtained, the random matrix A is applied to the batchjIs normalized to obtain a standard non-Hermite matrix
Figure FDA0002376425760000022
Figure FDA0002376425760000023
The expression of (a) is:
Figure FDA0002376425760000024
in the formula, Bj1、Bj2、Bjk、BjRRespectively, non-Hermite matrix
Figure FDA0002376425760000025
Vector of medium length d.
6. The method of claim 5, wherein the non-Hermite matrix is used for suspension system fault pre-diagnosis based on stochastic matrix theory and Euclidean distance
Figure FDA0002376425760000031
The characteristic value of (a) is obtained by the following steps:
is provided with
Figure FDA0002376425760000032
Has a characteristic value of
Figure FDA0002376425760000033
Wherein i 1.., n, each in increasing order;
according to the following formula, a non-Hermite matrix is obtained through calculation
Figure FDA0002376425760000034
Characteristic value of
Figure FDA0002376425760000035
Figure FDA0002376425760000036
In the formula (I), the compound is shown in the specification,
Figure FDA0002376425760000037
is composed of
Figure FDA0002376425760000038
The matrix of equivalent singular values of (a),
Figure FDA0002376425760000039
is composed of
Figure FDA00023764257600000310
The transpose matrix of (a) is,
Figure FDA00023764257600000311
is composed of
Figure FDA00023764257600000312
The transposed matrix of (1), wherein,
Figure FDA00023764257600000313
the following formula is used to solve:
Figure FDA00023764257600000314
in the formula, U is a Harley matrix.
7. The method for suspension system fault pre-diagnosis based on random matrix theory and Euclidean distance according to claim 6, characterized in that a non-Hermite matrix is defined
Figure FDA00023764257600000315
The MSR of the characteristic value of (1) is MSRj,qFor reflecting the matrix
Figure FDA00023764257600000316
The distribution of characteristic values of (a), which is obtained by the following formula:
Figure FDA00023764257600000317
in the formula, MSRj,qRepresents the state of health value of the suspension system at the q-th time,
Figure FDA00023764257600000318
is a characteristic value
Figure FDA00023764257600000319
Of (c) is used.
8. The method of claim 7, wherein the MSR is a method for predicting the suspension system fault based on the stochastic matrix theory and Euclidean distancej,qThe method is used for early warning and judgment of the suspension system, and obtains the fault early warning occurrence time of the suspension system through the following processes:
setting a threshold value theta when MSRj,q>θ, and MSRj,q-1<And when theta is reached, the q-1 moment is the moment of the fault early warning of the suspension system.
9. The method for pre-diagnosing the fault of the suspension system based on the random matrix theory and the Euclidean distance as claimed in claim 8, wherein a group of historical data matrixes obtained according to the fault early warning time are defined as H, wherein H is EZ×M，EZ×MThe characteristic value delta of H is diag (delta) representing a matrix set of Z rows and M columns1,...,δZ) Then, the expression of the contribution of each parameter in H is specifically as follows:
Figure FDA0002376425760000041
in the formula, #rIs the contribution, δ, of the r-th eigenvalue in the current staterIs the r-th eigenvalue of H, and Z is the number of H eigenvalues.
10. The method of claim 9, wherein psi is used for the suspension system fault pre-diagnosis based on the stochastic matrix theory and Euclidean distancerThe method is used for suspension system predictive fault diagnosis, and obtains a suspension system predictive fault diagnosis result through the following formula:
D＝minDt(13)
wherein D represents the predictive failure diagnosis result of the suspension system, DtAnd representing the distance between the contribution degree of each parameter and the contribution degree of each parameter in the existing various fault data, wherein:
Figure FDA0002376425760000042
in the formula, #r,f,hContribution degree of the r characteristic value of the f group in the h fault, VhThe number of sets of eigenvalues in the h-th fault.
CN202010067591.XA 2020-01-20 2020-01-20 Suspension system fault pre-diagnosis method based on random matrix theory and Euclidean distance Active CN111241632B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010067591.XA CN111241632B (en) 2020-01-20 2020-01-20 Suspension system fault pre-diagnosis method based on random matrix theory and Euclidean distance

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010067591.XA CN111241632B (en) 2020-01-20 2020-01-20 Suspension system fault pre-diagnosis method based on random matrix theory and Euclidean distance

Publications (2)

Publication Number Publication Date
CN111241632A true CN111241632A (en) 2020-06-05
CN111241632B CN111241632B (en) 2022-08-16

Family

ID=70869811

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010067591.XA Active CN111241632B (en) 2020-01-20 2020-01-20 Suspension system fault pre-diagnosis method based on random matrix theory and Euclidean distance

Country Status (1)

Country Link
CN (1) CN111241632B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116175281A (en) * 2023-04-26 2023-05-30 成都瑞雪丰泰精密电子股份有限公司 Vibration abnormality detection method for spindle system of machining center

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108152675A (en) * 2017-12-21 2018-06-12 华中科技大学 It is determined and fault zone localization method based on the fault moment of Random Matrices Theory
CN110288022A (en) * 2019-06-26 2019-09-27 华北水利水电大学 A kind of vibration fault diagnosis of hydro-turbine generating unit algorithm of extracting parameter non-linear relation

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108152675A (en) * 2017-12-21 2018-06-12 华中科技大学 It is determined and fault zone localization method based on the fault moment of Random Matrices Theory
CN110288022A (en) * 2019-06-26 2019-09-27 华北水利水电大学 A kind of vibration fault diagnosis of hydro-turbine generating unit algorithm of extracting parameter non-linear relation

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
陈伟彪等: "基于随机矩阵理论的故障时刻确定和故障区域定位方法", 《中国电机工程学报》 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116175281A (en) * 2023-04-26 2023-05-30 成都瑞雪丰泰精密电子股份有限公司 Vibration abnormality detection method for spindle system of machining center
CN116175281B (en) * 2023-04-26 2023-06-23 成都瑞雪丰泰精密电子股份有限公司 Vibration abnormality detection method for spindle system of machining center

Also Published As

Publication number Publication date
CN111241632B (en) 2022-08-16

Similar Documents

Publication Publication Date Title
US11144816B2 (en) Fault identifying method for sludge bulking based on a recurrent RBF neural network
CN107505133A (en) The probability intelligent diagnosing method of rolling bearing fault based on adaptive M RVM
CN103914064B (en) Based on the commercial run method for diagnosing faults that multi-categorizer and D-S evidence merge
CN112990545B (en) Traffic safety state prediction method for expressway intersection area
CN109298225B (en) Automatic identification model system and method for abnormal state of voltage measurement data
CN103103570B (en) Based on the aluminium cell condition diagnostic method of pivot similarity measure
CN107239887A (en) FMEA methods of risk assessments in the case of Factor Weight INFORMATION OF INCOMPLETE
CN112485747A (en) Error state evaluation method for capacitor voltage transformer
CN108022058A (en) A kind of wind energy conversion system state reliability estimation method
CN111178725A (en) Protective equipment state early warning method based on analytic hierarchy process
CN111241632B (en) Suspension system fault pre-diagnosis method based on random matrix theory and Euclidean distance
CN106650959A (en) Power distribution network repair ability assessment method based on improved grey clustering
CN116612098B (en) Insulator RTV spraying quality evaluation method and device based on image processing
CN106208043A (en) A kind of low and medium voltage distribution network key power consumption link discrimination method
CN105403777A (en) Classification and discrimination method of aging state of composite insulator
CN114118789A (en) Radar transmitter state evaluation method based on fuzzy comprehensive evaluation and comprehensive weighting
CN114734873B (en) Power battery monomer thermal runaway early warning method based on cloud online data
CN113406524B (en) Inconsistent fault diagnosis method and system for power battery system
CN111259490B (en) Suspension system early fault detection method based on random matrix theory and cumulative sum
CN108510002A (en) The detection method of rewinding material tractive transformer winding resistance to shorting impact capacity
CN113581015B (en) Safety early warning method and device for fuel cell system
CN114926702A (en) Small sample image classification method based on depth attention measurement
CN114838923A (en) Fault diagnosis model establishing method and fault diagnosis method for on-load tap-changer
CN113313406A (en) Power battery safety risk assessment method for big data of electric vehicle operation
CN111723851A (en) Production line fault detection method

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
GR01 Patent grant
GR01 Patent grant