CN115356631B - Motor state monitoring method and system under high-dimensional variable - Google Patents

Motor state monitoring method and system under high-dimensional variable Download PDF

Info

Publication number
CN115356631B
CN115356631B CN202211299744.9A CN202211299744A CN115356631B CN 115356631 B CN115356631 B CN 115356631B CN 202211299744 A CN202211299744 A CN 202211299744A CN 115356631 B CN115356631 B CN 115356631B
Authority
CN
China
Prior art keywords
dimensional
mode
motor
fault
components
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.)
Active
Application number
CN202211299744.9A
Other languages
Chinese (zh)
Other versions
CN115356631A (en
Inventor
江星星
宋秋昱
陈茜茜
高越
魏勇
郑建颖
朱忠奎
杨强
周振华
陈皓
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.)
Hrlm Technology Inc Co
Original Assignee
Hrlm Technology Inc Co
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 Hrlm Technology Inc Co filed Critical Hrlm Technology Inc Co
Priority to CN202211299744.9A priority Critical patent/CN115356631B/en
Publication of CN115356631A publication Critical patent/CN115356631A/en
Application granted granted Critical
Publication of CN115356631B publication Critical patent/CN115356631B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/34—Testing dynamo-electric machines
    • G01R31/343—Testing dynamo-electric machines in operation
    • 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

Abstract

The invention relates to a method and a system for monitoring the state of a motor under a high-dimensional variable, wherein the method comprises the steps of constructing a high-dimensional motor signal mode frequency searching model based on collected high-dimensional motor vibration signals, and determining mode frequency in the whole analysis frequency spectrum range of the high-dimensional motor vibration signals according to the mode frequency searching model; designing a high-dimensional variable single-step decomposition criterion driven by the mode frequency, and performing high-dimensional variable single-step decomposition operation on the searched mode frequency according to the high-dimensional variable single-step decomposition criterion to obtain a high-dimensional mode component corresponding to the mode frequency; establishing a motor high-dimensional fault feature extraction mechanism, extracting high-dimensional fault components from high-dimensional mode components by using the motor high-dimensional fault feature extraction mechanism, and performing manifold fusion on the high-dimensional fault components to obtain a manifold structure of fault features. The method solves the problems that the decomposition of variable variation partial modes in the traditional motor multichannel vibration signal analysis needs preset parameters and the information quality in the motor high-dimensional fault components is uneven.

Description

Motor state monitoring method and system under high-dimensional variable
Technical Field
The invention belongs to the technical field of intelligent operation and maintenance of motors, and particularly relates to a motor state monitoring method and system under a high-dimensional variable.
Background
The motor is used as a main driving machine of industrial equipment, is widely applied to various industries, and can directly determine the operation efficiency, stability and reliability of production equipment and systems if the motor can normally operate. The motor itself can be damaged by abnormal faults of the motor, the work of an electromechanical transmission system is influenced, the personal safety is endangered, and huge economic loss and social influence are caused. Therefore, the monitoring of the running state of the motor has important significance for ensuring the reliability, safety and high efficiency of the production process.
Early motor state monitoring methods mainly rely on experience judgment of workers, but the methods have strict high requirements on detection personnel, and are low in efficiency and poor in timeliness. With the modernization development of big data, the traditional motor state monitoring method is not enough to meet the important requirements of high safety and high reliability service. Therefore, developing an accurate and effective motor state monitoring method and timely diagnosing motor faults are one of the main challenges facing the intelligent operation and maintenance of the current motors.
The vibration signal is used as the appearance of the motor state and contains rich equipment state information, and the analysis method based on the vibration signal is widely applied to motor state monitoring. The method has obvious signal characteristics, can reflect the running state of the motor in real time, and avoids the defect that the traditional method excessively depends on prior knowledge and engineering experience. However, the traditional vibration signal monitoring method analyzes by collecting vibration data of a single channel, and the information coverage and reliability are insufficient, so that a wrong diagnosis result is easily caused by factors such as sensor faults, accidental errors and the like. With the rapid development of computer and sensor technologies, a multi-channel vibration signal can cover more complete and richer equipment state information, and the misjudgment risk caused by a single-channel error can be avoided. Therefore, the motor state monitoring method based on the multi-channel vibration signals has important engineering application value.
The core problem of the motor state monitoring method based on the multi-channel vibration signals is that real state information and fault characteristics are extracted from the motor high-dimensional variable signals under the background of strong noise. The current high-dimensional variable signal processing method mainly comprises multivariate empirical mode decomposition, multivariate fractional mode decomposition, multivariate fast iterative filtering and the like. The multivariate diversity mode decomposition is a self-adaptive signal decomposition method, which is based on clear mathematical theories such as wiener filtering, hilbert transform, frequency mixing and the like and aims to decompose all mode components through non-recursive screening. However, the decomposition effect is affected by the hyper-parameters, and certain prior knowledge is required as a precondition. In addition, there is an inevitable problem in the high-dimensional variable signal processing that the quality of the failure information contained in the failure mode component obtained by decomposing the high-dimensional variable signal is not uniform, and it is not favorable for the direct and effective failure diagnosis. Therefore, based on the multi-variable variational mode decomposition idea, the method for automatically decomposing the signal and extracting the high-dimensional features without prior knowledge under the condition of high-dimensional variables has great necessity.
Disclosure of Invention
Therefore, the technical problem to be solved by the invention is to overcome the problems that preset parameters are needed for multi-variable diversity mode decomposition in multi-channel vibration signal analysis of the motor and the information quality in high-dimensional fault components of the motor is uneven in the prior art, and provide a method and a system for monitoring the state of the motor under high-dimensional variables.
In order to solve the technical problem, the invention provides a method for monitoring the state of a motor under a high-dimensional variable, which comprises the following steps:
s1: constructing a high-dimensional motor signal mode frequency searching model based on the collected motor high-dimensional vibration signals, and determining mode frequency in the whole analysis frequency spectrum range of the motor high-dimensional vibration signals according to the high-dimensional motor signal mode frequency searching model;
s2: designing a high-dimensional variable single-step decomposition criterion driven by the mode frequency, and performing high-dimensional variable single-step decomposition operation on all searched mode frequencies according to the high-dimensional variable single-step decomposition criterion to obtain high-dimensional mode components corresponding to the mode frequencies;
s3: establishing a motor high-dimensional fault feature extraction mechanism, extracting high-dimensional fault components from all high-dimensional mode components by using the motor high-dimensional fault feature extraction mechanism, performing manifold fusion on the high-dimensional fault components to obtain a manifold structure of fault features, and monitoring the state of the motor.
In one embodiment of the present invention, in S1, a method for determining a mode frequency in the whole analysis spectrum range of a high-dimensional vibration signal of a motor according to a high-dimensional motor signal mode frequency search model includes:
s1.1: collecting variables relating to timetIsCHigh-dimensional vibration signal of dimensional motor
Figure 100002_DEST_PATH_IMAGE001
Wherein the subscriptcRepresenting a dimension sequence number;
s1.2: based on theCConstructing a second vibration signal with respect to the iteration number of 0 by using a high-dimensional vibration signal of the dimensional motoriA starting mode frequency
Figure 100002_DEST_PATH_IMAGE002
Mode frequency search function of
Figure 100002_DEST_PATH_IMAGE003
Comprises the following steps:
Figure 100002_DEST_PATH_IMAGE004
in the formula (I), the compound is shown in the specification,
Figure 100002_DEST_PATH_IMAGE005
is composed of
Figure 100002_DEST_PATH_IMAGE006
In the form of a transform in the fourier domain,
Figure 100002_DEST_PATH_IMAGE007
is a frequency parameter;
s1.3: when starting mode frequency
Figure 100002_DEST_PATH_IMAGE008
Automatically updating the next start mode frequency
Figure 100002_DEST_PATH_IMAGE009
And calculates the corresponding mode frequency search function
Figure 100002_DEST_PATH_IMAGE010
The value of (a), wherein,
Figure 100002_DEST_PATH_IMAGE011
updating the initial mode frequency for the signal sampling frequency
Figure 100002_DEST_PATH_IMAGE012
The formula of (1) is:
Figure 100002_DEST_PATH_IMAGE014
in the formula (I), the compound is shown in the specification,
Figure 100002_DEST_PATH_IMAGE015
is the frequency resolution;
s1.4: calculating the high-dimensional vibration signal analysis frequency spectrum range of the motor
Figure 100002_DEST_PATH_IMAGE016
The values of the pattern frequency search function corresponding to all the initial pattern frequencies are determined, and the value of the search function is changed from positive to negativekOne frequency point is the target mode frequency
Figure 100002_DEST_PATH_IMAGE017
Obtaining all target mode frequencies
Figure 100002_DEST_PATH_IMAGE018
。
In one embodiment of the present invention, in S2, the method for performing the high-dimensional variable single-step decomposition operation on all the searched mode frequencies according to the high-dimensional variable single-step decomposition criterion includes:
s2.1: constructing a high-dimensional variable single-step decomposition model as follows:
Figure 100002_DEST_PATH_IMAGE019
in the formula (I), the compound is shown in the specification,
Figure 100002_DEST_PATH_IMAGE020
is as followscDimension tokIndividual mode component>
Figure 100002_DEST_PATH_IMAGE021
In the form of a transform in the fourier domain,Kis the total number of high-dimensional mode components;
s2.2: all mode frequencies are converted into
Figure 100002_DEST_PATH_IMAGE022
Inputting the data into a high-dimensional variable single-step decomposition model, and calculating to obtain the transformation forms of all mode components in a Fourier domain
Figure 100002_DEST_PATH_IMAGE023
；
S2.3: for is to
Figure 663680DEST_PATH_IMAGE023
Performing inverse fast Fourier transform to obtain time domain representation of all high-dimensional mode components
Figure 100002_DEST_PATH_IMAGE024
。
In an embodiment of the present invention, in S3, the method for extracting high-dimensional fault components from all high-dimensional mode components by using the motor high-dimensional fault feature extraction mechanism and performing manifold fusion on the high-dimensional fault components includes:
s3.1: according to high dimensional fault coefficient mean value
Figure 100002_DEST_PATH_IMAGE025
In all high-dimensional mode components
Figure 100002_DEST_PATH_IMAGE026
A group of high-dimensional fault components are located;
s3.2: performing local tangent space arrangement algorithm on high-dimensional fault components to perform manifold fusion to obtain a manifold structure of fault characteristics
Figure 100002_DEST_PATH_IMAGE027
。
In one embodiment of the invention, in S3.1, the mean value of the fault coefficients is determined according to the high dimension
Figure 100002_DEST_PATH_IMAGE028
In all mode components
Figure 100002_DEST_PATH_IMAGE029
The method for locating a group of high-dimensional fault components comprises the following steps:
s3.1.1: calculating mode components
Figure 100002_DEST_PATH_IMAGE030
Is self-correlation function of
Figure 100002_DEST_PATH_IMAGE031
Comprises the following steps:
Figure 100002_DEST_PATH_IMAGE032
in the formula (I), the compound is shown in the specification,
Figure 100002_DEST_PATH_IMAGE033
is composed of
Figure 100002_DEST_PATH_IMAGE034
The time series after the hilbert transform,
Figure 100002_DEST_PATH_IMAGE035
is the amount of time lag;
s3.1.2: based on autocorrelation function
Figure 100002_DEST_PATH_IMAGE036
Establishing model Components
Figure 100002_DEST_PATH_IMAGE037
High dimensional failure coefficient of
Figure 100002_DEST_PATH_IMAGE038
Comprises the following steps:
Figure 100002_DEST_PATH_IMAGE039
in the formula (I), the compound is shown in the specification,
Figure 100002_DEST_PATH_IMAGE040
to represent
Figure 100002_DEST_PATH_IMAGE041
The amount of time lag in reaching the local maximum,
Figure 100002_DEST_PATH_IMAGE042
representing the total energy of the envelope signal;
s3.1.3: according to high dimensional fault coefficient
Figure 100002_DEST_PATH_IMAGE043
Calculate the firstkGroup mode component
Figure 100002_DEST_PATH_IMAGE044
High dimensional mean fault coefficient
Figure 100002_DEST_PATH_IMAGE045
Comprises the following steps:
Figure 100002_DEST_PATH_IMAGE046
wherein C represents the total number of dimensions;
s3.1.4: according to
Figure 100002_DEST_PATH_IMAGE047
The maximum value selects the corresponding high-dimensional mode component as the high-dimensional fault component data set
Figure 100002_DEST_PATH_IMAGE048
WhereinNIs the data length.
In one embodiment of the invention, in S3.2, the method for performing manifold fusion by using a local cutspace permutation algorithm on the high-dimensional fault components comprises the following steps:
s3.2.1: determining high dimensional fault component data sets
Figure 100002_DEST_PATH_IMAGE049
To middleiAnCOf dimensional data samplesmA neighboring point
Figure 100002_DEST_PATH_IMAGE050
Wherein
Figure 100002_DEST_PATH_IMAGE051
；
S3.2.2: computing
Figure 100002_DEST_PATH_IMAGE052
Front ofdThe largest right singular vector is used as a component
Figure 100002_DEST_PATH_IMAGE053
Is/are as followsdDimensional tangent space orthogonal basis vector
Figure 100002_DEST_PATH_IMAGE054
Wherein, in the step (A),
Figure 100002_DEST_PATH_IMAGE055
is composed of
Figure 100002_DEST_PATH_IMAGE056
The average value of (a) of (b),
Figure 100002_DEST_PATH_IMAGE057
is composed ofmA unit vector of dimensions;
s3.2.3: computing an alignment matrix
Figure 100002_DEST_PATH_IMAGE058
：
Figure 100002_DEST_PATH_IMAGE059
In the formula (I), the compound is shown in the specification,
Figure 100002_DEST_PATH_IMAGE060
is as followsiAnCDimensional data sample
Figure 100002_DEST_PATH_IMAGE061
Is aligned with the matrix 0-1 of (a),
Figure 100002_DEST_PATH_IMAGE062
is a firstiAnCDimensional data samples
Figure 808090DEST_PATH_IMAGE061
The matrix of coefficients of (a) is,
Figure 100002_DEST_PATH_IMAGE063
transposing the matrix;
s3.2.4: computing an alignment matrix
Figure 100002_DEST_PATH_IMAGE064
The feature vector corresponding to the 2 nd minimum feature value is taken as the manifold output of the high-dimensional fault feature
Figure 100002_DEST_PATH_IMAGE065
；
S3.2.5: is calculated at
Figure 100002_DEST_PATH_IMAGE066
All of the rangemCorresponding manifold output
Figure 100002_DEST_PATH_IMAGE067
Entropy of arrangement of
Figure 100002_DEST_PATH_IMAGE068
Selecting the manifold output corresponding to the minimum permutation entropy value as the final manifold structure, wherein the permutation entropy value
Figure 100002_DEST_PATH_IMAGE069
The calculation formula of (2) is as follows:
Figure 100002_DEST_PATH_IMAGE070
in the formula (I), the compound is shown in the specification,
Figure 100002_DEST_PATH_IMAGE071
is composed of
Figure 100002_DEST_PATH_IMAGE072
Reconstruction of the firstiProbability distribution of each permutation.
In addition, the invention also provides a motor state monitoring system under the high-dimensional variable, which comprises:
the mode frequency determining module is used for constructing a high-dimensional motor signal mode frequency searching model based on the collected motor high-dimensional vibration signals and determining the mode frequency in the whole analysis spectrum range of the motor high-dimensional vibration signals according to the high-dimensional motor signal mode frequency searching model;
the high-dimensional variable decomposition module is used for designing a high-dimensional variable single-step decomposition criterion driven by the mode frequency, and performing high-dimensional variable single-step decomposition operation on all searched mode frequencies according to the high-dimensional variable single-step decomposition criterion to obtain high-dimensional mode components corresponding to the mode frequency;
and the fault monitoring module is used for establishing a motor high-dimensional fault feature extraction mechanism, extracting high-dimensional fault components from all high-dimensional mode components by using the motor high-dimensional fault feature extraction mechanism, performing manifold fusion on the high-dimensional fault components to obtain a manifold structure of fault features, and monitoring the state of the motor.
In one embodiment of the present invention, the mode frequency determination module is configured to:
collecting time-related variablestIsCHigh-dimensional vibration signal of dimensional motor
Figure 491575DEST_PATH_IMAGE001
Wherein the subscriptcRepresenting a dimension sequence number;
based on theCConstructing a high-dimensional vibration signal of the dimensional motor about the number of iterations of 0iA starting mode frequency
Figure 637255DEST_PATH_IMAGE002
Mode frequency search function of
Figure 560736DEST_PATH_IMAGE003
Comprises the following steps:
Figure 766458DEST_PATH_IMAGE004
in the formula (I), the compound is shown in the specification,
Figure 38040DEST_PATH_IMAGE005
is composed of
Figure 205321DEST_PATH_IMAGE006
In the form of a transform in the fourier domain,
Figure 308275DEST_PATH_IMAGE007
is a frequency parameter;
when starting mode frequency
Figure 357002DEST_PATH_IMAGE008
Automatically updating the next start mode frequency
Figure 178196DEST_PATH_IMAGE009
And calculates the corresponding mode frequency search function
Figure 606291DEST_PATH_IMAGE010
The value of (a), wherein,
Figure 829331DEST_PATH_IMAGE011
updating the start mode frequency for the signal sampling frequency
Figure 100002_DEST_PATH_IMAGE073
The formula of (1) is as follows:
Figure 707682DEST_PATH_IMAGE014
in the formula (I), the compound is shown in the specification,
Figure 16172DEST_PATH_IMAGE015
is the frequency resolution;
calculating the analysis frequency spectrum range of the motor high-dimensional vibration signal
Figure 245028DEST_PATH_IMAGE016
The values of the pattern frequency search function corresponding to all the initial pattern frequencies are determined, and the value of the search function is changed from positive to negativekOne frequency point is the target mode frequency
Figure 791416DEST_PATH_IMAGE017
Obtaining all target mode frequencies
Figure 716034DEST_PATH_IMAGE018
。
In one embodiment of the invention, the high-dimensional variable decomposition module is configured to:
constructing a high-dimensional variable single-step decomposition model as follows:
Figure DEST_PATH_IMAGE074
in the formula (I), the compound is shown in the specification,
Figure DEST_PATH_IMAGE075
is a firstcDimension tokA mode component
Figure DEST_PATH_IMAGE076
In the form of a transform in the fourier domain,Kis the total number of high-dimensional mode components;
all mode frequencies
Figure 876933DEST_PATH_IMAGE022
Inputting the data into a high-dimensional variable single-step decomposition model, and calculating to obtain all the modelsForm of transformation of formula components in Fourier domain
Figure DEST_PATH_IMAGE077
；
To pair
Figure 830852DEST_PATH_IMAGE077
Performing inverse fast Fourier transform to obtain time domain representation of all high-dimensional mode components
Figure DEST_PATH_IMAGE078
。
In one embodiment of the invention, the fault monitoring module is configured to:
according to high dimensional fault coefficient mean value
Figure 624889DEST_PATH_IMAGE025
In all high-dimensional mode components
Figure 514216DEST_PATH_IMAGE026
A group of high-dimensional fault components are located;
performing local tangent space arrangement algorithm on high-dimensional fault components to perform manifold fusion to obtain manifold structure of fault characteristics
Figure 469403DEST_PATH_IMAGE027
。
Compared with the prior art, the technical scheme of the invention has the following advantages:
(1) The high-dimensional motor signal mode frequency searching model constructed by the method can automatically obtain all mode frequencies without priori knowledge, and overcomes the limitation of artificial preset parameters of the traditional multivariable variation model;
(2) The single-step decomposition criterion of the high-dimensional variable driven by the mode frequency does not need multiple iterations, and all high-dimensional mode components can be obtained only through one-time calculation, so that the decomposition efficiency is greatly improved;
(3) The motor high-dimensional fault feature extraction mechanism established by the invention overcomes the problem of quality difference of fault information in different dimensions, extracts real fault features and effectively improves the accuracy of motor fault diagnosis and state monitoring.
Drawings
In order that the present disclosure may be more readily and clearly understood, reference will now be made in detail to the present disclosure, examples of which are illustrated in the accompanying drawings.
Fig. 1 is a schematic flow chart of a method for monitoring a state of a motor under a high-dimensional variable according to an embodiment of the present invention.
Fig. 2 is a time domain waveform diagram of the 1 st, 5 th and 10 th channel signals in the collected motor high-dimensional channel vibration signal.
Fig. 3 is an envelope spectrogram of the 1 st, 5 th and 10 th channel signals in the acquired motor high-dimensional channel vibration signal.
FIG. 4 shows the search results of the high-dimensional motor signal pattern frequency search model.
Fig. 5 is a graph of the modal components of the 1 st, 5 th and 10 th channels obtained from the modal frequency driven high dimensional variable one step decomposition criterion.
FIG. 6 shows the mean results of the high-dimensional fault coefficients for different high-dimensional mode components.
Fig. 7 is a high-dimensional extracted fault feature manifold structure, in which (a) is a time domain waveform diagram, (b) is a frequency spectrum diagram, and (c) is an envelope spectrum diagram.
Detailed Description
The present invention is further described below in conjunction with the drawings and the embodiments so that those skilled in the art can better understand the present invention and can carry out the present invention, but the embodiments are not to be construed as limiting the present invention.
Referring to fig. 1, a method for monitoring a state of a motor under a high-dimensional variable according to an embodiment of the present invention includes the following steps:
s1: constructing a high-dimensional motor signal mode frequency searching model based on the collected motor high-dimensional vibration signals, and determining mode frequency in the whole analysis frequency spectrum range of the motor high-dimensional vibration signals according to the high-dimensional motor signal mode frequency searching model;
s2: designing a high-dimensional variable single-step decomposition criterion driven by the mode frequency, and performing high-dimensional variable single-step decomposition operation on all searched mode frequencies according to the high-dimensional variable single-step decomposition criterion to obtain high-dimensional mode components corresponding to the mode frequencies;
s3: establishing a motor high-dimensional fault feature extraction mechanism, extracting high-dimensional fault components from all high-dimensional mode components by using the motor high-dimensional fault feature extraction mechanism, performing manifold fusion on the high-dimensional fault components to obtain a manifold structure of fault features, and monitoring the state of the motor.
The method skillfully constructs a high-dimensional motor signal mode frequency searching model, and automatically determines the mode frequencies of all potential mode components in the high-dimensional vibration signal in the whole analysis spectrum range of the motor high-dimensional vibration signal according to the high-dimensional motor signal mode frequency searching model; furthermore, a single-step decomposition criterion of a high-dimensional variable driven by mode frequencies is designed, and all high-dimensional mode components corresponding to the mode frequencies can be decomposed only by one-time calculation based on all the searched mode frequencies; furthermore, a motor high-dimensional fault feature extraction mechanism is established, high-dimensional motor fault components are accurately positioned in all high-dimensional mode components obtained through decomposition according to a high-dimensional fault coefficient mean value, high-dimensional motor fault components are fused through a local tangent space arrangement algorithm, a manifold structure of fault features is obtained, and the motor state is effectively monitored.
Specifically, in S1, the method for determining the mode frequency in the whole analysis spectrum range of the motor high-dimensional vibration signal according to the high-dimensional motor signal mode frequency searching model includes:
s1.1: collecting time-related variablestIs/are as followsCHigh-dimensional vibration signal of dimensional motor
Figure DEST_PATH_IMAGE079
Wherein the subscriptcRepresenting dimension serial numbers, as an example, the time domain oscillogram and the envelope spectrogram of the 1 st, 5 th and 10 th channel signals in the acquired motor high-dimensional channel vibration signals are shown in fig. 2 and fig. 3, and the motor state information in different channels is interfered by noise of different degrees and cannot be directly identifiedState features;
s1.2: based on theCConstructing a high-dimensional vibration signal of the dimensional motor about the number of iterations of 0iA starting mode frequency
Figure DEST_PATH_IMAGE080
Mode frequency search function of
Figure DEST_PATH_IMAGE081
Comprises the following steps:
Figure DEST_PATH_IMAGE082
in the formula (I), the compound is shown in the specification,
Figure DEST_PATH_IMAGE083
is composed of
Figure DEST_PATH_IMAGE084
In the form of a transform in the fourier domain,
Figure DEST_PATH_IMAGE085
is a frequency parameter;
s1.3: when starting mode frequency
Figure 401298DEST_PATH_IMAGE008
Automatically updating the next start mode frequency
Figure 125540DEST_PATH_IMAGE009
And calculating the corresponding pattern frequency search function
Figure 185769DEST_PATH_IMAGE010
The value of (a), wherein,
Figure 631181DEST_PATH_IMAGE011
updating the start mode frequency for the signal sampling frequency
Figure DEST_PATH_IMAGE086
The formula of (1) is as follows:
Figure 661323DEST_PATH_IMAGE014
in the formula (I), the compound is shown in the specification,
Figure 567968DEST_PATH_IMAGE015
is the frequency resolution;
s1.4: calculating the high-dimensional vibration signal analysis frequency spectrum range of the motor
Figure 5290DEST_PATH_IMAGE016
The values of the pattern frequency search function corresponding to all the initial pattern frequencies are determined, and the value of the search function is changed from positive to negativekOne frequency point is the target mode frequency
Figure 935069DEST_PATH_IMAGE017
Obtaining all target mode frequencies
Figure 644268DEST_PATH_IMAGE018
As an example, the search result of the high-dimensional motor signal pattern frequency search model is shown in fig. 4.
Specifically, in S2, the method for performing the high-dimensional variable single-step decomposition operation on all searched mode frequencies according to the high-dimensional variable single-step decomposition criterion includes:
s2.1: constructing a high-dimensional variable single-step decomposition model as follows:
Figure DEST_PATH_IMAGE087
in the formula (I), the compound is shown in the specification,
Figure 532983DEST_PATH_IMAGE075
is as followscMaintenance ofkA mode component
Figure DEST_PATH_IMAGE088
In the form of a transform in the fourier domain,Kis the total number of high-dimensional mode components;
s2.2: all mode frequencies are converted into
Figure 528489DEST_PATH_IMAGE022
Inputting the data into a high-dimensional variable single-step decomposition model, and calculating to obtain the transformation forms of all mode components in a Fourier domain
Figure DEST_PATH_IMAGE089
；
S2.3: to pair
Figure 389305DEST_PATH_IMAGE089
Performing inverse fast Fourier transform to obtain time domain representation of all high-dimensional mode components
Figure DEST_PATH_IMAGE090
Illustratively, the modal frequency-driven high-dimensional variable one-step decomposition criterion results in the modal components of the 1 st, 5 th and 10 th channels as shown in fig. 5.
Specifically, in S3, the method for extracting high-dimensional fault components from all high-dimensional mode components by using the motor high-dimensional fault feature extraction mechanism and performing manifold fusion on the high-dimensional fault components to obtain a manifold structure of fault features includes:
s3.1: according to high dimensional fault coefficient mean value
Figure DEST_PATH_IMAGE091
In all high-dimensional mode components
Figure DEST_PATH_IMAGE092
A group of high-dimensional fault components are located, and the result of the mean value of the high-dimensional fault coefficients of different high-dimensional mode components is shown in fig. 6 as an example;
s3.2: performing local tangent space arrangement algorithm on high-dimensional fault components to perform manifold fusion to obtain a manifold structure of fault characteristics
Figure DEST_PATH_IMAGE094
. Illustratively, the extracted high-dimensional fault feature manifold structure is shown in fig. 7.
Specifically, in S3.1,according to high dimensional fault coefficient mean value
Figure 669239DEST_PATH_IMAGE028
In all mode components
Figure 287827DEST_PATH_IMAGE090
The method for locating a group of high-dimensional fault components comprises the following steps:
s3.1.1: computing model components
Figure 798442DEST_PATH_IMAGE090
Is self-correlation function of
Figure 968393DEST_PATH_IMAGE031
Comprises the following steps:
Figure 284973DEST_PATH_IMAGE032
in the formula (I), the compound is shown in the specification,
Figure 492488DEST_PATH_IMAGE033
is composed of
Figure 439585DEST_PATH_IMAGE034
The time series after the hilbert transform,
Figure 831252DEST_PATH_IMAGE035
is the amount of time lag;
s3.1.2: based on autocorrelation function
Figure 700593DEST_PATH_IMAGE036
Establishing model Components
Figure 25263DEST_PATH_IMAGE037
High dimensional failure coefficient of
Figure 408840DEST_PATH_IMAGE038
Comprises the following steps:
Figure 22224DEST_PATH_IMAGE039
in the formula (I), the compound is shown in the specification,
Figure 683537DEST_PATH_IMAGE040
represent
Figure 862715DEST_PATH_IMAGE041
The amount of time lag in reaching the local maximum,
Figure 151614DEST_PATH_IMAGE042
representing the total energy of the envelope signal;
s3.1.3: according to high dimensional fault coefficient
Figure 255223DEST_PATH_IMAGE043
Calculate the firstkGroup mode component
Figure 717298DEST_PATH_IMAGE044
High dimensional mean fault coefficient
Figure 219823DEST_PATH_IMAGE045
Comprises the following steps:
Figure 476361DEST_PATH_IMAGE046
wherein C represents the total number of dimensions;
s3.1.4: according to
Figure 67267DEST_PATH_IMAGE047
The maximum value selects the corresponding high-dimensional mode component as the high-dimensional fault component data set
Figure 67453DEST_PATH_IMAGE048
WhereinNIs the data length.
Specifically, in S3.2, the method for performing manifold fusion by the local tangent space arrangement algorithm on the high-dimensional fault component includes:
s3.2.1: determining high dimensional fault component data sets
Figure 424485DEST_PATH_IMAGE049
To middleiAnCOf dimensional data samplesmA neighboring point
Figure 843135DEST_PATH_IMAGE050
Wherein
Figure 183986DEST_PATH_IMAGE051
；
S3.2.2: calculating out
Figure 925546DEST_PATH_IMAGE052
Front of (2)dThe largest right singular vector is used as a component
Figure 199402DEST_PATH_IMAGE053
Is/are as followsdDimensional tangent space orthogonal basis vector
Figure 3934DEST_PATH_IMAGE054
Wherein, in the step (A),
Figure DEST_PATH_IMAGE095
is composed of
Figure DEST_PATH_IMAGE096
The average value of (a) of (b),
Figure DEST_PATH_IMAGE097
is composed ofmA unit vector of dimensions;
s3.2.3: computing an alignment matrix
Figure 271230DEST_PATH_IMAGE058
：
Figure 347639DEST_PATH_IMAGE059
In the formula (I), the compound is shown in the specification,
Figure 944842DEST_PATH_IMAGE060
is as followsiAnCDimensional data sample
Figure 654697DEST_PATH_IMAGE096
The 0-1 alignment matrix of (a),
Figure 501299DEST_PATH_IMAGE062
is as followsiAnCDimensional data sample
Figure 584661DEST_PATH_IMAGE096
The matrix of coefficients of (a) is,
Figure 833109DEST_PATH_IMAGE063
transposing the matrix;
s3.2.4: computing an alignment matrix
Figure 991163DEST_PATH_IMAGE064
The feature vector corresponding to the 2 nd minimum feature value is taken as the manifold output of the high-dimensional fault feature
Figure 325061DEST_PATH_IMAGE065
；
S3.2.5: is calculated at
Figure 743273DEST_PATH_IMAGE066
Range all ofmCorresponding manifold output
Figure 52419DEST_PATH_IMAGE067
Entropy of arrangement of (2)
Figure 366726DEST_PATH_IMAGE068
Selecting the manifold output corresponding to the minimum permutation entropy value as the final manifold structure, wherein the permutation entropy value
Figure 187920DEST_PATH_IMAGE069
The calculation formula of (2) is as follows:
Figure 613085DEST_PATH_IMAGE070
in the formula (I), the compound is shown in the specification,
Figure DEST_PATH_IMAGE098
is composed of
Figure 432530DEST_PATH_IMAGE072
The first of reconstructioniProbability distribution of individual permutations.
The high-dimensional motor signal mode frequency searching model constructed by the invention can automatically obtain all mode frequencies without prior knowledge, and overcomes the limitation of artificial preset parameters of the traditional multivariable variation model.
The mode frequency driven high-dimensional variable single-step decomposition criterion designed by the invention does not need multiple iterations, and all high-dimensional mode components can be obtained by one-time calculation, thereby greatly improving the decomposition efficiency.
The high-dimensional motor fault feature extraction mechanism overcomes the problem of quality difference of fault information in different dimensions, extracts real fault features and effectively improves the accuracy of motor fault diagnosis and state monitoring.
In the following, a system for monitoring a state of a motor under a high-dimensional variable disclosed in an embodiment of the present invention is introduced, and a system for monitoring a state of a motor under a high-dimensional variable described below and a method for monitoring a state of a motor under a high-dimensional variable described above may be referred to in a corresponding manner.
An embodiment of the present invention further provides a system for monitoring a state of a motor under a high-dimensional variable, including:
the mode frequency determining module is used for constructing a high-dimensional motor signal mode frequency searching model based on the collected motor high-dimensional vibration signals, and determining the mode frequency in the whole analysis spectrum range of the motor high-dimensional vibration signals according to the high-dimensional motor signal mode frequency searching model;
the high-dimensional variable decomposition module is used for designing a high-dimensional variable single-step decomposition criterion driven by the mode frequency, and performing high-dimensional variable single-step decomposition operation on all searched mode frequencies according to the high-dimensional variable single-step decomposition criterion to obtain high-dimensional mode components corresponding to the mode frequency;
and the fault monitoring module is used for establishing a motor high-dimensional fault feature extraction mechanism, extracting high-dimensional fault components from all high-dimensional mode components by using the motor high-dimensional fault feature extraction mechanism, performing manifold fusion on the high-dimensional fault components to obtain a manifold structure of fault features, and monitoring the state of the motor.
In one embodiment of the present invention, the mode frequency determination module is configured to:
collecting time-related variablestIs/are as followsCHigh-dimensional vibration signal of dimensional motor
Figure 714476DEST_PATH_IMAGE001
Wherein the subscriptcRepresenting a dimension number;
based on theCConstructing a high-dimensional vibration signal of the dimensional motor about the number of iterations of 0iA starting mode frequency
Figure 88213DEST_PATH_IMAGE002
Mode frequency search function of
Figure 582648DEST_PATH_IMAGE003
Comprises the following steps:
Figure DEST_PATH_IMAGE099
in the formula (I), the compound is shown in the specification,
Figure 510460DEST_PATH_IMAGE005
is composed of
Figure 228886DEST_PATH_IMAGE006
In the form of a transform in the fourier domain,
Figure 962356DEST_PATH_IMAGE007
is a frequency parameter;
when starting mode frequency
Figure 791640DEST_PATH_IMAGE008
Automatically update the dataNext start mode frequency
Figure 664306DEST_PATH_IMAGE009
And calculates the corresponding mode frequency search function
Figure 288054DEST_PATH_IMAGE010
The value of (a), wherein,
Figure 508820DEST_PATH_IMAGE011
updating the start mode frequency for the signal sampling frequency
Figure DEST_PATH_IMAGE100
The formula of (1) is:
Figure 738200DEST_PATH_IMAGE014
in the formula (I), the compound is shown in the specification,
Figure 524760DEST_PATH_IMAGE015
is the frequency resolution;
calculating the analysis frequency spectrum range of the motor high-dimensional vibration signal
Figure 788251DEST_PATH_IMAGE016
The values of the mode frequency search function corresponding to all the initial mode frequencies in the set are determined, and the value of the mode frequency search function is changed from positive to negativekOne frequency point is the target mode frequency
Figure 30401DEST_PATH_IMAGE017
Obtaining all target mode frequencies
Figure 404750DEST_PATH_IMAGE018
。
In one embodiment of the invention, the high-dimensional variable decomposition module is configured to:
constructing a high-dimensional variable single-step decomposition model as follows:
Figure 45816DEST_PATH_IMAGE074
in the formula (I), the compound is shown in the specification,
Figure 480209DEST_PATH_IMAGE075
is as followscDimension tokA mode component
Figure DEST_PATH_IMAGE101
In the form of a transform in the fourier domain,Kis the total number of high-dimensional mode components;
all mode frequencies
Figure 346007DEST_PATH_IMAGE022
Inputting the data into a high-dimensional variable single-step decomposition model, and calculating to obtain the transformation forms of all mode components in a Fourier domain
Figure DEST_PATH_IMAGE102
；
To pair
Figure DEST_PATH_IMAGE103
Performing inverse fast Fourier transform to obtain time domain representation of all high-dimensional mode components
Figure DEST_PATH_IMAGE104
。
In one embodiment of the present invention, the fault monitoring module is configured to:
according to high dimensional fault coefficient mean value
Figure 963196DEST_PATH_IMAGE025
In all high-dimensional mode components
Figure 724347DEST_PATH_IMAGE026
A group of high-dimensional fault components are located;
performing local tangent space arrangement algorithm on high-dimensional fault components to perform manifold fusion to obtain manifold structure of fault characteristics
Figure 332570DEST_PATH_IMAGE027
。
The system for monitoring the state of a motor under a high-dimensional variable of this embodiment is used to implement the method for monitoring the state of a motor under a high-dimensional variable described above, and therefore, the specific implementation of the system can be seen from the above section of the embodiment of the method for monitoring the state of a motor under a high-dimensional variable, and therefore, the specific implementation thereof can refer to the description of the corresponding section of the embodiment, and will not be further described herein.
In addition, since the system for monitoring the state of the motor under the high-dimensional variable of this embodiment is used for implementing the method for monitoring the state of the motor under the high-dimensional variable, the function of the system corresponds to that of the method, and details are not described here.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
It should be understood that the above examples are only for clarity of illustration and are not intended to limit the embodiments. Various other modifications and alterations will occur to those skilled in the art upon reading the foregoing description. And are neither required nor exhaustive of all embodiments. And obvious variations or modifications of the invention may be made without departing from the spirit or scope of the invention.

Claims (10)

1. A method for monitoring the state of a motor under a high-dimensional variable is characterized by comprising the following steps:
s1: constructing a high-dimensional motor signal mode frequency searching model based on the collected motor high-dimensional vibration signals, and determining mode frequency in the whole analysis frequency spectrum range of the motor high-dimensional vibration signals according to the high-dimensional motor signal mode frequency searching model;
s2: designing a high-dimensional variable single-step decomposition criterion driven by the mode frequency, and performing high-dimensional variable single-step decomposition operation on all searched mode frequencies according to the high-dimensional variable single-step decomposition criterion to obtain high-dimensional mode components corresponding to the mode frequencies;
s3: establishing a motor high-dimensional fault feature extraction mechanism, extracting high-dimensional fault components from all high-dimensional mode components by using the motor high-dimensional fault feature extraction mechanism, performing manifold fusion on the high-dimensional fault components to obtain a manifold structure of fault features, and monitoring the state of the motor.
2. The method for monitoring motor state under high dimensional variable according to claim 1, wherein in S1, the method for determining the mode frequency in the whole analysis spectrum range of the motor high dimensional vibration signal according to the high dimensional motor signal mode frequency searching model comprises:
s1.1: collecting time-related variablestIsCHigh-dimensional vibration signal of dimensional motor
Figure DEST_PATH_IMAGE001
Wherein the subscriptcRepresenting a dimension sequence number;
s1.2: based on theCConstructing a high-dimensional vibration signal of the dimensional motor about the number of iterations of 0iA starting mode frequency
Figure DEST_PATH_IMAGE002
Mode frequency search function of
Figure DEST_PATH_IMAGE003
Comprises the following steps:
Figure DEST_PATH_IMAGE004
in the formula (I), the compound is shown in the specification,
Figure DEST_PATH_IMAGE005
is composed of
Figure DEST_PATH_IMAGE006
In the form of a transform in the fourier domain,
Figure DEST_PATH_IMAGE007
is a frequency parameter;
s1.3: when starting mode frequency
Figure DEST_PATH_IMAGE008
Automatically updating the next start mode frequency
Figure DEST_PATH_IMAGE009
And calculates the corresponding mode frequency search function
Figure DEST_PATH_IMAGE010
The value of (a), wherein,
Figure DEST_PATH_IMAGE011
updating the initial mode frequency for the signal sampling frequency
Figure DEST_PATH_IMAGE012
The formula of (1) is as follows:
Figure DEST_PATH_IMAGE014
in the formula (I), the compound is shown in the specification,
Figure DEST_PATH_IMAGE015
is the frequency resolution;
s1.4: calculating the high-dimensional vibration signal analysis frequency spectrum range of the motor
Figure DEST_PATH_IMAGE016
The values of the mode frequency search function corresponding to all the initial mode frequencies in the set are determined, and the value of the mode frequency search function is changed from positive to negativekOne frequency point is the target mode frequency
Figure DEST_PATH_IMAGE017
Obtaining all target mode frequencies
Figure DEST_PATH_IMAGE018
。
3. The method for monitoring the state of a motor under high-dimensional variable according to claim 2, wherein in S2, the method for performing the high-dimensional variable single-step decomposition operation on all searched mode frequencies according to the high-dimensional variable single-step decomposition criterion comprises:
s2.1: the high-dimensional variable single-step decomposition model is constructed as follows:
Figure DEST_PATH_IMAGE019
in the formula (I), the compound is shown in the specification,
Figure DEST_PATH_IMAGE020
is a firstcMaintenance ofkA mode component
Figure DEST_PATH_IMAGE021
In the form of a transform in the fourier domain,Kis the total number of high-dimensional mode components;
s2.2: all mode frequencies
Figure DEST_PATH_IMAGE022
Inputting the data into a high-dimensional variable single-step decomposition model, and calculating to obtain the transformation forms of all mode components in a Fourier domain
Figure DEST_PATH_IMAGE023
；
S2.3: to pair
Figure 100460DEST_PATH_IMAGE023
Performing inverse fast Fourier transform to obtain time domain representation of all high-dimensional mode components
Figure DEST_PATH_IMAGE024
。
4. The method for monitoring the state of the motor under the high-dimensional variable according to claim 3, wherein in S3, the method for extracting the high-dimensional fault components from all the high-dimensional mode components by using the motor high-dimensional fault feature extraction mechanism and performing manifold fusion on the high-dimensional fault components comprises the following steps:
s3.1: according to high dimensional fault coefficient mean value
Figure DEST_PATH_IMAGE025
In all high-dimensional mode components
Figure 660011DEST_PATH_IMAGE024
A group of high-dimensional fault components are located;
s3.2: performing local tangent space arrangement algorithm on high-dimensional fault components to perform manifold fusion to obtain manifold structure of fault characteristics
Figure DEST_PATH_IMAGE026
。
5. The method of claim 4, wherein in S3.1, the mean value of the fault coefficients is determined according to the high-dimensional variables
Figure DEST_PATH_IMAGE027
In all mode components
Figure 144475DEST_PATH_IMAGE024
The method for locating a group of high-dimensional fault components comprises the following steps:
s3.1.1: computing model components
Figure 398739DEST_PATH_IMAGE024
Is self-correlation function of
Figure DEST_PATH_IMAGE028
Comprises the following steps:
Figure DEST_PATH_IMAGE029
in the formula (I), the compound is shown in the specification,
Figure DEST_PATH_IMAGE030
is composed of
Figure DEST_PATH_IMAGE031
The time series after the hilbert transform,
Figure DEST_PATH_IMAGE032
is the amount of time lag;
s3.1.2: based on autocorrelation function
Figure DEST_PATH_IMAGE033
Establishing model Components
Figure DEST_PATH_IMAGE034
High dimensional failure coefficient of
Figure DEST_PATH_IMAGE035
Comprises the following steps:
Figure DEST_PATH_IMAGE036
in the formula (I), the compound is shown in the specification,
Figure DEST_PATH_IMAGE037
to represent
Figure DEST_PATH_IMAGE038
The amount of time lag when the local maximum is reached,
Figure DEST_PATH_IMAGE039
representing the total energy of the envelope signal;
s3.1.3: according to high dimensional fault coefficient
Figure DEST_PATH_IMAGE040
Calculate the firstkGroup mode component
Figure DEST_PATH_IMAGE041
High dimensional fault coefficient mean of
Figure DEST_PATH_IMAGE042
Comprises the following steps:
Figure DEST_PATH_IMAGE043
wherein C represents the total number of dimensions;
s3.1.4: according to
Figure DEST_PATH_IMAGE044
Selecting corresponding high-dimensional mode components as high-dimensional fault component data sets by maximum values
Figure DEST_PATH_IMAGE045
WhereinNIs the data length.
6. The method for monitoring the state of a motor under high-dimensional variables according to claim 5, wherein in S3.2, the method for performing manifold fusion by a local tangent space arrangement algorithm on high-dimensional fault components comprises the following steps:
s3.2.1: determining high dimensional fault component data sets
Figure DEST_PATH_IMAGE046
To middleiAnCOf dimensional data samplesmA neighboring point
Figure DEST_PATH_IMAGE047
Wherein
Figure DEST_PATH_IMAGE048
；
S3.2.2: computing
Figure DEST_PATH_IMAGE049
Front of (2)dThe largest right singular vector is used as a component
Figure DEST_PATH_IMAGE050
Is/are as followsdWei-Jie (food cutting)Space orthogonal basis vector
Figure DEST_PATH_IMAGE051
Wherein, in the step (A),
Figure DEST_PATH_IMAGE052
is composed of
Figure DEST_PATH_IMAGE053
The average value of (a) of (b),
Figure DEST_PATH_IMAGE054
is composed ofmA unit vector of dimensions;
s3.2.3: computing an alignment matrix
Figure DEST_PATH_IMAGE055
：
Figure DEST_PATH_IMAGE056
In the formula (I), the compound is shown in the specification,
Figure DEST_PATH_IMAGE057
is a firstiAnCDimensional data samples
Figure 511314DEST_PATH_IMAGE053
The 0-1 alignment matrix of (a),
Figure DEST_PATH_IMAGE058
is as followsiAnCDimensional data sample
Figure 114202DEST_PATH_IMAGE053
The matrix of coefficients of (a) is,
Figure DEST_PATH_IMAGE059
transposing the matrix;
s3.2.4: computing an alignment matrix
Figure DEST_PATH_IMAGE060
The feature vector corresponding to the 2 nd minimum feature value is taken as the manifold output of the high-dimensional fault feature
Figure DEST_PATH_IMAGE061
；
S3.2.5: is calculated at
Figure DEST_PATH_IMAGE062
All of the rangemCorresponding manifold output
Figure DEST_PATH_IMAGE063
Entropy of arrangement of
Figure DEST_PATH_IMAGE064
Selecting the manifold output corresponding to the minimum permutation entropy value as the final manifold structure, wherein the permutation entropy value
Figure DEST_PATH_IMAGE065
The calculation formula of (2) is as follows:
Figure DEST_PATH_IMAGE066
in the formula (I), the compound is shown in the specification,
Figure DEST_PATH_IMAGE067
is composed of
Figure DEST_PATH_IMAGE068
The first of reconstructioniProbability distribution of individual permutations.
7. A motor state monitoring system under high dimensional variables, comprising:
the mode frequency determining module is used for constructing a high-dimensional motor signal mode frequency searching model based on the collected motor high-dimensional vibration signals, and determining the mode frequency in the whole analysis spectrum range of the motor high-dimensional vibration signals according to the high-dimensional motor signal mode frequency searching model;
the high-dimensional variable decomposition module is used for designing a high-dimensional variable single-step decomposition criterion driven by the mode frequency, and performing high-dimensional variable single-step decomposition operation on all searched mode frequencies according to the high-dimensional variable single-step decomposition criterion to obtain high-dimensional mode components corresponding to the mode frequency;
and the fault monitoring module is used for establishing a motor high-dimensional fault feature extraction mechanism, extracting high-dimensional fault components from all high-dimensional mode components by using the motor high-dimensional fault feature extraction mechanism, performing manifold fusion on the high-dimensional fault components to obtain a manifold structure of fault features, and monitoring the state of the motor.
8. The system of claim 7, wherein the mode frequency determination module is configured to:
collecting time-related variablestIsCHigh-dimensional vibration signal of dimensional motor
Figure 493754DEST_PATH_IMAGE001
Wherein the subscriptcRepresenting a dimension number;
based on theCConstructing a second vibration signal with respect to the iteration number of 0 by using a high-dimensional vibration signal of the dimensional motoriA starting mode frequency
Figure 348446DEST_PATH_IMAGE002
Mode frequency search function of
Figure 673117DEST_PATH_IMAGE003
Comprises the following steps:
Figure DEST_PATH_IMAGE069
in the formula (I), the compound is shown in the specification,
Figure 653099DEST_PATH_IMAGE005
is composed of
Figure 328800DEST_PATH_IMAGE006
In the form of a transform in the fourier domain,
Figure 190446DEST_PATH_IMAGE007
is a frequency parameter;
when starting mode frequency
Figure 575815DEST_PATH_IMAGE008
Automatically updating the next start mode frequency
Figure DEST_PATH_IMAGE070
And calculating the corresponding pattern frequency search function
Figure 317244DEST_PATH_IMAGE010
The value of (a), wherein,
Figure 483171DEST_PATH_IMAGE011
updating the initial mode frequency for the signal sampling frequency
Figure 882928DEST_PATH_IMAGE070
The formula of (1) is as follows:
Figure 385453DEST_PATH_IMAGE014
in the formula (I), the compound is shown in the specification,
Figure 641991DEST_PATH_IMAGE015
is the frequency resolution;
calculating the analysis frequency spectrum range of the motor high-dimensional vibration signal
Figure 486758DEST_PATH_IMAGE016
All start mode frequenciesThe value of the corresponding pattern frequency search function is determined to be the first to change from positive to negativekOne frequency point is the target mode frequency
Figure 424627DEST_PATH_IMAGE017
Obtaining all target mode frequencies
Figure 47238DEST_PATH_IMAGE018
。
9. The system of claim 8, wherein the high-dimensional variable decomposition module is configured to:
constructing a high-dimensional variable single-step decomposition model as follows:
Figure DEST_PATH_IMAGE071
in the formula (I), the compound is shown in the specification,
Figure 867820DEST_PATH_IMAGE020
is a firstcMaintenance ofkA mode component
Figure DEST_PATH_IMAGE072
In the form of a transform in the fourier domain,Kis the total number of high-dimensional mode components;
all mode frequencies
Figure 598884DEST_PATH_IMAGE022
Inputting the data into a high-dimensional variable single-step decomposition model, and calculating to obtain the transformation forms of all mode components in a Fourier domain
Figure DEST_PATH_IMAGE073
；
To pair
Figure 468008DEST_PATH_IMAGE073
Performing fast Fourier inversionIn the alternative, a time-domain representation of all high-dimensional mode components is obtained>
Figure 741863DEST_PATH_IMAGE072
。
10. The system of claim 9, wherein the fault monitoring module is configured to:
according to high dimensional fault coefficient mean value
Figure 809045DEST_PATH_IMAGE025
In all high-dimensional mode components
Figure 108964DEST_PATH_IMAGE072
A group of high-dimensional fault components are positioned;
performing local tangent space arrangement algorithm on high-dimensional fault components to perform manifold fusion to obtain manifold structure of fault characteristics
Figure 654215DEST_PATH_IMAGE026
。
CN202211299744.9A 2022-10-24 2022-10-24 Motor state monitoring method and system under high-dimensional variable Active CN115356631B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211299744.9A CN115356631B (en) 2022-10-24 2022-10-24 Motor state monitoring method and system under high-dimensional variable

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211299744.9A CN115356631B (en) 2022-10-24 2022-10-24 Motor state monitoring method and system under high-dimensional variable

Publications (2)

Publication Number Publication Date
CN115356631A CN115356631A (en) 2022-11-18
CN115356631B true CN115356631B (en) 2023-04-14

Family

ID=84007945

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211299744.9A Active CN115356631B (en) 2022-10-24 2022-10-24 Motor state monitoring method and system under high-dimensional variable

Country Status (1)

Country Link
CN (1) CN115356631B (en)

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103048041B (en) * 2012-12-20 2014-11-26 北京信息科技大学 Fault diagnosis method of electromechanical system based on local tangent space and support vector machine
CN103953490A (en) * 2014-04-23 2014-07-30 浙江工业大学 Implementation method for monitoring status of hydraulic turbine set based on HLSNE
CN106096154B (en) * 2016-06-17 2019-05-28 北京信息科技大学 A kind of wavelet packet time-domain signal manifold learning method for diagnosing faults
CN113702044B (en) * 2021-08-13 2022-04-19 华中科技大学 Bearing fault detection method and system
CN114186578A (en) * 2021-08-30 2022-03-15 华北电力大学 Gas turbine electric actuating mechanism fault diagnosis method based on signal decomposition

Also Published As

Publication number Publication date
CN115356631A (en) 2022-11-18

Similar Documents

Publication Publication Date Title
JP2012528392A5 (en)
CN109765055B (en) Rolling bearing fault detection method and system based on EWT, spectrum effective value and KNN
CN112798280B (en) Rolling bearing fault diagnosis method and system
EP2239642B1 (en) Analysis method
CN113177537B (en) Fault diagnosis method and system for rotary mechanical equipment
CN113255835A (en) Hydropower station pump equipment anomaly detection method
CN112649196B (en) Signal variation modal decomposition preset scale parameter selection method based on frequency domain information
CN116304909A (en) Abnormality detection model training method, fault scene positioning method and device
CN112000081A (en) Fault monitoring method and system based on multi-block information extraction and Mahalanobis distance
CN112180315A (en) Fault feature extraction method, device and system for optical fiber current transformer
CN103914386B (en) Software defect positioning method based on input parameter characteristic spectrum
CN115356631B (en) Motor state monitoring method and system under high-dimensional variable
CN111291918A (en) Rotating machine degradation trend prediction method based on stationary subspace exogenous vector autoregression
CN111160393A (en) Data-driven modularized modeling method for carrier rocket health assessment model
Wang et al. Fast spectral correlation based on sparse representation self-learning dictionary and its application in fault diagnosis of rotating machinery
CN110222390B (en) Gear crack identification method based on wavelet neural network
CN110674791B (en) Forced oscillation layered positioning method based on multi-stage transfer learning
US11449781B2 (en) Plant abnormality prediction system and method
CN114088387A (en) Gear fault diagnosis method, storage medium and device
CN115358286A (en) Helicopter vibration health state monitoring method
CN114357667A (en) Engine starting state fault detection method based on RESID recursive structure identification
CN110441081B (en) Intelligent diagnosis method and intelligent diagnosis system for faults of rotating machinery
Prist et al. Online fault detection: A smart approach for industry 4.0
CN113702038B (en) Fault diagnosis method and system for explosion-proof motor bearing
CN115555920B (en) Online chatter detection method and system based on adaptive variation modal decomposition

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