CN109033031B - Bearing state detection method based on high-dimensional random matrix - Google Patents
Bearing state detection method based on high-dimensional random matrix Download PDFInfo
- Publication number
- CN109033031B CN109033031B CN201810816943.XA CN201810816943A CN109033031B CN 109033031 B CN109033031 B CN 109033031B CN 201810816943 A CN201810816943 A CN 201810816943A CN 109033031 B CN109033031 B CN 109033031B
- Authority
- CN
- China
- Prior art keywords
- matrix
- normal
- data
- state
- detection
- 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
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
Abstract
The invention provides a bearing state detection method based on a high-dimensional random matrix, which is characterized by comprising the following steps of: establishing a monitoring network, collecting operation data to form a data source X, selecting a normal operation space-time section from the data source X, constructing a normal database X1, constructing a normal original measurement matrix X2, constructing a normal judgment matrix Z, solving a normal judgment matrix Z characteristic value, and constructing a normal state visual parameter database; selecting unknown space-time sections from a data source, constructing an original measurement matrix D1, constructing a detection matrix D2, constructing a detection decision matrix D3, solving characteristic values of the detection decision matrix, and constructing a visual reference database of a detection state; and comparing the visual reference database of the detection state with the visual parameter database of the normal state, and judging whether the abnormal condition exists or not, so that the bearing state is judged to be normal, abnormal or fault, and the measurement efficiency is high.
Description
Technical Field
The invention belongs to the technical field of rolling bearing carrier running state detection, and particularly relates to a bearing state detection method based on a high-dimensional random matrix.
Background
The existing mechanical equipment is more precise, more complex and more expensive, and can only obtain the operation data under the operation state, the data of the fault state is difficult to obtain, and a full-life model can not be established, so that the state of the bearing can not be obtained. However, a plurality of bearings are used in mechanical equipment, but whether the bearings are in failure or abnormal in the operation process cannot be known, in the actual production process, the machine needs to be stopped for inspection, and one bearing is checked, so that the working efficiency is low; moreover, a sudden failure may even affect the service life of the entire mechanical device.
Disclosure of Invention
The invention aims to provide a bearing state detection method based on a high-dimensional random matrix to solve the problems in the background art.
In order to solve the above technical problem, an embodiment of the present invention provides a bearing state detection method based on a high-dimensional random matrix, which is characterized by including the following steps:
s1: under the condition of lacking fault state signals, designing a multi-point monitoring network aiming at a rolling bearing to be detected in work, collecting vibration signals of all monitoring points of the rolling bearing to be detected in the work process, storing the collected real-time data in a high-dimensional matrix form to obtain a data source X, selecting a normal operation state space-time section from the data source X, and constructing a normal state reference database X1;
s2: intercepting a space-time section from a normal state reference database X1 according to specific time t and frequency f, and performing feature extraction on data in X1, wherein the data after feature extraction form a high-dimensional matrix which is called a normal original measurement matrix X2;
s3: normalizing the normal original measurement matrix X2 to obtain a matrix X2 ', and then performing singular value decomposition on the matrix X2' to obtain a normal measurement matrix X3; carrying out standardization processing on the normal measurement matrix X3 to form a normal 'decision matrix' Z, and carrying out eigenvalue solution on the normal 'decision matrix' Z; representing the characteristic value of a normal 'decision matrix' Z on a complex plane by using a data visualization processing method; by adopting the single-ring theory, the radius of the outer ring is 1, the radius of the inner ring is (1-c) L/2 The ring completes the construction of a normal state visual reference database;
s4: intercepting a space-time section with an unknown state from a data source X according to the same time t and frequency f, and forming an original detection matrix D1 with the same scale as that of X2 by the obtained data;
s5: carrying out normalization processing and singular value decomposition processing on the original detection matrix D1 to construct a detection matrix D2; standardizing the D2 to construct a decision matrix D3; solving the eigenvalue of a 'decision matrix' D3, carrying out data visualization technical processing on the eigenvalue of the 'decision matrix' D3, and drawing an outer ring radius of 1 and an inner ring radius of (1-c) by adopting a single-ring theory L/2 The detection state visual reference database is obtained;
s6: through the normal state visual reference database constructed in the step S3, it is found that the eigenvalues of all normal "decision matrices" Z are between the inner ring and the outer ring of the single-ring theory, and the average radius of the eigenvalues of the normal "decision matrices" Z in the single-ring theory is greater than the radius of the inner ring; comparing the detection state visual reference database with the normal state visual reference database, and if the conditions of the detection state visual reference database and the normal state visual reference database are the same, the characteristic value points of D3 are all positioned between the inner ring and the outer ring of the circular ring, and the rolling bearing to be detected is normal, selecting to stop detection or detect the next group of data; if the detected state of the rolling bearing is found to be visually referred to the database, the characteristic value point D3 exists in the inner ring in the single-ring theory, and the average radius of the characteristic value D3 in the single-ring theory is smaller than the radius of the inner ring, the condition of the rolling bearing to be detected is indicated to be in fault.
Further, in step S1, a multi-point monitoring network is designed for the rolling bearing to be tested, including an output end and an input end, to perform all-dimensional operation data acquisition, and the acquired data is used to construct an operation state high-dimensional matrix
wherein x is ij The data measured at the moment j of the ith monitoring point of the rolling bearing are shown, i is the serial number of the monitoring point of the rolling bearing, and i is 1,2,3.. N; j is a detected time scale, and j is 1,2,3.. T;
in step S1, in data source X, the data under normal operating conditions constitute a normal-state reference database X1 in a high-dimensional matrix
wherein x is ij Representing data measured at the j-th moment of the ith monitoring point, wherein i is a serial number of the monitoring point of the rolling bearing, and is 1,2,3.. n; j is detectionJ-1, 2,3.
Further, in step S2, the spatio-temporal section data is first cut from the database X1 according to the specific time t and frequency f, then the feature extraction is performed on the spatio-temporal section data, a high-dimensional matrix is constructed by using the data after feature extraction, and is called as a normal original measurement matrix X2,
wherein x is ij Is data at any position in a high-dimensional original detection matrix obtained by reconstructing data obtained after feature extraction, x ij Representing data measured at the j-th moment of the ith monitoring point, wherein i is a serial number of the monitoring point of the rolling bearing, and is 1,2,3.. n; j is the time scale of detection, j 1,2,3.
Further, in step S3, the normal raw measurement matrix X2 is normalized by the following normalization formula; the normalization formula is as follows:
wherein, i is more than or equal to 1 and less than or equal to n is a matrix X 2 The line vectors of (a) are,
represents the average value, σ (X), of the normal raw measurement matrix X2i ij ) Representing the standard deviation of a normal raw measurement matrix X2i, and the new matrix formed is marked as matrix X 2 '；
Matrix X is decomposed by following singular value decomposition formula 2 Singular value decomposition processing is carried out to obtain a normal measurement matrix X3; the singular value decomposition formula is:
wherein the matrix U is a Haar unitary matrix, the matrix X 2 ' H Is a matrix X 2 ' arrangement ofThe matrix is obtained after the above processing
The normal measurement matrix X3 is normalized line by line using a matrix normalization process formula, where each line is normalized using the following normalization process formula:
wherein X 3i Refers to decision matrix X 3 Row i of (1), Z i The number is the ith row of a 'decision matrix' Z; a normal "decision matrix" Z is obtained,
solving the characteristic value of the 'decision matrix' Z as an evaluation index of the rolling bearing running state, and expressing the characteristic value on a complex plane by using a data visualization processing method; according to the single-ring theory, the radius of the outer ring is 1, the radius of the inner ring is (1-c) L/2 Wherein L is 1, c is n/t' e (0, 1).
Further, in the step S4, the method of constructing the raw detection matrix D1 is the same as the method of constructing the normal raw measurement matrix X2 in the step S2.
Further, in the step S5, the method of performing normalization processing and singular value decomposition processing on the original detection matrix D1 is the same as the method of performing normalization processing agent singular value decomposition processing on the normal original measurement matrix X2 in the step S3, and a detection matrix D2 is constructed; normalizing the D2 to construct a 'decision matrix' D3, wherein the method for normalizing the D2 in the step S5 is the same as the method for normalizing the normal measurement matrix X3 line by line in the step S3; solving the eigenvalue of a 'decision matrix' D3 and converting the eigenvalue of the 'decision matrix' D3Carrying out data visualization technical processing, adopting a single-ring theory to draw an outer ring radius of 1 and an inner ring radius of (1-c) L/2 And obtaining a visual reference database of the detection state.
Further, in the step S6, when some of the D3 characteristic value points in the to-be-detected state of the rolling bearing break through the inner ring of the single ring theory, but the average radius of the D3 characteristic value in the single ring theory is larger than the radius of the inner ring, it indicates that the state of the to-be-detected rolling bearing is abnormal, and the larger the number of the inner rings of which the D3 characteristic value points break through the single ring theory, the more serious the abnormal degree is, until the fault occurs.
The technical scheme of the invention has the following beneficial effects: designing a plurality of detection networks for a bearing to be detected, collecting operation data to form a data source, selecting a normal operation space-time section from the data source, constructing a normal database, constructing a normal original measurement matrix, constructing a normal judgment matrix, solving a characteristic value of the normal judgment matrix, and constructing a normal state visual parameter database; selecting unknown space-time sections from a data source, constructing an original measurement matrix, constructing a detection decision matrix, solving characteristic values of the detection decision matrix, and constructing a visual reference database of a detection state; comparing the visual reference database of the detection state with the visual parameter database of the normal state, and judging whether the visual reference database of the detection state is abnormal or not, so as to judge whether the bearing state is normal, abnormal or failure; if the time-space section is normal, the detection of other future time-space sections can be continued or stopped; if the fault occurs, the machine is stopped and the bearing is replaced; if the bearing is abnormal, the bearing is replaced according to the requirement, the bearing state is detected by establishing a comparison database, the running state of the bearing can be judged, and the measuring efficiency is high.
Drawings
FIG. 1 is a flow chart of a high-dimensional random matrix-based bearing condition detection method of the present invention;
FIG. 2 is a normal state visualization parameter database of the present invention;
FIG. 3 is an abnormal state visualization parameter database of the present invention;
FIG. 4 is a failure status visualization parameter database of the present invention.
Detailed Description
To make the technical problems, technical solutions and advantages of the present invention more apparent, the following detailed description is given with reference to the accompanying drawings and specific embodiments.
In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", "front", "rear", and the like indicate orientations or positional relationships based on those shown in the drawings, and are only for convenience of description and simplicity of description, but do not indicate or imply that the referred device or element must have a specific orientation, be constructed in a specific orientation, and be operated, and thus, should not be construed as limiting the present invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.
In the description of the present invention, it should be noted that, unless otherwise explicitly specified or limited, the terms "mounted," "connected," and "connected" are to be construed broadly, e.g., as meaning either a fixed connection, a removable connection, or an integral connection; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meanings of the above terms in the present invention can be understood in specific cases to those skilled in the art.
As shown in figure 1 of the drawings, in which,
a bearing state detection method based on a high-dimensional random matrix is characterized by comprising the following steps:
s1: under the condition of lacking fault state signals, a multi-point monitoring network is designed for the rolling bearing to be detected in work, in the working process, vibration signals of all monitoring points of the rolling bearing to be detected are collected, the collected real-time data are stored in a high-dimensional matrix form, namely the data source X, a normal operation state space-time section is selected from the data source X, and a normal state reference database X1 is constructed.
In step S1, a multi-point monitoring network is designed for the rolling bearing to be tested, including an output end and an input end, to perform all-directional acquisition of operation data, and the acquired data is used to construct a high-dimensional matrix of operation status
wherein x is ij The data measured at the moment j of the ith monitoring point of the rolling bearing is represented, i is the serial number of the monitoring point of the rolling bearing, and i is 1,2,3.. N; j is the time scale of detection, j 1,2,3.
In step S1, in data source X, the data under normal operating conditions constitute a normal-state reference database X1 in a high-dimensional matrix
wherein x is ij Representing data measured at the j-th moment of the ith monitoring point, wherein i is a serial number of the monitoring point of the rolling bearing, and is 1,2,3.. n; j is the time scale of detection, j 1,2,3.
S2: and (3) intercepting a space-time section from a normal state reference database X1 according to specific time t and frequency f, carrying out feature extraction on data in X1, and forming a high-dimensional matrix by the data after feature extraction, wherein the matrix is called a normal original measurement matrix X2.
In the step S2, the spatio-temporal profile data is first cut out from the database X1 according to the specific time t and frequency f, then feature extraction is performed on the spatio-temporal profile data, a high-dimensional matrix called as a normal original measurement matrix X2 is constructed by using the data after feature extraction,
wherein x is ij Is a high dimension obtained by reconstructing data obtained after feature extractionData of arbitrary position in the original detection matrix, x ij Representing data measured at the j-th moment of the ith monitoring point, wherein i is a serial number of the monitoring point of the rolling bearing, and is 1,2,3.. n; j is the time scale of detection, j 1,2,3.
S3: normalizing the normal original measurement matrix X2 to obtain a matrix X2 ', and then performing singular value decomposition on the matrix X2' to obtain a normal measurement matrix X3; carrying out standardization processing on the normal measurement matrix X3 to form a normal 'decision matrix' Z, and carrying out eigenvalue solution on the normal 'decision matrix' Z; representing the characteristic value of a normal 'decision matrix' Z on a complex plane by using a data visualization processing method; by adopting the single-ring theory, the radius of the outer ring is 1, the radius of the inner ring is (1-c) L/2 And (4) completing the construction of a normal state visual reference database.
In the step S3, a normal raw measurement matrix X2 is normalized by the following normalization formula; the normalization formula is:
wherein, i is more than or equal to 1 and less than or equal to n is a matrix X 2 The line vectors of (a) are,
represents the average value, σ (X), of the normal raw measurement matrix X2i ij ) Representing the standard deviation of a normal raw measurement matrix X2i, and the new matrix formed is marked as matrix X 2 '；
Matrix X is decomposed by following singular value decomposition formula 2 Singular value decomposition processing is carried out to obtain a normal measurement matrix X3; the singular value decomposition formula is:
wherein the matrix U is a Haar unitary matrix, the matrix X 2 ' H Is a matrix X 2 The transposed matrix of' is obtained after the above-mentioned processing
Haar matrix: one of the common random matrices is the Haar matrix, if a p orthogonal random matrix H p The distribution of p conforms to a Haar measure h p It is called a Haar matrix. For a P × P matrix a, if the distribution of matrix a conforms to a standard gaussian distribution, then matrix P ═ a (a' a) -1/2 And P ═ A (AA') -1/2 A are all Haar matrices.
The normal measurement matrix X3 is normalized line by line using a matrix normalization process formula, where each line is normalized using the following normalization process formula:
wherein, X 3i Refers to decision matrix X 3 Row i of (1), Z i The number is the ith row of a 'decision matrix' Z; a normal "decision matrix" Z is obtained,
solving the characteristic value of a 'decision matrix' Z as an evaluation index of the rolling bearing running state, and expressing the characteristic value on a complex plane by using a data visualization processing method; according to the single-ring theory, the radius of the outer ring is 1, the radius of the inner ring is (1-c) L/2 Wherein L is 1, c is n/t' e (0, 1).
S4: intercepting a space-time section with an unknown state from a data source X according to the same time t and frequency f, and forming an original detection matrix D1 with the same scale as that of X2 by the obtained data; the method of constructing the original detection matrix D1 is the same as the method of constructing the normal original measurement matrix X2 in step S2.
S5: carrying out normalization processing and singular value processing on an original detection matrix D1Performing solution processing to construct a detection matrix D2; d2 is standardized, and a 'decision matrix' D3 is constructed; solving the eigenvalue of a 'decision matrix' D3, carrying out data visualization technical processing on the eigenvalue of the 'decision matrix' D3, and drawing an outer ring radius of 1 and an inner ring radius of (1-c) by adopting a single-ring theory L/2 The detection state visual reference database is obtained; the method of performing normalization processing and singular value decomposition processing on the original detection matrix D1 is the same as the method of performing normalization processing agent singular value decomposition processing on the normal original measurement matrix X2 in step S3, and a detection matrix D2 is constructed; normalizing the D2 to construct a 'decision matrix' D3, wherein the method for normalizing the D2 in the step S5 is the same as the method for normalizing the normal measurement matrix X3 line by line in the step S3; solving the eigenvalue of a 'decision matrix' D3, carrying out data visualization technical processing on the eigenvalue of the 'decision matrix' D3, and drawing an outer ring radius of 1 and an inner ring radius of (1-c) by adopting a single-ring theory L/2 And obtaining a visual reference database of the detection state. Single-ring theory: the Single Ring theory (Single Ring theory) is a data processing method for processing Non-Hermitian matrices (Non-Hermitian Matrix) proposed in 2009 by guionet, Krishnapur and Manjunath et al.
S6: through the normal state visual reference database constructed in the step S3, it is found that the eigenvalues of all the normal "decision matrices" Z in the database are located between the inner ring and the outer ring of the single-ring theory, and the average radius of the eigenvalues of the normal "decision matrices" Z in the single-ring theory is greater than the radius of the inner ring; comparing the detection state visual reference database with the normal state visual reference database, and if the conditions of the detection state visual reference database and the normal state visual reference database are the same, the characteristic value points of D3 are all positioned between the inner ring and the outer ring of the circular ring, and the rolling bearing to be detected is normal, stopping detection or detecting the next group of data can be selected; if the detected state of the rolling bearing is found to be visually referred to the database, the characteristic value point D3 exists in the inner ring in the single-ring theory, and the average radius of the characteristic value D3 in the single-ring theory is smaller than the radius of the inner ring, the condition of the rolling bearing to be detected is indicated to be in fault.
In the step S6, when some of the D3 characteristic value points in the to-be-tested state of the rolling bearing break through the inner ring of the single ring theory, but the average radius of the D3 characteristic value in the single ring theory is larger than the radius of the inner ring, it indicates that the state of the to-be-tested rolling bearing is abnormal, and the larger the number of the inner rings of which the D3 characteristic value points break through the single ring theory, the more serious the abnormal degree is, until the fault occurs.
In the embodiment, for a brand-new complex mechanical device, the rolling bearing lacks fault data, and the collected data is data of the rolling bearing during production and operation, so that the rolling bearing cannot be modeled according to a traditional method. A multipoint monitoring network is designed for a bearing to be detected, the multipoint monitoring network comprises an input end, a fan end and the like, operation data are collected in all directions, data in the operation of a rolling bearing are collected, and the collected data are subjected to matrix construction, so that a data source X of equipment is formed.
As shown in fig. 1, firstly, a normal state reference structure is used, normal operation spatio-temporal profile data is selected from X to construct a normal operation reference database X1, feature extraction is performed on the extracted data, the selected features include a series of indexes such as mean, variance, kurtosis and skewness, the data obtained by feature extraction is constructed into a 400 × 500 high-dimensional matrix, that is, n/t' is 0.8 and is called a normal measurement matrix X2, so that the high-dimensional matrix is selected, that is, information in the measurement data is mined as completely as possible, and information loss caused by data simplification is avoided.
The normal operation data form a normal original measurement matrix, the normal original measurement matrix is subjected to unitization by using a normalization formula, then singular value decomposition is performed by using a singular value decomposition formula to obtain a normal measurement matrix X3, and the normal measurement matrix is subjected to line-by-line unitization by using a normalization processing formula to obtain a normal 'decision matrix' Z of 400X 400.
Is aligned withSolving eigenvalue of the normal state 'decision matrix' Z to obtain 400 eigenvalues, representing the eigenvalues on a complex plane, and combining a single-ring theory (the outer diameter of the single-ring theory is 1, and the inner diameter is (1-c)) L/2 Since the simplest single-ring theory is constructed, L1, c n/t e (0,1), the inner diameter is (1/5) 1/2 ) And obtaining a normal state visual reference database, as shown in fig. 2.
For the detection sample state structure in fig. 1, the data processing process is the same as before, except that the measured data is in an abnormal state, the high-dimensional matrix scale of the abnormal data is still maintained at 400 × 500, then normalization processing, singular value decomposition processing and normalization processing are performed according to the processing process to obtain a "decision matrix" D3, the characteristic value of D3 is solved to obtain a detection state visualization reference database, as shown in fig. 3 and 4, and comparison between fig. 3 and 4 and fig. 2 shows that a significant difference can be found. Under a normal state, D3 eigenvalue data points drawn according to the eigenvalue of the 'decision matrix' D3 are converged between the inner ring and the outer ring of the circular ring; when only a few points are breakthrough inner rings or outer rings, the proportion of the data points in breakthrough inner rings or outer rings and the severity of breakthrough are used for measuring the state of the equipment, and the bearing is considered to be abnormal, as shown in fig. 3; when a plurality of points break through the inner ring or the outer ring and the average radius of the characteristic value data points of D3 is smaller than the radius of the inner ring, the rolling bearing is considered to have a fault or be in a failure state, the rolling bearing needs to be stopped for equipment inspection, and the bearing is disassembled and replaced, as shown in FIG. 4; if the bearing is required to be ensured not to break down, when the bearing is detected to have an abnormal state, the machine is stopped to replace the bearing.
While the foregoing is directed to the preferred embodiment of the present invention, it will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the invention as defined in the appended claims.
Claims (7)
1. A bearing state detection method based on a high-dimensional random matrix is characterized by comprising the following steps:
s1: under the condition of lacking fault state signals, designing a multi-point monitoring network for a rolling bearing to be detected in work, collecting vibration signals of monitoring points of the rolling bearing to be detected in the work process, storing the collected real-time data in a high-dimensional matrix form to obtain a data source X, selecting a normal operation state space-time section from the data source X, and constructing a normal state reference database X1;
s2: intercepting a space-time section from a normal state reference database X1 according to specific time t and frequency f, and performing feature extraction on data in X1, wherein the data after feature extraction form a high-dimensional matrix which is called a normal original measurement matrix X2;
s3: normalizing the normal original measurement matrix X2 to obtain a matrix X2 ', and then performing singular value decomposition on the matrix X2' to obtain a normal measurement matrix X3; carrying out standardization processing on the normal measurement matrix X3 to form a normal 'decision matrix' Z, and carrying out eigenvalue solution on the normal 'decision matrix' Z; representing the characteristic value of a normal 'decision matrix' Z on a complex plane by using a data visualization processing method; by adopting the single-ring theory, the radius of the outer ring is 1, the radius of the inner ring is (1-c) L/2 The ring completes the construction of a normal state visual reference database; wherein, L is 1, c is n/t' epsilon (0, 1);
s4: intercepting a space-time section with an unknown state from a data source X according to the time t and the frequency f which are the same as those in the step 2, and forming an original detection matrix D1 with the same scale as that of X2 by the obtained data;
s5: carrying out normalization processing and singular value decomposition processing on the original detection matrix D1 to construct a detection matrix D2; d2 is standardized, and a 'decision matrix' D3 is constructed; solving the eigenvalue of a 'decision matrix' D3, carrying out data visualization technical processing on the eigenvalue of the 'decision matrix' D3, and drawing an outer ring radius of 1 and an inner ring radius of (1-c) by adopting a single-ring theory L/2 The detection state visual reference database is obtained;
s6: through the normal state visual reference database constructed in the step S3, it is found that all eigenvalues of the constructed normal "decision matrix" Z are between the inner ring and the outer ring of the single-ring theory, and the average radius of the eigenvalues of the normal "decision matrix" Z in the single-ring theory is greater than the radius of the inner ring; comparing the detection state visual reference database with the normal state visual reference database, and if the conditions of the detection state visual reference database and the normal state visual reference database are the same, the characteristic value points of D3 are all positioned between the inner ring and the outer ring of the circular ring, and the rolling bearing to be detected is normal, stopping detection or detecting the next group of data can be selected; if the detected state of the rolling bearing is found to be visually referred to the database, the characteristic value point D3 exists in the inner ring in the single-ring theory, and the average radius of the characteristic value D3 in the single-ring theory is smaller than the radius of the inner ring, the condition of the rolling bearing to be detected is indicated to be in fault.
2. The method as claimed in claim 1, wherein in step S1, a multi-point monitoring network is designed for the rolling bearing to be tested, the network includes an output terminal and an input terminal, the operation data is collected in all directions, and the collected data is used to construct the operation state high-dimensional matrix
wherein x is ij The data measured at the moment j of the ith monitoring point of the rolling bearing are shown, i is the serial number of the monitoring point of the rolling bearing, and i is 1,2,3.. N; j is a detected time scale, and j is 1,2,3.. T;
in step S1, in data source X, the data under normal operating conditions constitute a normal-state reference database X1 in a high-dimensional matrix
wherein x is ij Representing the number measured at time j of the ith monitoring pointAccording to the formula, i is the serial number of a rolling bearing monitoring point, and i is 1,2,3.. n; j is the time scale of detection, j 1,2,3.
3. The method as claimed in claim 1, wherein in step S2, the spatio-temporal profile data is first extracted from the database X1 according to a specific time t and frequency f, and then feature extraction is performed on the spatio-temporal profile data, and a high-dimensional matrix is constructed as a normal original measurement matrix X2 using the data after feature extraction,
wherein x is ij Is data at any position in a high-dimensional original detection matrix obtained by reconstructing data obtained after feature extraction, x ij Representing data measured at the jth moment of the ith monitoring point, wherein i is a monitoring point serial number of a rolling bearing high-dimensional matrix X2, and i is 1,2,3.. n; j is the time scale of detection, j is 1,2,3.. t ', and n and t' are the row and column numbers of the normal raw measurement matrix X2, respectively.
4. The method for detecting the condition of the bearing based on the high-dimensional random matrix as claimed in claim 1, wherein in step S3, the normal raw measurement matrix X2 is normalized by the following normalization formula; the normalization formula is as follows:
wherein x is ij The data at any position in a high-dimensional original detection matrix is obtained by reconstructing the data obtained after the characteristic extraction; t' is the normal raw measurement matrix X 2 The number of columns; i is more than or equal to 1 and less than or equal to n is a matrix X 2 The line vectors of (a) are,
represents the normal raw measurement matrix X2 No i Mean value of the rows, σ (x) ij ) Represents the normal raw measurement matrix X2 th i The standard deviation of the rows, the new matrix formed is marked as matrix X 2 '；
Matrix X is decomposed by following singular value decomposition formula 2 Singular value decomposition processing is carried out to obtain a normal measurement matrix X3; the singular value decomposition formula is:
wherein the matrix U is a Haar unitary matrix, the matrix X 2 ' H Is a matrix X 2 The transposed matrix of' is obtained after the above-mentioned processing
The normal measurement matrix X3 is normalized line by line using a matrix normalization process formula, where each line is normalized using the following normalization process formula:
wherein X 3i Refers to decision matrix X 3 Row i of (1), Z i The number is the ith row of a 'decision matrix' Z; a normal "decision matrix" Z is obtained,
solving the characteristic value of the 'decision matrix' Z as an evaluation index of the rolling bearing running state, and expressing the characteristic value on a complex plane by using a data visualization processing method; according to the single-ring theory, the radius of the outer ring is 1, the radius of the inner ring is (1-c) L/2 Of a ring ofIn the formula, L is 1, and c is n/t ∈ (0, 1).
5. The method for detecting the condition of the bearing based on the high-dimensional random matrix as claimed in claim 1, wherein in the step S4, the method for constructing the original detection matrix D1 is the same as the method for constructing the normal original measurement matrix X2 in the step S2.
6. The method for detecting the condition of the bearing based on the high-dimensional random matrix as claimed in claim 1, wherein in step S5, the method for performing the normalization processing and singular value decomposition processing on the original detection matrix D1 is the same as the method for performing the normalization processing agent singular value decomposition processing on the normal original measurement matrix X2 in step S3, so as to construct the detection matrix D2; normalizing the D2 to construct a 'decision matrix' D3, wherein the method for normalizing the D2 in the step S5 is the same as the method for normalizing the normal measurement matrix X3 line by line in the step S3; solving the eigenvalue of a 'decision matrix' D3, carrying out data visualization technical processing on the eigenvalue of the 'decision matrix' D3, and drawing an outer ring radius of 1 and an inner ring radius of (1-c) by adopting a single-ring theory L/2 And obtaining a visual reference database of the detection state.
7. The method as claimed in claim 1, wherein in step S6, when some of the D3 eigenvalue points in the to-be-tested state of the rolling bearing break through the inner ring of the single-ring theory, but the average radius of the D3 eigenvalue in the single-ring theory is greater than the radius of the inner ring, it indicates that the state of the rolling bearing under test is abnormal, and the larger the number of the inner rings of the single-ring theory that the D3 eigenvalue points break through, the more serious the abnormal degree is, until the fault occurs.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810816943.XA CN109033031B (en) | 2018-07-24 | 2018-07-24 | Bearing state detection method based on high-dimensional random matrix |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810816943.XA CN109033031B (en) | 2018-07-24 | 2018-07-24 | Bearing state detection method based on high-dimensional random matrix |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109033031A CN109033031A (en) | 2018-12-18 |
| CN109033031B true CN109033031B (en) | 2022-08-23 |
Family
ID=64644621
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810816943.XA Active CN109033031B (en) | 2018-07-24 | 2018-07-24 | Bearing state detection method based on high-dimensional random matrix |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109033031B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110704987B (en) * | 2019-10-21 | 2023-08-22 | 南通大学 | Bearing abnormal state assessment method based on similar working condition of failure data mining |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107132454A (en) * | 2017-05-04 | 2017-09-05 | 国网上海市电力公司 | The abnormal quick determination method of power network based on random matrix spectral radius method |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105258947B (en) * | 2015-11-06 | 2017-06-23 | 北京航空航天大学 | A kind of Fault Diagnosis of Roller Bearings under operating mode disturbed conditions based on compressed sensing |
| CN106548021B (en) * | 2016-11-03 | 2018-01-05 | 华中科技大学 | A kind of electronic mutual inductor error state appraisal procedure based on higher-dimension random matrix |
| CN108173670B (en) * | 2016-12-07 | 2020-06-02 | 华为技术有限公司 | Method and device for detecting network |
| CN107024352A (en) * | 2017-05-03 | 2017-08-08 | 哈尔滨理工大学 | A kind of Rolling Bearing Fault Character extracting method based on slip entropy ICA algorithm |
-
2018
- 2018-07-24 CN CN201810816943.XA patent/CN109033031B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107132454A (en) * | 2017-05-04 | 2017-09-05 | 国网上海市电力公司 | The abnormal quick determination method of power network based on random matrix spectral radius method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109033031A (en) | 2018-12-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106780121B (en) | Power consumption abnormity identification method based on power consumption load mode analysis | |
| CN112505549B (en) | New energy automobile battery abnormity detection method based on isolated forest algorithm | |
| CN108445410A (en) | A kind of method and device of monitoring accumulator group operating status | |
| CN111579121B (en) | Method for diagnosing faults of temperature sensor in new energy automobile battery pack on line | |
| CN115469176A (en) | Digital twin power grid risk assessment system based on data model | |
| CN116756595B (en) | Conductive slip ring fault data acquisition and monitoring method | |
| CN113542017A (en) | Network fault positioning method based on network topology and multiple indexes | |
| CN116678552B (en) | Abnormality monitoring method for optical fiber stress sensor in variable temperature environment | |
| CN110704987A (en) | Bearing abnormal state evaluation method based on similar working conditions of failure data mining | |
| CN115267589B (en) | Multi-parameter joint diagnosis method for battery faults of electric vehicle | |
| CN110930057A (en) | Quantitative evaluation method for reliability of distribution transformer test result based on LOF algorithm | |
| CN112417763A (en) | Defect diagnosis method, device and equipment for power transmission line and storage medium | |
| CN109033031B (en) | Bearing state detection method based on high-dimensional random matrix | |
| CN115037603A (en) | Diagnosis evaluation method, device and system of electricity consumption information acquisition equipment | |
| CN114610553A (en) | Intelligent equipment detection system and detection method thereof | |
| CN114487705A (en) | Power grid equipment fault positioning detection method | |
| CN113837591A (en) | Equipment health assessment method oriented to multi-working-condition operation conditions | |
| CN117113135A (en) | Carbon emission anomaly monitoring and analyzing system capable of sorting and classifying anomaly data | |
| WO2023029382A1 (en) | Strong-robustness signal early-degradation feature extraction and device running state monitoring method | |
| CN115788771A (en) | Intelligent operation and maintenance management system of wind power generation system based on Internet of things technology | |
| CN116032003A (en) | Power grid inspection method and system based on power private network data processing | |
| CN113032915B (en) | Method and device for evaluating health state of spindle bearing of manufacturing equipment | |
| CN112532481B (en) | Network quality monitoring and analyzing method, system and storage medium | |
| CN114997375A (en) | Insulator surface metal pollution model building system in air type switch cabinet | |
| CN114760332A (en) | Efficient equipment monitoring and analyzing 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 |