Bearing fault detection method based on Gershgorin circle theorem

Abstract

The present invention discloses a bearing fault detection method based on the Gershgorin circle theorem, which relates to the technical field of fault diagnosis. The technical solution identifies abnormal eigenvalues ​​T at an early stage and issues early warnings by establishing an autocorrelation matrix and calculating the eigenvalue T distribution. Secondly, the technical solution overcomes the dependence of traditional methods on high sampling rates, and uses fast Fourier transform and noise filtering technology to simplify data processing and improve system efficiency. Finally, by calculating and evaluating the health index Rhzs, the dependence on experience is reduced, making the detection process more automated and data-driven. The evaluation results are based on historical data and health threshold classification, which improves the accuracy and reliability of fault detection. Overall, the solution optimizes the effect of bearing fault detection through precise data processing and automated analysis.

Description

Technical Field

[0001] The invention relates to the technical field of fault diagnosis, and in particular to a bearing fault detection method based on the Gershgorin circle theorem. Background Art

[0002] Aircraft engine bearings are key rotating components, and their performance directly affects the reliability and safety of the engine. However, bearings are prone to wear, fatigue and other faults during long-term high-load operation, which can lead to serious accidents. Therefore, early detection of bearing failures is particularly important.

[0003] Traditional detection methods include vibration analysis, temperature monitoring and acoustic emission technology, but these methods have the following technical disadvantages:

[0004] 1. Early faults are difficult to detect: Traditional methods are slow to detect minor or early faults, and usually require the fault to develop to a certain extent before it can be easily detected, which may lead to untimely warnings;

[0005] 2. High sampling rate required: Many vibration analysis methods require data with high sampling rate, which results in large data processing volume, high hardware requirements, and limited real-time performance.

[0006] 3. Reliance on experience: The selection of parameters and interpretation of analysis results of traditional methods often rely on experience and are highly subjective. Different operators may come to different conclusions. Summary of the invention

[0007] Technical issues solved

[0008] In view of the deficiencies in the prior art, the present invention provides a bearing fault detection method based on the Gershgorin circle theorem, which solves the technical shortcomings of the background technology that initial faults are difficult to detect, require a high sampling rate and rely on experience.

[0009] Technical Solution

[0010] To achieve the above objectives, the present invention is implemented through the following technical solutions: A bearing fault detection method based on the Gershgorin circle theorem includes the following steps:

[0011] S1. Install the bearing to be tested on the test bench in advance, use the multi-channel sensor system to monitor the bearing in real time, and collect the vibration signal data of the bearing during operation; upload the vibration signal data of the healthy bearing as standard data to the reference database;

[0012] S2. Then, the vibration signal data collected from the bearing to be inspected is input into the signal processing module; the original signal is decomposed into different frequency components by using the fast Fourier transform signal processing technology, and the background noise is removed to obtain clear characteristic signal data;

[0013] S3, then, the processed characteristic signal data is used to construct an autocorrelation matrix or a covariance matrix; the eigenvalue distribution of the matrix is ​​calculated using the Gershgorin circle theorem, and a Gershgorin disk is constructed according to the position of the eigenvalue to determine whether there is an abnormal eigenvalue distribution;

[0014] S4. If an abnormal characteristic value distribution is detected, the system will issue a warning signal and further analyze the health status of the current bearing by comparing the Gershgorin disk of the standard bearing in the reference database; for the abnormal characteristic value distribution found, the corresponding fault characteristic factor is calculated, and the bearing fault coefficient Bxgs is constructed based on the change trend of the characteristic factor;

[0015] S5. Finally, the bearing fault coefficient Bxgs is correlated with historical data using machine learning technology, and the bearing health index Rhzs is fitted after dimensionless processing. The health index Rhzs is compared with the preset health threshold H to comprehensively evaluate the current bearing fault level, and trigger maintenance or replacement operations based on the evaluation results.

[0016] Preferably,

[0017] S11. Pre-install the bearing to be tested on the test bench, and deploy a multi-channel sensor system according to the installation position of the bearing to be tested, monitor the vibration signal of the bearing to be tested in real time through the multi-channel sensor system, collect and mark the vibration signal data of different channels, and associate and mark these signal data with corresponding channel information;

[0018] Preferably,

[0019] S21, inputting the vibration signal data of the bearing to be inspected into the signal processing module, applying the fast Fourier transform technology to decompose the input original vibration signal in the frequency domain, obtaining signal data of different frequency components, and storing the data of these frequency components in sequence;

[0020] S22, processing the signal data of different frequency components that have been decomposed, applying a noise filtering algorithm, removing background noise components, retaining valid characteristic signal data, and transmitting the denoised characteristic signal data as output to a subsequent analysis module.

[0021] Preferably,

[0022] S31. Input the processed characteristic signal data into the data processing system, and construct an autocorrelation matrix related to the vibration signal of the bearing to be inspected through an autocorrelation analysis method, wherein each element of the matrix represents the signal correlation of the bearing to be inspected at different time points.

[0023] Preferably,

[0024] S32. From the constructed autocorrelation matrix, by collecting all diagonal elements Ai and off-diagonal elements Aj, using the Gershgorin circle theorem calculation formula, the center C and radius R of the matrix are obtained, and finally the center and radius of all Gershgorin disks are calculated row by row, and their distribution on the complex plane is inferred, so as to obtain the eigenvalue T of the matrix. The specific calculation process is as follows:

[0025] Given an n×n square matrix A, whose elements are aij (elements in the i-th row and j-th column):

[0026]

[0027] Then obtain the eigenvalue T of matrix A by solving the following characteristic equation:

[0028] det(A-TI)=0;

[0029] Where I is the n×n identity matrix, T is the eigenvalue to be found, and det represents the determinant;

[0030] Then solve the characteristic equation, calculate the determinant of the matrix and expand it, and finally get n eigenvalues ​​and plot them on the complex plane. Compare the position of each eigenvalue to check whether it falls within the corresponding Gershgorin disk, and determine whether there is an abnormal eigenvalue distribution in the current matrix. Specifically:

[0031] Normality: If all eigenvalues ​​fall inside the corresponding Gershgorin disk, then the eigenvalues ​​T of the matrix are normally distributed.

[0032] Abnormal situation: If one or more eigenvalues ​​T fall outside the disk, it means that the distribution of the eigenvalues ​​T of the matrix is ​​abnormal, indicating that the matrix properties have changed.

[0033] Preferably,

[0034] S41. After detecting abnormal distribution of characteristic values, an early warning signal is issued, and the health status of the current bearing is analyzed by comparing it with the Gershgorin disk of the standard bearing in the reference database. The specific process is as follows:

[0035] S411. Comparative analysis: compare the characteristic value distribution of the current bearing with the Gershgorin disk of the standard bearing to determine whether the current characteristic value exceeds the normal range.

[0036] S412. Issue an early warning: When an abnormal distribution is detected, the system generates and issues an early warning signal to prompt the operator or maintenance system to conduct further inspection.

[0037] Preferably,

[0038] S42. According to the distribution of abnormal characteristic values ​​found, the corresponding fault characteristic factors are calculated, and the bearing fault coefficient Bxgs is constructed based on the change trend of the characteristic factors. The specific contents are as follows:

[0039] Extract relevant fault characteristic factors, including abnormal frequency component Yc, deviation from normal value Pl, eigenvalue offset Tpy and matrix condition number Jzt. After dimensionless processing of the extracted relevant fault characteristic factors, calculate the bearing fault coefficient Bxgs by the following formula:

[0040]

[0041] Preferably,

[0042] S51. Extract the records and related parameters of the bearing fault coefficient Bxgs from the historical data, and perform dimensionless processing on the bearing fault coefficient Bxgs and related parameters, where the related parameters include the radial clearance Jjx between the inner and outer rings of the bearing, the viscosity change Nbh of the lubricating oil, the kurtosis Zxh of the vibration signal, and the thermal conductivity Rdl; obtain the bearing health index Rhzs by calculating the following formula:

[0043]

[0044] Preferably,

[0045] S52. Compare and evaluate the preset health threshold H with the health index Rhzs. The specific contents are as follows:

[0046] If the health index Rhzs> the preset health threshold H, it means that the current bearing health index is abnormal and lower than the preset threshold; at this time, the health status of the bearing is marked as the first status result;

[0047] If the health index Rhzs = the preset health threshold, it means that the current bearing health index is normal and just above the preset threshold. At this time, the health status of the bearing is marked as the second status result;

[0048] If the health index Rhzs<the preset health threshold, it means that the current bearing health index is normal and higher than the preset threshold. At this time, the health status of the bearing is marked as the third status result.

[0049] Preferably,

[0050] S53: Trigger maintenance or replacement operations according to the evaluation results. The specific contents are as follows:

[0051] If the health status is the first status result: mark the current bearing as an object that needs maintenance or replacement, arrange for professionals to conduct further inspection, determine the specific cause of the failure, and record relevant information for subsequent processing;

[0052] If the health status is the second status result: mark the current bearing as an object that needs further observation. Arrange inspectors to conduct a detailed inspection of the bearing to evaluate whether its condition can be improved through adjustment and repair;

[0053] If the health status is the third status result: continue to use the bearing and proceed to the next working step.

[0054] Beneficial Effects

[0055] The present invention provides a bearing fault detection method based on the Gershgorin circle theorem. It has the following beneficial effects:

[0056] (1) This bearing fault detection method based on the Gershgorin circle theorem effectively solves the problem of difficulty in initial fault detection by comprehensively utilizing vibration signal data and the Gershgorin circle theorem. Traditional detection methods often have difficulty in discovering potential problems at the beginning of a fault, while this method identifies abnormal eigenvalues ​​at an early stage by establishing an autocorrelation matrix and calculating the eigenvalue distribution, thereby providing timely warnings. By comparing the Gershgorin disks of healthy bearings and those of the bearing to be inspected, the system can accurately determine whether there is a fault, providing a scientific basis for early fault diagnosis.

[0057] (2) This bearing fault detection method based on the Gershgorin circle theorem, in addition, this scheme overcomes the problem of dependence on high sampling rate; traditional methods often require high-frequency data sampling to improve detection accuracy, which not only increases the complexity and cost of the system, but may also lead to data processing bottlenecks; by using fast Fourier transform technology, frequency domain information is quickly acquired, and combined with noise filtering algorithm to process signal data, the data acquisition and processing process is greatly simplified, so that while ensuring detection accuracy, the dependence on high sampling rate is reduced, and the overall performance of the system is optimized.

[0058] (3) The bearing fault detection method based on the Gershgorin circle theorem, finally, this technical solution reduces the dependence on experience; traditional bearing fault detection often needs to rely on experienced technicians to analyze and judge the data, while this solution uses machine learning technology to perform dimensionless processing on the bearing fault coefficient Bxgs and calculate the bearing health index Rhzs, making the detection and evaluation process more automated and data-driven; the evaluation results are not only based on historical data for accurate predictions, but also can be classified by setting health thresholds, thereby effectively reducing the subjectivity of human judgment and improving the reliability and consistency of fault detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1It is a schematic diagram of the steps of the bearing fault detection method based on the Gershgorin circle theorem of the present invention; DETAILED DESCRIPTION

[0060] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0061] Example 1

[0062] See also Figure 1 The present invention provides a bearing fault detection method based on the Gershgorin circle theorem, comprising the following steps:

[0063] S1. Install the bearing to be tested on the test bench in advance, use the multi-channel sensor system to monitor the bearing in real time, and collect the vibration signal data of the bearing during operation; upload the vibration signal data of the healthy bearing as standard data to the reference database;

[0064] S2. Then, the vibration signal data collected from the bearing to be inspected is input into the signal processing module; the original signal is decomposed into different frequency components by using the fast Fourier transform signal processing technology, and the background noise is removed to obtain clear characteristic signal data;

[0065] S3, then, the processed characteristic signal data is used to construct an autocorrelation matrix or a covariance matrix; the eigenvalue distribution of the matrix is ​​calculated using the Gershgorin circle theorem, and a Gershgorin disk is constructed according to the position of the eigenvalue to determine whether there is an abnormal eigenvalue distribution;

[0066] S4. If an abnormal characteristic value distribution is detected, the system will issue a warning signal and further analyze the health status of the current bearing by comparing the Gershgorin disk of the standard bearing in the reference database; for the abnormal characteristic value distribution found, the corresponding fault characteristic factor is calculated, and the bearing fault coefficient Bxgs is constructed based on the change trend of the characteristic factor;

[0067] S5. Finally, the bearing fault coefficient Bxgs is correlated with historical data using machine learning technology, and the bearing health index Rhzs is fitted after dimensionless processing. The health index Rhzs is compared with the preset health threshold H to comprehensively evaluate the current bearing fault level, and trigger maintenance or replacement operations based on the evaluation results.

[0068] In this embodiment, firstly, a multi-channel sensor system is used to monitor the bearing in real time and collect vibration signals to ensure the comprehensiveness and accuracy of the data; secondly, the application of fast Fourier transform technology makes signal processing more efficient, and the clear characteristic signal data obtained after removing the background noise improves the accuracy of the analysis; thirdly, by constructing an autocorrelation matrix and applying the Gershgorin circle theorem to analyze the distribution of eigenvalues, it is possible to accurately determine whether there is an abnormal eigenvalue distribution, issue a warning signal in time and conduct further health status analysis; finally, by combining machine learning technology to perform historical data correlation analysis on the bearing fault coefficient, the generated health index is compared with the preset threshold value to accurately determine the fault level, and trigger maintenance or replacement operations based on the evaluation results, thereby realizing automated and precise fault detection and management. These improvements effectively overcome the shortcomings of traditional detection methods in initial fault detection, high sampling rate dependence and experience judgment, and improve the accuracy of fault diagnosis and system efficiency.

[0069] Example 2

[0070] S11. Pre-install the bearing to be tested on the test bench, and deploy a multi-channel sensor system according to the installation position of the bearing to be tested, monitor the vibration signal of the bearing to be tested in real time through the multi-channel sensor system, collect and mark the vibration signal data of different channels, and associate and mark these signal data with corresponding channel information;

[0071] In this embodiment

[0072] Example 3

[0073] S21, inputting the vibration signal data of the bearing to be inspected into the signal processing module, applying the fast Fourier transform technology to decompose the input original vibration signal in the frequency domain, obtaining signal data of different frequency components, and storing the data of these frequency components in sequence;

[0074] S22, processing the signal data of different frequency components that have been decomposed, applying a noise filtering algorithm, removing background noise components, retaining valid characteristic signal data, and transmitting the denoised characteristic signal data as output to a subsequent analysis module.

[0075] In this embodiment, by pre-installing the bearing to be inspected on the test bench and deploying a multi-channel sensor system, comprehensive real-time monitoring and accurate marking of the bearing vibration signal are achieved; this method can not only collect the vibration signal data of each channel, but also ensure that these data are effectively associated and labeled with the corresponding channel information, thereby improving the accuracy and completeness of the data; in addition, by selecting healthy bearings for vibration signal collection and labeling, and then transmitting these standard data to the reference database, reliable benchmark data is provided for subsequent fault detection; this approach significantly enhances the detection accuracy of the system and the effectiveness of comparative analysis, overcomes the problem of lack of unified marking and reference for standard data in traditional methods, and makes fault detection more scientific and systematic.

[0076] Example 4

[0077] S31. Input the processed characteristic signal data into the data processing system, and construct an autocorrelation matrix related to the vibration signal of the bearing to be inspected through an autocorrelation analysis method, wherein each element of the matrix represents the signal correlation of the bearing to be inspected at different time points.

[0078] S32. From the constructed autocorrelation matrix, by collecting all diagonal elements Ai and off-diagonal elements Aj, using the Gershgorin circle theorem calculation formula, the center C and radius R of the matrix are obtained, and finally the center and radius of all Gershgorin disks are calculated row by row, and their distribution on the complex plane is inferred, so as to obtain the eigenvalue T of the matrix. The specific calculation process is as follows:

[0079] Given an n×n square matrix A, whose elements are aij (elements in the i-th row and j-th column):

[0080]

[0081] Then obtain the eigenvalue T of matrix A by solving the following characteristic equation:

[0082] det(A-TI)=0;

[0083] Where I is the n×n identity matrix, T is the eigenvalue to be found, and det represents the determinant;

[0084] Then solve the characteristic equation, calculate the determinant of the matrix and expand it, and finally get n eigenvalues ​​and plot them on the complex plane. Compare the position of each eigenvalue to check whether it falls within the corresponding Gershgorin disk, and determine whether there is an abnormal eigenvalue distribution in the current matrix. Specifically:

[0085] Normality: If all eigenvalues ​​fall inside the corresponding Gershgorin disk, then the eigenvalues ​​T of the matrix are normally distributed.

[0086] Abnormal situation: If one or more eigenvalues ​​T fall outside the disk, it means that the distribution of the eigenvalues ​​T of the matrix is ​​abnormal, indicating that the matrix properties have changed.

[0087] In this embodiment, the processed characteristic signal data is input into the data processing system and the autocorrelation matrix is ​​constructed, thereby realizing the systematic analysis of the vibration signal of the bearing to be inspected; by extracting the diagonal and non-diagonal elements in the autocorrelation matrix, the center and radius of the matrix are accurately calculated using the Gershgorin circle theorem, thereby accurately inferring the distribution of the matrix eigenvalues ​​on the complex plane; this method not only ensures the accuracy of the eigenvalue calculation, but also can effectively identify the abnormal distribution in the matrix by comparing whether the eigenvalue falls within the Gershgorin disk, thereby revealing possible faults or abnormal states of the system; compared with the traditional method, this improvement enhances the sensitivity and accuracy of fault detection, while avoiding the potential risks caused by the failure to timely discover the abnormal distribution of eigenvalues, thereby improving the reliability and practicality of the overall fault detection system.

[0088] Example 5

[0089] S41. After detecting abnormal distribution of characteristic values, an early warning signal is issued, and the health status of the current bearing is analyzed by comparing it with the Gershgorin disk of the standard bearing in the reference database. The specific process is as follows:

[0090] S411. Comparative analysis: compare the characteristic value distribution of the current bearing with the Gershgorin disk of the standard bearing to determine whether the current characteristic value exceeds the normal range.

[0091] S412. Issue an early warning: When an abnormal distribution is detected, the system generates and issues an early warning signal to prompt the operator or maintenance system to conduct further inspection.

[0092] S42. According to the distribution of abnormal characteristic values ​​found, the corresponding fault characteristic factors are calculated, and the bearing fault coefficient Bxgs is constructed based on the change trend of the characteristic factors. The specific contents are as follows:

[0093] Extract relevant fault characteristic factors, including abnormal frequency component Yc, deviation from normal value Pl, eigenvalue offset Tpy and matrix condition number Jzt. After dimensionless processing of the extracted relevant fault characteristic factors, calculate the bearing fault coefficient Bxgs by the following formula:

[0094]

[0095] In this embodiment, after detecting an abnormal eigenvalue distribution, the present invention effectively improves the response speed and accuracy of the system by issuing an early warning signal and comparing it with the Gershgorin disk of a standard bearing; by comparing and analyzing whether the eigenvalue of the current bearing exceeds the normal range, the abnormal distribution can be accurately identified, so that an early warning signal is issued in time to remind the operator or the maintenance system to conduct further inspection, thereby effectively preventing the expansion of potential faults; further, based on the abnormal eigenvalue distribution, the fault characteristic factors are calculated, including the abnormal frequency component Yc, the deviation from the normal value Pl, the eigenvalue offset Tpy and the matrix condition number Jzt, and the bearing fault coefficient Bxgs is calculated after dimensionless processing, which can comprehensively evaluate the severity and development trend of the fault; these parameters not only provide detailed fault information, but also make the detection results more scientific and systematic, thereby improving the reliability and accuracy of the overall detection, and effectively reducing the possibility of false detection and missed detection;

[0096] Among them, the abnormal frequency component Yc refers to the specific frequency component in the vibration signal, whose value deviates from the normal operating range; by analyzing these frequency components, abnormal vibration modes such as mechanical imbalance, gear meshing problems or defects in bearing rolling elements can be identified; these abnormal frequency components Yc are usually related to specific types of faults, so monitoring them helps to detect potential problems of bearings at an early stage and locate the source of faults;

[0097] Deviation from normal value Pl refers to the degree of deviation of certain measured parameters or characteristic values ​​from their expected normal values; deviation indicates the degradation of equipment performance or the early signs of failure; monitoring these deviations from normal value Pl helps to track changes in system performance and detect abnormal behavior or potential problems of equipment in advance, so as to carry out timely maintenance and adjustments;

[0098] The eigenvalue offset Tpy indicates the change in the matrix eigenvalue. The change in the eigenvalue reflects the change in the system state, such as the geometric deformation or stiffness change of the bearing. By analyzing the eigenvalue offset, the dynamic changes of the system can be detected more accurately, and the deterioration of the structure or material can be identified, thereby helping to predict and prevent possible failures.

[0099] The matrix condition number Jzt is an indicator that describes the stability and numerical accuracy of the matrix. A higher condition number usually means that the numerical calculation of the matrix may be unstable, resulting in increased calculation errors. For fault detection, the matrix condition number Jzt is used to reflect the stability and reliability of the data. An abnormal condition number may indicate data quality problems or instability of system parameters, which in turn affects the accuracy of fault diagnosis. Therefore, monitoring the matrix condition number Jzt helps ensure the reliability and accuracy of the diagnostic results.

[0100] Example 6

[0101] S51. Extract the records and related parameters of the bearing fault coefficient Bxgs from the historical data, and perform dimensionless processing on the bearing fault coefficient Bxgs and related parameters, where the related parameters include the radial clearance Jjx between the inner and outer rings of the bearing, the viscosity change Nbh of the lubricating oil, the kurtosis Zxh of the vibration signal, and the thermal conductivity Rdl; obtain the bearing health index Rhzs by calculating the following formula:

[0102]

[0103] S52. Compare and evaluate the preset health threshold H with the health index Rhzs. The specific contents are as follows:

[0104] If the health index Rhzs> the preset health threshold H, it means that the current bearing health index is abnormal and lower than the preset threshold; at this time, the health status of the bearing is marked as the first status result;

[0105] If the health index Rhzs = the preset health threshold, it means that the current bearing health index is normal and just above the preset threshold. At this time, the health status of the bearing is marked as the second status result;

[0106] If the health index Rhzs<the preset health threshold, it means that the current bearing health index is normal and higher than the preset threshold. At this time, the health status of the bearing is marked as the third status result.

[0107] S53: Trigger maintenance or replacement operations according to the evaluation results. The specific contents are as follows:

[0108] If the health status is the first status result: mark the current bearing as an object that needs maintenance or replacement, arrange for professionals to conduct further inspection, determine the specific cause of the failure, and record relevant information for subsequent processing;

[0109] If the health status is the second status result: mark the current bearing as an object that needs further observation. Arrange inspectors to conduct a detailed inspection of the bearing to evaluate whether its condition can be improved through adjustment and repair;

[0110] If the health status is the third status result: continue to use the bearing and proceed to the next working step.

[0111] In this embodiment, by extracting the bearing fault coefficient Bxgs and related parameters from historical data and performing dimensionless processing on them, the present invention significantly improves the calculation accuracy of the bearing health index Rhzs; among them, the comprehensive analysis of parameters such as radial clearance Jjx, viscosity change of lubricating oil Nbh, kurtosis Zxh of vibration signal and thermal conductivity Rdl provides a comprehensive fault assessment basis, making the assessment of the health index more scientific and targeted; when compared with the preset health threshold H, the health status of the bearing can be accurately identified, and it can be clearly marked as a state that requires maintenance, replacement or continued use, thereby optimizing the maintenance decision-making process and reducing maintenance costs; this method effectively avoids the limitations of judging by a single indicator, improves the reliability and response speed of fault detection, and provides a scientific basis for the maintenance and replacement of bearings;

[0112] Among them, the radial clearance Jjx is the space between the inner and outer rings of the bearing, which directly affects the running accuracy and stability of the bearing; excessive radial clearance Jjx may lead to increased relative movement of the internal parts of the bearing, resulting in increased vibration and noise, accelerated wear and failure; therefore, monitoring the radial clearance Jjx helps prevent failures caused by excessive clearance;

[0113] The significance of collecting the viscosity change Nbh of the lubricating oil is that the viscosity of the lubricating oil affects the lubrication effect and the friction characteristics of the bearing; the viscosity change may occur due to temperature fluctuations, oil aging or the presence of contaminants; too low viscosity of the lubricating oil may lead to insufficient lubrication, while too high viscosity may lead to increased flow resistance. Monitoring the viscosity change helps to evaluate the lubrication status and prevent insufficient lubrication or excessive damage to the bearing;

[0114] The kurtosis Zxh of the vibration signal is a statistical measure of the intensity of sharp pulses in the vibration signal; high kurtosis usually indicates irregular impacts or severe vibrations, which may be a sign of bearing failure; by analyzing the kurtosis, potential mechanical failures, including impacts, imbalances or defects, can be identified, and timely measures can be taken to prevent further damage;

[0115] Thermal conductivity Rd is used to measure the ability of a material to conduct heat. For bearings, thermal conductivity Rd can reflect the thermal management performance of its material. Low thermal conductivity Rd may lead to heat accumulation, causing overheating and lubricant degradation, affecting bearing performance and life. Monitoring thermal conductivity Rd helps evaluate the thermal management of bearings and prevent damage caused by overheating.

[0116] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. Bearing fault detection method based on Gershgorin circle theorem, characterized by: The following steps are involved: S1. Install the bearing to be tested on the test bench in advance, use the multi-channel sensor system to monitor the bearing in real time, and collect the vibration signal data of the bearing during operation; upload the vibration signal data of the healthy bearing as standard data to the reference database; S2. Then, the vibration signal data collected from the bearing to be inspected is input into the signal processing module; the original signal is decomposed into different frequency components by using the fast Fourier transform signal processing technology, and the background noise is removed to obtain clear characteristic signal data; S3, then, the processed characteristic signal data is used to construct an autocorrelation matrix or a covariance matrix; the eigenvalue distribution of the matrix is ​​calculated using the Gershgorin circle theorem, and a Gershgorin disk is constructed according to the position of the eigenvalue to determine whether there is an abnormal eigenvalue distribution; S4. If an abnormal characteristic value distribution is detected, the system will issue a warning signal and further analyze the health status of the current bearing by comparing the Gershgorin disk of the standard bearing in the reference database; for the abnormal characteristic value distribution found, the corresponding fault characteristic factor is calculated, and the bearing fault coefficient Bxgs is constructed based on the change trend of the characteristic factor; S5. Finally, the bearing failure coefficient Bxgs is correlated with historical data using machine learning technology, and after dimensionless processing, the bearing health index Rhzs is fitted; Compare the health index Rhzs with the preset health threshold H, comprehensively evaluate the current bearing fault level, and trigger maintenance or replacement operations based on the evaluation results; S41. After detecting abnormal distribution of characteristic values, an early warning signal is issued, and the health status of the current bearing is analyzed by comparing it with the Gershgorin disk of the standard bearing in the reference database. The specific process is as follows: S411, comparative analysis: comparing the characteristic value distribution of the current bearing with the Gershgorin disk of the standard bearing to determine whether the current characteristic value exceeds the normal range; S412, issuing an early warning: when an abnormal distribution is detected, the system generates and issues an early warning signal to prompt the operator or maintenance system to conduct further inspection; S42. According to the distribution of abnormal characteristic values ​​found, the corresponding fault characteristic factors are collected, and the bearing fault coefficient Bxgs is constructed based on the change trend of the characteristic factors. The specific contents are as follows: Extract relevant fault characteristic factors, including abnormal frequency component Yc, deviation from normal value Pl, eigenvalue offset Tpy and matrix condition number Jzt. After dimensionless processing of the extracted relevant fault characteristic factors, calculate the bearing fault coefficient Bxgs by the following formula: S51. Extract the records and related parameters of the bearing fault coefficient Bxgs from the historical data, and perform dimensionless processing on the bearing fault coefficient Bxgs and related parameters, where the related parameters include the radial clearance Jjx between the inner and outer rings of the bearing, the viscosity change Nbh of the lubricating oil, the kurtosis Zxh of the vibration signal, and the thermal conductivity Rdl; obtain the bearing health index Rhzs by calculating the following formula:

2. The bearing fault detection method based on the Gershgorin circle theorem according to claim 1 is characterized in that: S11. Install the bearing to be inspected on the test bench in advance, and deploy a multi-channel sensor system according to the installation position of the bearing to be inspected. The vibration signal of the bearing to be inspected is monitored in real time through the multi-channel sensor system, and the vibration signal data of different channels are collected and marked, and these signal data are associated and marked with corresponding channel information.

3. The bearing fault detection method based on Gershgorin circle theorem according to claim 1 is characterized in that: S21, inputting the vibration signal data of the bearing to be inspected into the signal processing module, applying the fast Fourier transform technology to decompose the input original vibration signal in the frequency domain, obtaining signal data of different frequency components, and storing the data of these frequency components in sequence; S22, processing the signal data of different frequency components that have been decomposed, applying a noise filtering algorithm, removing background noise components, retaining valid characteristic signal data, and transmitting the denoised characteristic signal data as output to a subsequent analysis module.

4. The bearing fault detection method based on Gershgorin circle theorem according to claim 1 is characterized in that: S31. Input the processed characteristic signal data into the data processing system, and construct an autocorrelation matrix related to the vibration signal of the bearing to be inspected through an autocorrelation analysis method, wherein each element of the matrix represents the signal correlation of the bearing to be inspected at different time points.

5. The bearing fault detection method based on Gershgorin circle theorem according to claim 1 is characterized in that: S32. From the constructed autocorrelation matrix, by collecting all diagonal elements Ai and off-diagonal elements Aj, using the Gershgorin circle theorem calculation formula, the center C and radius R of the matrix are obtained, and finally the center and radius of all Gershgorin disks are calculated row by row, and their distribution on the complex plane is inferred, so as to obtain the eigenvalue T of the matrix. The specific calculation process is as follows: Given an n×n square matrix A, whose elements are aij (elements in the i-th row and j-th column): Then obtain the eigenvalue T of matrix A by solving the following characteristic equation: det(A-TI)=0; Where I is the n×n identity matrix, T is the eigenvalue to be found, and det represents the determinant; Then solve the characteristic equation, calculate the determinant of the matrix and expand it, and finally get n eigenvalues ​​and plot them on the complex plane. Compare the position of each eigenvalue to check whether it falls within the corresponding Gershgorin disk, and determine whether there is an abnormal eigenvalue distribution in the current matrix. Specifically: Normal case: If all eigenvalues ​​fall inside the corresponding Gershgorin disk, it means that the eigenvalue T distribution of the matrix is ​​normal; Abnormal situation: If one or more eigenvalues ​​T fall outside the disk, it means that the distribution of the eigenvalues ​​T of the matrix is ​​abnormal, indicating that the matrix properties have changed.

6. The bearing fault detection method based on Gershgorin circle theorem according to claim 1, characterized in that: S52. Compare and evaluate the preset health threshold H with the health index Rhzs. The specific contents are as follows: If the health index Rhzs> the preset health threshold H, it means that the current bearing health index is abnormal and lower than the preset threshold; at this time, the health status of the bearing is marked as the first status result; If the health index Rhzs = the preset health threshold, it means that the current bearing health index is normal and just above the preset threshold. At this time, the health status of the bearing is marked as the second status result; If the health index Rhzs<the preset health threshold, it means that the current bearing health index is normal and higher than the preset threshold. At this time, the health status of the bearing is marked as the third status result.

7. The bearing fault detection method based on Gershgorin circle theorem according to claim 1 is characterized in that: S53: Trigger maintenance or replacement operations according to the evaluation results. The specific contents are as follows: If the health status is the first status result: mark the current bearing as an object that needs maintenance or replacement, arrange for professionals to conduct further inspection, determine the specific cause of the failure, and record relevant information for subsequent processing; If the health status is the second status result: mark the current bearing as an object that needs further observation, arrange for inspectors to conduct a detailed inspection of the bearing to evaluate whether its condition can be improved through adjustment and repair; If the health status is the third status result: continue to use the bearing and proceed to the next working step.

Images