A rolling bearing fault detection method

Abstract

The present application discloses a rolling bearing fault detection method, which relates to the field of rolling bearing fault detection; the method comprises: obtaining information data of a rolling bearing to be tested; using a parameter self-consistent variational modal decomposition method to perform modal decomposition on vibration signal data to obtain a modal function set; calculating the relevant kurtosis value under each mode according to the vibration signal data and the modal function set, then screening multiple modal functions in the modal function set to obtain valid modal functions, and then performing reconstruction processing and Hilbert demodulation processing, and judging whether the rolling bearing to be tested has a fault according to the obtained envelope spectrum; if so, determining the occurrence location of the fault according to the envelope spectrum and the theoretical value of the fault characteristic frequency; the present invention can improve the accuracy of the fault detection result.

Description

Technical Field

[0001] The present application relates to the field of rolling bearing fault detection, and in particular to a rolling bearing fault detection method. Background Art

[0002] In recent years, the manufacturing industry has developed rapidly and has become an important engine of economic growth. As the core foundation of the manufacturing industry, mechanical equipment is widely used in daily transportation such as subways, trains, and cars, as well as in high-tech fields such as integrated circuits, military industry, and aerospace. With the continuous advancement of science and technology and the continuous increase in production scale, various types of mechanical equipment have become more and more complex.

[0003] Rolling bearings are indispensable rotating parts in mechanical equipment and are also one of the important sources of failure in mechanical equipment. Compared with other parts, rolling bearings have significant life discreteness. Some bearings can still operate normally even if they exceed their design life, while others will have various failures before reaching their design life. Therefore, the fault detection method of rolling bearings has always been one of the important development directions in mechanical fault detection. According to statistics, about 30% of the faults in rotating machinery are caused by rolling bearings; in the detection of induction motor faults, the faults caused by rolling bearings account for about 40%; and in gearbox faults, the failure rate caused by bearings is second only to gears, accounting for about 20%.

[0004] Therefore, managers and researchers from companies around the world have begun to try to understand the true operating status of bearings, how to obtain their fault locations and types, and to change from regular maintenance to condition-based maintenance. Statistics show that after the equipment applies condition monitoring and fault detection technology, the accident rate has been reduced by 75% and the maintenance cost has been reduced by 25% to 50%.

[0005] However, as fault detection technology continues to mature, some researchers have discovered new problems. In engineering practice, the vibration signal obtained by the sensor is generated by the superposition of multi-frequency information and noise interference. It is difficult to directly obtain the key features related to the fault from the original vibration signal, and it is impossible to accurately determine whether the bearing has a fault. At present, advanced signal processing technology is one of the important tools to solve the above problems. By decomposing the original vibration signal, it can be split into more basic and easier to handle superposition components. This type of technology can not only separate fault-related information and noise, but also further reveal the potential characteristics of complex signals, such as frequency, amplitude, phase, etc.

[0006] As the most classic and widely used signal decomposition method, empirical mode decomposition has made great contributions to the fields of bearing condition detection and performance degradation assessment. However, both the method itself and its related improvements have problems such as mode aliasing, mode splitting, and endpoint effect. Based on this, a new variational mode decomposition algorithm was proposed in 2014 to replace it. With the help of the Wiener filter group, this method can better suppress noise and overcome the aliasing problem. However, it is worth noting that although it is an adaptive algorithm, its performance depends on the pre-setting of key parameters. Moreover, from the current research, its parameter problem has not been well solved and is still subject to great limitations. Summary of the invention

[0007] The purpose of this application is to provide a rolling bearing fault detection method, which can improve the accuracy of fault detection results.

[0008] To achieve the above objectives, this application provides the following solutions:

[0009] The present application provides a rolling bearing fault detection method, the rolling bearing fault detection method comprising:

[0010] Acquire information data of the rolling bearing to be tested; the information data includes: vibration signal data and bearing component parameter data; the bearing component parameter data includes: number of rolling elements, rolling element diameter, pitch diameter and contact angle;

[0011] Using a parameter self-consistent variational modal decomposition method, the vibration signal data is modally decomposed to obtain a modal function set; the modal function set includes multiple modal functions;

[0012] Calculating the correlation-kurtosis value under each mode according to the vibration signal data and the modal function set; the correlation-kurtosis value is determined based on the correlation coefficient and the kurtosis value;

[0013] Screening multiple modal functions in the modal function set according to the correlation-kurtosis value to obtain a valid modal function;

[0014] Reconstructing the effective mode function to determine a characteristic signal;

[0015] Performing Hilbert demodulation on the characteristic signal to obtain an envelope spectrum;

[0016] Determining whether the rolling bearing to be tested has a fault according to the envelope spectrum;

[0017] If so, the location of the fault is determined based on the envelope spectrum and the theoretical value of the fault characteristic frequency; the theoretical value of the fault characteristic frequency is determined based on the bearing part parameter data and the rotation speed; the theoretical value of the fault characteristic frequency includes: the outer ring fault characteristic frequency, the inner ring fault characteristic frequency and the rolling element fault characteristic frequency.

[0018] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0019] The present application provides a rolling bearing fault detection method, which obtains information data of a rolling bearing to be tested; adopts a parameter self-consistent variational modal decomposition method to perform modal decomposition on vibration signal data to obtain a modal function set; calculates the correlation-kurtosis value under each mode according to the vibration signal data and the modal function set, then screens multiple modal functions in the modal function set to obtain a valid modal function, and then performs reconstruction processing and Hilbert demodulation processing, and judges whether the rolling bearing to be tested has a fault according to the obtained envelope spectrum; if so, determines the location of the fault according to the envelope spectrum and the theoretical value of the fault characteristic frequency; the present invention combines parameter self-consistent variational modal decomposition, correlation-kurtosis index and Hilbert envelope spectrum, breaks through the parameter preset problem, can extract fault characteristics under different complex working conditions, has high efficiency and intuitiveness, and thus improves the accuracy of fault detection results. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0021] Figure 1 This is a flow chart of a rolling bearing fault detection method in an embodiment of the present application;

[0022] Figure 2 It is a schematic diagram of the specific steps of the rolling bearing fault detection method in practical application;

[0023] Figure 3 This is a schematic diagram of the effective mode function screening principle;

[0024] Figure 4 Schematic diagram of envelope spectrum of normal signal;

[0025] Figure 5 Schematic diagram of envelope spectrum of fault signal;

[0026] Figure 6 This is a schematic diagram of the normal signal collected;

[0027] Figure 7 This is a schematic diagram of the collected outer ring fault signal;

[0028] Figure 8 This is a schematic diagram of the collected inner ring fault signal;

[0029] Fig. 9 It is a schematic diagram of the collected ball or rolling element fault signal;

[0030] Fig.10 Schematic diagram of the modal function of a normal signal and its associated kurtosis value;

[0031] Fig.11 Schematic diagram of the modal function of the outer ring fault signal and its correlation-kurtosis value;

[0032] Fig.12 Schematic diagram of the modal function of the inner ring fault signal and its correlation-kurtosis value;

[0033] Fig.13 Schematic diagram of the modal function of the rolling element fault signal and its related-kurtosis value;

[0034] Fig.14 This is a schematic diagram showing that the detection result is a normal signal;

[0035] Fig.15 The detection result is a schematic diagram of the outer ring fault signal;

[0036] Fig.16 The detection result is a schematic diagram of the inner ring fault signal;

[0037] Fig.17 The detection result is a schematic diagram of the rolling element failure signal. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0039] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0040] The present invention combines parameter self-consistent variational mode decomposition, correlation-kurtosis index and Hilbert envelope spectrum, breaks through the parameter preset problem, can extract fault characteristics under different complex working conditions, has high efficiency and intuitiveness, and significantly improves the accuracy of fault detection.

[0041] like Figure 1 As shown, a rolling bearing fault detection method, the method comprising:

[0042] Step 100: Obtain information data of the rolling bearing to be tested. The information data includes: vibration signal data and bearing component parameter data; the bearing component parameter data includes: the number of rolling elements, rolling element diameter, pitch diameter and contact angle.

[0043] Step 200: Using a parameter self-consistent variational modal decomposition method, modal decomposition is performed on the vibration signal data to obtain a modal function set. The modal function set includes multiple modal functions.

[0044] Among them, a parameter self-consistent variational modal decomposition method is used to perform modal decomposition on the vibration signal data to obtain a modal function set, which specifically includes:

[0045] Determine initial parameters; the initial parameters include: Lagrange multiplier, convergence tolerance, noise tolerance and threshold.

[0046] It is determined whether the current number of iterations is 1, and a first determination result is obtained.

[0047] If the first judgment result is yes, the modal function at the current number of iterations is determined based on the Lagrange multiplier and the parameter combination at the current number of iterations; the parameter combination includes: a penalty parameter and a center frequency; the center frequency is determined based on the spectrum of the current residual signal.

[0048] The signal kurtosis at the current iteration number is determined according to the vibration signal data and the modal function at the current iteration number.

[0049] It is determined whether an iteration stop condition is met to obtain a second determination result; the iteration stop condition is that the signal kurtosis at the current iteration number is less than or equal to a threshold.

[0050] If the second judgment result is no, the modal number K at the current iteration number is increased by 1, and the process returns to the step of "determining the modal function at the current iteration number according to the Lagrange multiplier and the parameter combination at the current iteration number".

[0051] If the second judgment result is yes, then the modal function set is determined according to the parameter combination at the current iteration number and the modal function at the current iteration number.

[0052] If the first judgment result is no, then:

[0053] Obtain parameter data; the parameter data includes: the modal number K at the previous iteration number, the parameter combination at the previous iteration number, and the modal function at the previous iteration number.

[0054] Let the value of the current modal number be 1:K.

[0055] According to the parameter data and the current number of modes, the modal function under the current number of iterations is determined.

[0056] The stopping condition is to perform a summation calculation based on the modal function at the current iteration number and the modal function at the previous iteration number, and determine whether the calculation result is less than the convergence tolerance to obtain a third determination result.

[0057] If the third judgment result is no, the Lagrange multiplier is updated according to the noise tolerance and the modal function at the current iteration number, the updated Lagrange multiplier is obtained, and the process returns to the step of "determining the modal function at the current iteration number according to the parameter data and the current modal number".

[0058] If the third judgment result is yes, then the modal function set is determined according to the modal function at the current iteration number and the parameter combination at the current iteration number.

[0059] Step 300: Calculate the correlation-kurtosis value under each mode according to the vibration signal data and the modal function set. The correlation-kurtosis value is determined based on the correlation coefficient and the kurtosis value.

[0060] Specifically, according to the modal function set and vibration signal data, the formula is used Calculate the correlation coefficient for each mode.

[0061] According to the set of modal functions, the formula is used Calculate the kurtosis value for each mode.

[0062] For any mode, the correlation-kurtosis value is determined based on the correlation coefficient and the kurtosis value.

[0063] in, For the The correlation coefficient under each mode; is the total number of sampling points; is the sampling point number; For the In the modal The mode function of the sampling points; for The mean of the corresponding function values; is the signal after the vibration signal is clipped; for The mean of the corresponding signal values; for The standard deviation of the corresponding function value; For the The kurtosis value under each mode.

[0064] The calculation formula of correlation-kurtosis value is:

[0065] .

[0066] in, is the correlation-kurtosis value; For the The kurtosis value under each mode; For the The correlation coefficient under each mode; is the total number of modes; For the The kurtosis value under each mode; For the The correlation coefficient under each mode.

[0067] Step 400: Screen multiple modal functions in the modal function set according to the correlation-kurtosis value to obtain valid modal functions.

[0068] Step 500: Reconstruct the effective mode function to determine the characteristic signal.

[0069] Specifically, for the effective mode function, based on the formula Perform reconstruction processing to determine the characteristic signal.

[0070] in, is the characteristic signal; is the signal after the vibration signal is clipped; For the In the modal The effective mode function of the sampling points; is the number of valid mode functions.

[0071] Step 600: Perform Hilbert demodulation on the characteristic signal to obtain an envelope spectrum.

[0072] Specifically, the characteristic signal is subjected to Hilbert transform processing to obtain a transformed signal.

[0073] Construct an analytical signal based on the transformed signal and the characteristic signal.

[0074] Calculate the amplitude of the analytical signal and obtain the envelope signal.

[0075] Perform fast Fourier transform on the envelope signal to obtain the envelope spectrum.

[0076] Step 700: Determine whether the rolling bearing to be tested has a fault according to the envelope spectrum.

[0077] If yes, then execute step 800: determine the location of the fault according to the envelope spectrum and the theoretical value of the fault characteristic frequency. The theoretical value of the fault characteristic frequency is determined according to the bearing component parameter data and the rotation speed; the theoretical value of the fault characteristic frequency includes: the outer ring fault characteristic frequency, the inner ring fault characteristic frequency and the rolling element fault characteristic frequency.

[0078] The calculation formula of the theoretical value of fault characteristic frequency is:

[0079] ;

[0080] ;

[0081] ;

[0082] in, is the characteristic frequency of outer race fault; is the number of rolling elements; is the rolling element diameter; is the pitch diameter; is the contact angle; is the rotation speed; is the characteristic frequency of inner race fault; is the characteristic frequency of rolling element failure.

[0083] In the prior art, the following three algorithms can be used for fault detection:

[0084] 1. The detection algorithm based on empirical mode decomposition includes the following steps: first, determine the required interpolation technology and stopping criteria, use empirical mode decomposition to obtain a series of intrinsic mode functions, then calculate the kurtosis index, and finally select the component with the maximum kurtosis value for Hilbert transform to obtain fault information. The algorithm is simple to implement; no prior knowledge is required; and it is intuitive. However, its performance depends on the setting of interpolation technology and stopping criteria to adjust the modal function and the number of modes, which has great limitations; this adjustment method depends on the experience and skills of the operator and is highly subjective; the indicators used only focus on the signal distribution density and ignore some effective fault information. Moreover, this manual operation will make it difficult for the algorithm to cope with noise changes under complex working conditions, resulting in feature masking or loss.

[0085] 2. The detection algorithm based on variational modal decomposition includes the following steps: first, the parameters (number of modes, penalty parameters, initial center frequency) are preset, and the modal function is obtained by variational modal decomposition. Secondly, the approximate entropy of the modal function is calculated to obtain the eigenvector. Finally, the diagnostic model is established in combination with the particle swarm optimization algorithm to detect the fault. The algorithm has a strong mathematical foundation and is adaptive; its accuracy and stability are high. However, the parameters rely on manual settings. If the settings are improper, it will lead to abnormal decomposition and volatility in the results; its indicators ignore the key characteristics of the fault. The detection algorithm of variational modal decomposition performs variational solutions for the modal function and its center frequency under the mathematical framework. Compared with empirical mode decomposition, variational modal decomposition introduces the Wiener filtering theory, and iteratively obtains the number of modes and the center frequency through mathematical formulas, which has greater robustness to noise. However, the algorithm does not solve the problem of establishing the number of modes, and in order to better separate the multi-frequency components, the algorithm introduces new parameters to constrain the decomposition. Both the number of modes and these new parameters require prior knowledge and are preset in advance.

[0086] 3. The detection algorithm based on improved variational modal decomposition includes the following steps: First, the Beiying algorithm is used to find the best parameter combination of variational modal decomposition, that is, the number of modes and penalty parameters. Secondly, the Spearman correlation coefficient and wavelet packet decomposition are used to denoise the signal, and finally the filtered signal is Hilbert demodulated. It has a mathematical basis; it can find some optimal parameters; it has noise robustness; and it can intuitively obtain fault information. However, the accuracy of the Beiying algorithm in finding the best parameter combination depends on the establishment of the objective function, and it requires multiple decompositions to obtain the optimal value, which requires a large amount of calculation; this indicator ignores the characteristics of bearing faults.

[0087] In view of the main defects of the above three existing methods, the method proposed in the present invention uses quadratic terms, Wiener filtering, kurtosis and other technologies to adjust the parameters, and realize parameter adaptation while iteratively obtaining the modal function. Secondly, based on the indicator defects of the existing methods, the fault information content in the modal function is quantified by introducing the correlation-kurtosis indicator, and the effective modal function is extracted. Finally, the reconstructed signal is Hilbert demodulated to make the fault information more intuitive. The main advantage of the present invention is that, on the mathematical framework of the original variational modal decomposition, the influence of each parameter is comprehensively considered, and a new definition method or iteration is established through a theoretical basis, which avoids parameter pre-setting and can flexibly cope with feature extraction under different complex working conditions. At the same time, due to the simultaneous adaptation of multiple parameters, the new algorithm has stronger noise robustness, can separate clearer and more comprehensive fault features, and improve the accuracy of fault detection. Figure 2 As shown, the specific steps are as follows.

[0088] 1) Collect vibration signals of rolling bearings , and obtain the relevant parameters of the bearing parts: the number of rolling elements Z, the rolling element diameter d, the pitch diameter D and the contact angle .

[0089] 2) Vibration signal Perform preprocessing.

[0090] The vibration signal collected in 1) According to the number of sampling points To crop:

[0091] .

[0092] 3) Use parameterized self-consistent variational mode decomposition to process the signal Decompose and obtain the modal function and its quantity The detailed decomposition process is as follows.

[0093] enter: .

[0094] Initialization: Initial Lagrange multipliers , convergence tolerance , noise margin , threshold .

[0095] Let the number of iterations , the number of modes .

[0096] Outer loop:

[0097] .

[0098] if do :

[0099] Inner loop:

[0100] .

[0101] Let the initial mode function , initial penalty parameter .

[0102] Get the initial center frequency:

[0103]

[0104] Update the modal function:

[0105]

[0106] Update center frequency:

[0107]

[0108] Update the penalty parameter:

[0109]

[0110] Until the kurtosis of the remaining signal is less than or equal to the threshold, the inner loop stops:

[0111]

[0112] else:

[0113] for do:

[0114] Update the modal function:

[0115]

[0116] Update center frequency:

[0117]

[0118] Update the penalty parameter:

[0119]

[0120] end for

[0121] end if

[0122] Update the Lagrange multipliers:

[0123]

[0124] Until satisfied , stop the outer loop.

[0125] 4) Calculate the correlation-kurtosis value of all mode functions , the mode functions are screened according to the correlation-kurtosis criterion, such as Figure 3 shown.

[0126] First, calculate all mode functions and process signals Correlation coefficient .

[0127] Secondly, the kurtosis values ​​for all modes Perform calculations.

[0128] Then, by calculating the correlation coefficient and kurtosis value , get the correlation-kurtosis value of each mode .

[0129] Finally, only of As an effective mode function ( ) to be retained. indivual( ) Effective mode function .

[0130] 5) Determine whether there is a valid modal function, that is Is it greater than 0? If it exists, reconstruct it and obtain the characteristic signal Otherwise, directly The specific calculation formula is as follows.

[0131]

[0132] 6) Perform Hilbert demodulation to obtain the envelope spectrum.

[0133] First, the characteristic signal Perform a Hilbert transform operation.

[0134]

[0135] Next, build the analytical signal , the formula is as follows:

[0136]

[0137] Then calculate The corresponding envelope signal can be obtained by .

[0138]

[0139] Finally, perform a fast Fourier transform (FFT) to obtain the envelope spectrum .

[0140]

[0141] 7) Determine whether there is a fault based on the envelope spectrum.

[0142] like Figure 4-Figure 5 As shown in the figure, if obvious periodic peaks are observed in the envelope spectrum, the bearing is faulty; otherwise, it is in normal condition.

[0143] Then, by comparing the maximum kurtosis value in the envelope spectrum The theoretical value of the fault characteristic frequency can be used to determine the location of the fault.

[0144] In practical applications, taking bearing parts as an example, the specific operation steps are:

[0145] 1) Obtain multiple bearing parts with the following parameters: number of rolling elements (Z) is 9, rolling element diameter (d) is 7.94 mm, pitch diameter (D) is 39.04 mm, and contact angle is 0.

[0146] Then, the bearing parts are divided into four parts, one part remains unchanged, and the other three parts are linearly cut at the outer ring, inner ring and ball parts.

[0147] 2) Use the accelerometer sensor to collect the vibration signal of the bearing, recorded as . Among them, set the rotation speed 29.95r / s; sampling frequency The sampling frequency is 51.2kHz; the sampling interval T is 20s. The vibration signals of four types of bearing parts (normal, outer ring fault, inner ring fault, and ball fault) are collected using a self-made experimental platform. The specific data examples are as follows Figure 6-Figure 9 shown.

[0148] 3) According to the number of sampling points is 32768, for the vibration signal collected in 2) Perform preprocessing and obtain processing signals .

[0149] 4) First, initialize: initial Lagrange multiplier , convergence tolerance =10 -7 , noise margin , threshold Then, we use the parameterized self-consistent variational mode decomposition to Decompose and obtain the modal function and its quantity .

[0150] 5) Calculate all modal functions Correlation-kurtosis value ,like Figure 10-13 Then, The modal functions of are considered as valid modal functions. Among them, the number of valid modal functions of normal signals is The number of fault signals of the outer ring, inner ring and sphere are and 3.

[0151] 6) According to the number of effective mode functions , calculate the characteristic signal : There is no effective mode function in the normal signal, that is , then process the signal Directly assigned to characteristic signal ; Among the fault signals of the outer ring, inner ring and ball, , then reconstruct the effective mode function and obtain the characteristic signal .

[0152] 7) Characteristic signal Perform Hilbert demodulation to obtain its envelope spectrum.

[0153] 8) By observing the periodic peaks in the envelope spectrum, such as Figure 14-17 As shown in the figure, it is determined that there are obvious fault features in the fault signals of the outer ring, inner ring and ball, and the corresponding bearings are faulty. Then, the maximum peak value of its envelope spectrum is obtained, that is, The values ​​are 109.38Hz, 160.84Hz and 148.41Hz. Then, according to the specific parameters in 1), the theoretical value of the fault characteristic frequency is obtained: the outer ring fault characteristic frequency 107.36Hz; inner ring fault characteristic frequency 162.18Hz; rolling element fault characteristic frequency is 153.35Hz. By By comparing with the theoretical value, the specific situation of the fault was determined, that is, the fault location was located in the outer ring, inner ring and sphere of the bearing, which was consistent with the theory.

[0154] The present invention constructs the expression of the initial center frequency for the first time based on the quadratic term and Wiener filtering. The kurtosis characteristics of noise at different intensities are used to achieve the number of modes. The parameters are self-consistent. According to the orthogonality between the modal functions and the noise, an iterative form of the penalty parameter is established, which can independently adjust the bandwidth of each modal function. For the first time, a parameter self-consistent variational modal decomposition algorithm is proposed, which can efficiently and accurately separate fault characteristics and interference without parameter priors.

[0155] Combining the correlation coefficient and kurtosis, a new fault screening index, the correlation-kurtosis index, was established. This index can comprehensively consider the characteristics of different types of faults and characterize the richness of fault information.

[0156] Since the mathematical framework of variational modal decomposition is adopted, the present invention can avoid the subjectivity introduced by empirical operations, thereby ensuring efficient and accurate feature extraction under different complex working conditions. Since the number of modes, penalty parameters and initial center frequency are all adjusted in an adaptive manner, the present invention can achieve effective separation of multi-frequency components without the need for prior knowledge. Since the iterative framework of variational modal decomposition is improved, the number of modes and penalty parameters are adaptive while solving the modal function, and the present invention avoids the huge amount of calculation caused by multiple runs to find the best parameter combination. In addition, since the parameter adaptation in the present invention is based on modal characteristics or noise characteristics, the problem of setting the objective function is solved.

[0157] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0158] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A rolling bearing fault detection method, characterized in that: The rolling bearing fault detection method comprises: Acquire information data of the rolling bearing to be tested; the information data includes: vibration signal data and bearing component parameter data; the bearing component parameter data includes: number of rolling elements, rolling element diameter, pitch diameter and contact angle; Using a parameter self-consistent variational modal decomposition method, the vibration signal data is modally decomposed to obtain a modal function set; the modal function set includes multiple modal functions; Calculating the correlation-kurtosis value under each mode according to the vibration signal data and the modal function set; the correlation-kurtosis value is determined based on the correlation coefficient and the kurtosis value; Screening multiple modal functions in the modal function set according to the correlation-kurtosis value to obtain a valid modal function; Reconstructing the effective mode function to determine a characteristic signal; Performing Hilbert demodulation on the characteristic signal to obtain an envelope spectrum; Determining whether the rolling bearing to be tested has a fault according to the envelope spectrum; If yes, the location of the fault is determined according to the envelope spectrum and the theoretical value of the fault characteristic frequency; the theoretical value of the fault characteristic frequency is determined according to the bearing component parameter data and the rotation speed; the theoretical value of the fault characteristic frequency includes: the outer ring fault characteristic frequency, the inner ring fault characteristic frequency and the rolling element fault characteristic frequency; The vibration signal data is modally decomposed using a parameter self-consistent variational modal decomposition method to obtain a modal function set, which specifically includes: Determine initial parameters; the initial parameters include: Lagrange multiplier, convergence tolerance, noise tolerance and threshold; Determine whether the current number of iterations is 1, and obtain a first determination result; If the first judgment result is yes, the modal function at the current number of iterations is determined according to the Lagrange multiplier and the parameter combination at the current number of iterations; the parameter combination includes: a penalty parameter and a center frequency; the center frequency is determined according to the spectrum of the current residual signal; the initial center frequency: ； in, is the initial center frequency; is the processed signal in the form of spectrum; is the n+1th iteration The modal function under each mode; For the nth iteration The modal function under each mode; is the Lagrange multiplier for the first iteration; Determining the signal kurtosis at the current iteration number according to the vibration signal data and the modal function at the current iteration number; Determine whether an iteration stop condition is met to obtain a second determination result; the iteration stop condition is that the signal kurtosis at the current iteration number is less than or equal to the threshold; If the second judgment result is no, the modal number K at the current iteration number is increased by 1, and the process returns to the step of "determining the modal function at the current iteration number according to the Lagrange multiplier and the parameter combination at the current iteration number"; and the penalty parameter is updated: ； in, is the updated penalty parameter; is the initial penalty parameter; is the mode function in spectral form; for conjugation of; If the second judgment result is yes, determining a modal function set according to the parameter combination at the current iteration number and the modal function at the current iteration number; If the first judgment result is no, then: Acquire parameter data; the parameter data includes: the modal number K at the last iteration number, the parameter combination at the last iteration number, and the modal function at the last iteration number; Let the value of the current modal number be 1: K; Determine the modal function at the current number of iterations according to the parameter data and the current number of modes; According to the stop condition, a sum calculation is performed based on the modal function at the current iteration number and the modal function at the previous iteration number, and it is determined whether the calculation result is less than the convergence tolerance to obtain a third determination result; If the third judgment result is no, the Lagrange multiplier is updated according to the noise tolerance and the modal function at the current iteration number to obtain an updated Lagrange multiplier, and the process returns to the step of "determining the modal function at the current iteration number according to the parameter data and the current modal number"; If the third judgment result is yes, determining a modal function set according to the modal function at the current iteration number and the parameter combination at the current iteration number; According to the vibration signal data and the modal function set, the correlation-kurtosis value under each mode is calculated, specifically including: According to the modal function set and the vibration signal data, the formula is used Calculate the correlation coefficient under each mode; According to the set of modal functions, the formula Calculate the kurtosis value under each mode; For any mode, determining a correlation-kurtosis value according to the correlation coefficient and the kurtosis value; in, For the The correlation coefficient under each mode; is the total number of sampling points; is the sampling point number; For the In the modal The mode function of the sampling points; for The mean of the corresponding function values; is the signal after the vibration signal is clipped; for The mean of the corresponding signal values; for The standard deviation of the corresponding function value; For the The kurtosis value under each mode; The calculation formula of the correlation-kurtosis value is: ； in, is the correlation-kurtosis value; For the The kurtosis value under each mode; For the The correlation coefficient under each mode; is the total number of modes; For the The kurtosis value under each mode; For the The correlation coefficient under each mode; Keep only of Retained as a valid modal function, thus obtaining Valid modal functions ; ; For the effective mode function, based on the formula Perform reconstruction processing to determine characteristic signals; in, is the characteristic signal; is the signal after the vibration signal is clipped; For the In the modal The effective mode function of the sampling points; is the number of valid mode functions.

2. The rolling bearing fault detection method according to claim 1, characterized in that: Performing Hilbert demodulation on the characteristic signal to obtain an envelope spectrum specifically includes: Performing Hilbert transform processing on the characteristic signal to obtain a transformed signal; constructing an analytical signal according to the transformed signal and the characteristic signal; Calculating the amplitude of the analytical signal to obtain an envelope signal; Performing fast Fourier transform on the envelope signal to obtain the envelope spectrum.

3. The rolling bearing fault detection method according to claim 1, characterized in that: The calculation formula of the theoretical value of the fault characteristic frequency is: ； ； ； in, is the characteristic frequency of outer race fault; is the number of rolling elements; is the rolling element diameter; is the pitch diameter; is the contact angle; is the rotation speed; is the characteristic frequency of inner race fault; is the characteristic frequency of rolling element failure.

Images