CN115436058B - Bearing fault feature extraction method, device, equipment and storage medium - Google Patents
Bearing fault feature extraction method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN115436058B CN115436058B CN202211057305.7A CN202211057305A CN115436058B CN 115436058 B CN115436058 B CN 115436058B CN 202211057305 A CN202211057305 A CN 202211057305A CN 115436058 B CN115436058 B CN 115436058B
- Authority
- CN
- China
- Prior art keywords
- signal
- stfd
- bearing
- fractal dimension
- calculating
- 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
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M13/00—Testing of machine parts
- G01M13/04—Bearings
- G01M13/045—Acoustic or vibration analysis
-
- 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
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computing Systems (AREA)
- Acoustics & Sound (AREA)
- Algebra (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The invention discloses a method, a device, equipment and a computer storage medium for extracting bearing fault characteristics, which comprise the steps of collecting bearing vibration signals and carrying out windowing interception on the bearing vibration signals; calculating fractal dimension values of the intercepted signal fragments by using a Katz method; tracking the impact position generated by bearing faults in the bearing vibration signals according to the fractal dimension values, and extracting an envelope curve of the impact position; and further removing interference on the envelope line according to a peak value searching algorithm to obtain the bearing vibration fault characteristics. According to the method, the vibration sequence of the bearing is converted into the short-time fractal dimension sequence through the fractal dimension, other interference signals are restrained in the conversion process, the signals of bearing faults are more prominent, the interference signals are restrained, finally, all the signals are processed according to the peak value searching algorithm, the influence of the interference signals on the bearing fault signals is further restrained, and the bearing fault signals are more highlighted.
Description
Technical Field
The present invention relates to the field of rolling bearing fault diagnosis technologies, and in particular, to a method, an apparatus, a device, and a computer storage medium for extracting bearing fault characteristics.
Background
Rolling bearings are the most common bearings. They are commonly used to carry heavy objects in harsh environments and are one of the most prone to failure. 44% of large induction motor failures are related to bearing failures. Bearing failure can lead to productivity and economic loss and even catastrophic consequences. Therefore, bearing failure detection is of great significance. Most of the many methods that have been developed are based on sound, temperature, wear and vibration. Among other things, vibration signatures have been found to be a powerful tool for bearing failure detection. When the rolling elements pass through the defective area, small collisions occur, resulting in a series of mechanical pulses. These impulses cause the system natural mechanical frequency to resonate at high speeds. The resonant frequency is then modulated, and the fault signature frequency (Fault Characteristic Frequency: FCF) and detected primary information are named by the bearing fault pulse frequency. One of the most common detection techniques, the signal is enveloped by amplitude demodulation of the vibration signal. The bearing health can then be determined based on the frequency spectrum. Hilbert transform is commonly used for this purpose. However, the hilbert transform is susceptible to noise and interference. In order to obtain reliable detection results, a pretreatment is required and filtration is usually necessary. Typically, a high resonant frequency technique (high resonance frequency technique: HRFT) is used. The main steps of the method include filtering the vibration signal around the resonance frequency, enveloping the filtered signal, and fourier transforming the demodulated signal. The design of bandpass filters is usually based on unknown resonant frequencies, so that proper center frequency and bandwidth are important, and selecting an appropriate solution is a challenging task.
The most widely used envelope methods in the literature for bearing condition monitoring include hilbert transform, EO and MMA. Although they have their own advantages, they all have their own disadvantages, summarized as follows: the hilbert transform (1) is susceptible to vibration interference and noise, and therefore requires pre-filtering, which involves a challenging filter parameter selection process, (2) EO approaches are limited by single components and narrow bands, whose effectiveness is often affected by multiple interferences, (3) for MMA techniques, SE structures, computational effort is high, processing efficiency for multiple interferences is low, limiting their application in bearing fault signature extraction.
From the above, it can be seen how to reduce the influence of the interference signal and improve the fault characteristics is a problem to be solved at present.
Disclosure of Invention
The invention aims to provide a bearing fault feature extraction method, device and equipment and a computer storage medium, which solve the problem of high difficulty in bearing fault feature extraction in the prior art.
In order to solve the technical problems, the invention provides a method for extracting bearing fault characteristics, which comprises the following steps:
collecting a bearing vibration signal, and carrying out windowing interception on the bearing vibration signal;
calculating fractal dimension values of the intercepted bearing vibration signals by using a Katz method;
tracking the impact position generated by bearing faults in the bearing vibration signal according to the fractal dimension value, and extracting an envelope curve of the impact position;
calculating an average value and a standard deviation of each envelope, and calculating a peak value of an envelope signal based on the average value and the standard deviation;
and setting the peak value of the envelope signal with the difference value of the peak value and the standard deviation smaller than 1 as 1, wherein the rest envelope signals are bearing fault characteristics.
Preferably, the calculating the fractal dimension value of the intercepted bearing vibration signal using the Katz method includes:
according to the bearing vibration signal s= { s 1 ,s 2 ,...,s N -calculating the euclidean distance between successive signals as:
wherein N is the total number of sampling points in the bearing vibration signal, s i Is any point in the vibration signal, s i The coordinates are (x) i ,y i ),i＝1,2,...,N；
According to the Euclidean distance between the continuous signals, calculating the plane range d of the waveform and the total length L of the curve, wherein the calculation formula is as follows:
normalizing the total length L of the curve and the planar extent d of the waveform by the average distance a between the successive signals and using the mathematical definition formula of KatzFD Calculating fractal dimension:
where k=n-1 is the number of steps in the curve, a=l/k.
Preferably, the tracking the impact position generated by the bearing fault in the bearing vibration signal according to the fractal dimension value, and extracting the envelope curve of the impact position comprises:
determining the width of a moving window according to the fractal dimension value;
intercepting the bearing vibration signals from beginning to end by utilizing a moving window to obtain a plurality of signal segments;
calculating the fractal dimension value of each bearing vibration section;
and splicing fractal dimension values of all the signal segments to obtain the short-time fractal dimension sequence.
Preferably, said capturing said bearing vibration sequence from beginning to end using said moving window, obtaining a plurality of signal segments comprises:
the matrix formed by the plurality of bearing vibration sections is as follows:
wherein sw is m For each signal segment, N win For window length, N win ＝int(αf s ) Int (-) is the integer part of the value, α is a preset constant, f s Is wave-shapedSampling frequency.
Preferably, the step of splicing the fractal dimension values of all the signal segments to obtain the short-time fractal dimension signal includes:
calculating the fractal dimension value of each signal segment to obtain a short-time fractal dimension vector:
STFD＝[STFD(1),STFD(2),...,STFD(m),...,]；
converting it into the short-time fractal dimension sequence: STFD = STFD original -MIN+1；
Wherein the vector STFD original For the calculated signal STFD sequence, MIN is the sum STFD original The smallest component of the same length, 1 is the same as STFD original The unit vectors of the same length, STFD, are the resulting STFD sequences.
Preferably, said setting the peak value of the envelope signal to 1, which has a difference between the peak value of the envelope signal and the standard deviation of less than 1, the remaining envelope signals being bearing failure characteristics includes:
the formula defining the peak search algorithm:
where μ is the average of the signal x (t), σ is the standard deviation of the signal x (t), E { } is the mathematical expectation operator, x (t) =s (t) +i (t) =a mp e -βt sinω r t+A i cosω i t is; wherein A is mp For the amplitude, ω, of the vibration of the pulse r For the vibration frequency of the pulse A i To the amplitude of the disturbance, ω i Is the frequency of the disturbance;
calculating the standard deviation sigma and the peak value K of each signal;
judging whether the amplitude is greater than or equal to 1+sigma, if the amplitude is greater than or equal to 1+sigma, the amplitude is kept unchanged, and if the amplitude is less than 1+sigma, the amplitude is set to be 1;
judging whether the peak value difference of the two iterations meets a preset value, if so, stopping calculating the peak value, and if not, continuing the iterations;
and stopping iteration until all signal calculation is completed, and obtaining the bearing vibration fault characteristics.
Preferably, the calculation steps of the peak search algorithm are:
s71: initializing, let i=1,
s72: calculating the standard deviation sigma of the signal segment x (i) i And peak value K i ；
S73: judging whether the amplitude in the signal section is more than or equal to 1+sigma i ；
S74: if greater than or equal to 1+sigma i The amplitude remains unchanged; if less than 1+sigma i Let the amplitude set to 1;
s75: judging that i is more than or equal to N, and stopping iteration if the i is more than or equal to N;
s76: if the result is not true, the method comprises the steps of, then determine if |K is satisfied i -K i-1 And if the I is less than or equal to epsilon, stopping calculating K i -K i-1 Wherein ε is a threshold minimum constant;
s77: if not, i=i+1 is set to step S72.
The invention also provides a device for extracting the bearing fault characteristics, which comprises the following steps:
the signal acquisition module is used for acquiring bearing vibration signals and carrying out windowing interception on the bearing vibration signals;
the fractal dimension calculating module is used for calculating the fractal dimension value of the intercepted bearing vibration signal by using a Katz method;
the envelope extraction module is used for tracking the impact position generated by bearing faults in the bearing vibration signals according to the fractal dimension values and extracting an envelope of the impact position;
a calculation peak module for calculating an average value and a standard deviation of each envelope, and calculating a peak value of the envelope signal based on the average value and the standard deviation;
and the interference suppression module is used for setting the peak value of the envelope signal with the difference value of the peak value and the standard deviation of the envelope signal smaller than 1 as 1, and the rest envelope signals are bearing fault characteristics.
The invention also provides a device for extracting the bearing fault characteristics, which comprises:
a memory for storing a computer program; and the processor is used for realizing the steps of the method for extracting the bearing fault characteristics when executing the computer program.
The invention also provides a computer readable storage medium having stored thereon a computer program which when executed by a processor performs the steps of a method of bearing fault signature extraction as described above.
According to the method for extracting the bearing fault characteristics, firstly, bearing vibration signals during bearing operation are collected, windowing interception is carried out, the fractal dimension value of the intercepted bearing vibration signals is calculated by using a Katz method, the bearing vibration signals are converted into short-time fractal dimension sequences by using the fractal dimension value, the envelope curve of impact positions generated by faults is extracted by using the fractal dimension values, the fault signals and the interference signals are distinguished, the impact of the interference signals is well restrained, and the fault signals are highlighted; and finally, setting the amplitude value smaller than the standard deviation as 1 according to a peak value searching algorithm, so that fault signals are more highlighted. Finally, the vibration fault characteristics of the bearing are obtained. According to the method, the vibration sequence of the bearing is converted into the short-time fractal dimension sequence through the fractal dimension, other interference signals are restrained in the conversion process, the signals of bearing faults are more prominent, the interference signals are restrained, finally, all the signals are processed according to the peak value searching algorithm, the influence of the interference signals on the bearing fault signals is further restrained, the bearing fault signals are more prominent, the signals are not required to be filtered by a filter, various interference signals cannot be restrained, the calculation amount is huge, the practical range is wider, and the online bearing state monitoring can be realized.
Drawings
For a clearer description of embodiments of the invention or of the prior art, the drawings that are used in the description of the embodiments or of the prior art will be briefly described, it being apparent that the drawings in the description below are only some embodiments of the invention, and that other drawings can be obtained from them without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a first embodiment of a method for bearing fault extraction provided by the present invention;
FIG. 2 is a flow chart of a second embodiment of a method for bearing fault extraction provided by the present invention;
FIG. 3 is a comparison graph of bearing failure feature extraction using STFD transforms: (a) a pulse signal; (b) STFD obtained by Katz, sevcik and Higuchi methods;
FIG. 4 (a) curve u of (different) STFD sequences generated at different window lengths; (b) A plot of the evolution of correlation coefficients between the analog signal envelope and the generated STFD sequence at different window lengths;
fig. 5 is a graph of interference suppression for STFD transformation of signal mix (pulse and random interference): (a) periodic interference; (b) analog pulses; (c) a mixture of (a) and (b); (d) STFD representation of signal mixing;
FIG. 6 is a graph of STFD and STFD-KPSA conversion results for simulated fault orientation signals with multiple cyclic disturbances: (a) signal interference mixing; (b) STFD sequence without KPSA; (c) spectrum of b, (d) STFD-KPSA sequence; (e) the spectrum of d;
FIG. 7 is a diagram of the detectability of the STFD-KPSA method;
FIG. 8 is a graph of a proposed method applied to an analog signal, (a) synthesizing a signal; (b) a hilbert envelope spectrum of a; (c) an STFD sequence; (d) STFD-KPSA results; (e) the spectrum of c; (f) the spectrum of d;
FIG. 9 is a graph of STFD-KPSA performance in detecting side-positioned outer race faults: (a) a raw vibration signal; (b) a hilbert envelope spectrum of the original signal; (c) STFD-KPSA results of the original signal; (d) spectrum of STFD-KPSA results;
FIG. 10 is a graph of the performance of STFD-KPSA to detect bottom out failure: (a) a raw vibration signal; (b) a hilbert envelope spectrum of the original signal; (c) STFD-KPSA results of the original signal; (d) spectrum of STFD-KPSA results;
FIG. 11 is a graph of the performance of STFD-KPSA to detect an inner circle failure: (a) a raw vibration signal; (b) a hilbert envelope spectrum of the original signal; (c) STFD-KPSA of the original signal; (d) spectrum of STFD-KPSA results;
fig. 12 is a block diagram of a device for extracting bearing fault characteristics according to an embodiment of the present invention.
Detailed Description
The core of the invention is to provide a bearing fault extraction method, which utilizes fractal dimension to convert into short-time fractal dimension sequence, effectively inhibits the influence of interference signals, and utilizes peak search algorithm to further inhibit interference and improve fault characteristics.
In order to better understand the aspects of the present invention, the present invention will be described in further detail with reference to the accompanying drawings and detailed description. It will be apparent that the described embodiments are only some, but not all, embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Referring to fig. 1, fig. 1 is a flowchart of a first embodiment of a method for extracting bearing fault characteristics according to the present invention; the specific operation steps are as follows:
step S101: collecting a bearing vibration signal, and carrying out windowing interception on the bearing vibration signal;
and acquiring a bearing vibration fault signal by using a vibration sensor, and carrying out windowing interception on the acquired bearing vibration signal.
Step S102: calculating fractal dimension values of the intercepted bearing vibration signals by using a Katz method;
step S103: tracking the impact position generated by bearing faults in the bearing vibration signal according to the fractal dimension value, and extracting an envelope curve of the impact position;
step S104: calculating an average value and a standard deviation of each envelope, and calculating a peak value of an envelope signal based on the average value and the standard deviation;
step S105: and setting the peak value of the envelope signal with the difference value of the peak value and the standard deviation smaller than 1 as 1, wherein the rest envelope signals are bearing fault characteristics.
In the embodiment, firstly, a vibration sensor is used for collecting signals of bearing vibration, then a Katz algorithm is used for calculating the distance between the signals and the total length of a curve, further fractal dimension values between the signals are calculated, and then the bearing signals are converted into short-time fractal dimension sequences based on the fractal dimension values; the fractal dimension conversion is adopted to inhibit the interference signal, so that the fault signal is more prominent. And removing the interference signal according to a peak value searching algorithm, reserving the fault signal, and finally obtaining a more perfect fault signal. The invention can monitor the bearing fault state on line in real time.
Based on the above embodiments, the present embodiment describes the method for extracting the bearing fault characteristics in detail, please refer to fig. 2, fig. 2 is a flowchart of a second specific embodiment of the method for extracting the bearing fault characteristics provided by the present invention; the specific operation steps are as follows:
step S201: collecting bearing vibration signals by using a vibration sensor;
step S202: calculating Euclidean distance and total length of a curve between signals according to the bearing vibration signals;
referring to fig. 3, fig. 3 is a comparison chart of the present invention for extracting bearing failure features using STFD transformation, wherein: (a) a pulse signal; (b) And (c) and (d) respectively using Katz, sevcik and Higuchi methods, and comparing to obtain the best effect by using the Katz method, and uniformly differentiating peaks of the images.
The acquired vibration signal is s= { s 1 ,s 2 ,...,s N Where N is the total number of sample points in the signal. An arbitrary point in the signal is defined by s i Is represented by the coordinates (x) i ,y i ) I=1, 2,..n. The euclidean distance between two consecutive points is
KatzFD of the N-point signal is mathematically defined as:
where L is the total length of the curve, derived from the sum of Euclidean distances of all adjacent points, i.e
d is the planar extent of the waveform and can generally be considered as the furthest distance of all distances between the first point (point 1) and all subsequent points of the signal. Mathematically, d can be written as
The FD value obtained from equation (2) is affected by the unit of measurement. To overcome this difficulty, the kaz proposal normalizes L and d by the average distance (denoted by a) between consecutive points. The average distance a is defined as a=l/k, where k is the number of steps in the curve, i.e. k=n-1. In equation (2), division of a by L and d yields
Step S203: calculating a fractal dimension value according to the Euclidean distance and the total length of the curve;
step S204: windowing and intercepting the bearing vibration signal by using the fractal dimension value to obtain a plurality of bearing signal segments;
FD is used to track the location of the impact caused by bearing failure and extract the envelope of the impact. The typical vibration signal generated by a bearing failure is periodic, including a sharp rise corresponding to an impact between the defect location surfaces. To extract periodically affected features using FD, STFD was studied.
The moving window is used to truncate the original signal to obtain a STFD sequence, wherein each STFD value corresponds to a windowed signal portion. The length of the window is selected to be N win ＝int(αf s ) Where int (-) represents the integer part of the value, is a user pre-specified constant α, f s Is the sampling frequency of the waveform. The constant α is empirically set. Since the fault-induced pulses are usually weak and masked by background noise and vibration disturbances, the window must be shorter to obtain satisfactory results. This is because too long a window may make the STFD sequence excessively smooth. Therefore, in bearing failure detection applications, the constant α should be smaller than the constant used for acoustic signal processing to ensure that the initial pulse is not missed. However, it should be noted that N win And not too short. Too short a window may result in many imaginary oscillations, obviously failing to correctly identify the effects, as shown in fig. 4, fig. 4 (a) STFD sequences generated at different window lengths (different α); fig. 4 (b) simulates the evolution of correlation coefficients between the envelope signal and the generated STFD sequence at different window lengths.
To obtain a point-to-point time sequence of STFD values, the window is shifted along the input vector one sample step at a time. W [ n ] for window function]And (3) representing. Then, sequence sw m [n]＝s[n]w[m-n]Is the signal s [ n ] at time m]Length of N win . Representing all signal segments in matrix form, the following expression can be obtained
For each signal segment w m By STFD (m) The corresponding FD value represented can be calculated by equation (5). Vector STFD can be obtained by calculating FD values for all signal segments as follows
STFD＝[STFD(1),STFD(2),...,STFD(m),...,] (7)
Applying the following transform in the subsequent subsection without affecting the dynamic range of the STFD sequence
STFD＝STFD original -MIN+1 (8)
Wherein the vector STFD original Is a signal STFD sequence calculated using equation (5), MIN is a signal corresponding to STFD original Vectors of equal length (size) with all components equal to STFD original 1 represents the minimum component of (1) and STFD as well original The unit vectors of the same length, STFD, represent the obtained STFD sequence. According to equation (8), the minimum value of the STFD sequence is always limited to 1.
To convert the original vibration signal into an STFD sequence, the FD value of each signal segment is calculated by equation (5) and assigned to the midpoint of the signal segment. However, this will result in a length of N-N of the STFD sequence obtained win +1 is shorter than the length N of the original signal. To preserve the signal length, the original signal is spread by replication: (a) First half of the number of points at the beginning, i.e. odd N win Of (N) win -1)/2 points, (N) win /2, even), and (b) the last (N) at the end of the original signal win -1)/2, odd number N win (N win -2/2, even N win ) Points. The length of the signal is then extended to n+n win -1. Thus, the size of the obtained STFD sequence is the same as the original signal.
Step S205: calculating the fractal dimension value of each bearing signal segment;
step S206: splicing fractal dimension values of all bearing signal segments to obtain the short-time fractal dimension sequence;
step S207: and removing interference on the fractal dimension sequence according to a peak value searching method to obtain bearing vibration fault characteristics.
Periodic disturbances can be expressed asAnd A is im And omega im Representing the amplitude and frequency of the disturbance, respectively. To evaluate the effect of disturbances on the conversion envelope demodulation, consider an analog pulse caused by a single fault contaminated by disturbances with amplitudes greater than the pulse x (t) =s (t) +i (t) =a mp e -βt sin ω r t+A i cosω i t, wherein A mp Is the amplitude of the pulse and,A i >A mp indicating that the fault-related pulses are severely masked by the disturbances. Analysis of the signal denoted by X (t).
The envelope (or instantaneous amplitude) of the analysis signal can be calculated by the following formula:
equation (10) shows that the main component, i.e., the interference, is envelope demodulated, but pulse demodulation is unsuccessful. It can be concluded that conventional demodulation is adversely affected by strong interference. Thus, an envelope method with interference suppression capability is highly desired.
Interference suppression based on STFD transformation. Referring to fig. 5, consider random interference (i (t) =2cos (2pi ft)) as shown in fig. 5 (a). Fig. 5 (b) and (c) are respectively a pulse signal and a signal mixture consisting of the pulse signal and the disturbance. For most practical applications, the resonance frequency is much higher than the frequency of the periodic disturbance, i.e. it can be observed that the disturbance frequency is much lower than the frequency of the pulses, as shown in fig. 5 (a) and (b). Let us look at the signal segment within one short window, the nth window in fig. 5 (a). Euclidean distance of the curve within the nth window, n=1, 2.. n (defined by equation (3)), the plane range is defined by d n (defined by equation (4)). If the vibration disturbance frequency is sufficiently low, L is selected as long as the appropriate window length n Almost equal to d in FIG. 5 (a) n This will result in STFD (n) ≡1 (so that d n ≈L n The satisfaction is. It can be concluded that the interference has little effect on the STFD representation. For pulse signals, the planar extent (d n ) And Euclidean distance (L n ) The difference is greater because of the significant morphological appearance change, as shown in FIG. 5 (c), the STFD (n) is derived according to equation (5)>1. Since the resonance frequency is typically much higher than the frequency of the vibration disturbance, the pulsesThe resulting STFD values are much larger than the interference-related STFD values, indicating that the pulse-related STFD values are more pronounced in the STFD representation than the interference-related STFD values. Thus, the envelope of the pulse can be highlighted and interference suppressed, as shown in fig. 5 (d).
Interference suppression is performed by a combination of STFD and KPSA. However, in some cases, the effect on the high frequency interference is still not negligible. For example, a composite signal consisting of a plurality of disturbances (cos (2π 1000 t), cos (2π620 t) and cos (2π 300 t)) and analog pulses is shown in FIG. 6 (a). The STFD transform results and their spectra are shown in fig. 6 (b) and (c), respectively. It can be seen that the composite effect of the interference is not successfully eliminated, since the STFD value associated with the interference has a greater impact on the STFD representation due to the high frequency. Therefore, the spectrum in fig. 6 (c) is mainly controlled by the interference frequency difference, which will confuse or even mislead the detection result.
In order to solve the problems caused by the multiple disturbances of relatively high frequencies, it is necessary to improve the proposed method in order to further suppress the influence of the disturbances. The idea of KPSA is to extract useful information and to remove unwanted signal components as iteratively as possible until a stopping criterion is reached. The criterion is kurtosis, as it has proven to be a valid indicator of impulse. It is defined as:
where μ and σ are the mean and standard deviation of the signal x (t), respectively, and E { } is the mathematical expectation operator. Taking into account the uncertainty of the signal, the standard deviation sigma is introduced i Is a tolerance range of (c). Thus, in each iteration of the KPSA, e.g., iteration i, the STFD transform result is calculated (i.e., STFD i Vector representing STFD sequence) standard deviation σ i And kurtosis K i . Higher than 1+sigma i The amplitude of the points of (a) remains unchanged. Lower than 1+sigma i The amplitude of the point of (c) is "removed", i.e., set to 1.0 (since the minimum FD value is 1 according to equation (6)). In this way, a new vector STFD is constructed *i Representing the extracted presence in iteration iInformation is used. However, some low energy points, i.e. amplitudes below 1+σ i May also be caused by a fault. Thus, a new iteration i+1 is required to further extract information of such low energy pulses, as long as K i And K i-1 The difference between them is still significant, i.e. the impulsive difference between two successive iterations is desirable. In other words, in each iteration, e.g., i+1, the above procedure is applied to the signal remainder of iteration i, i.e., STFD i +1＝STFD i -STFD * i +1, and the iterative process continues until the pulse nature of the signal remainder becomes insignificant, i.e., |K i -K i-1 +.ltoreq.ε (ε is a pre-specified small positive number). The final signal residual is then considered to be comprised of the STFD component caused by noise and/or interference and is therefore removed. The final STFD sequence after KPSA application (hereinafter STFD-KPSA) is given by the following equation:
application of the KPSA algorithm described above to the STFD sequence in fig. 6 (b) will result in the results in fig. 6 (d). As shown, the unwanted signal components have been removed, leaving only peaks associated with the simulated fault. A more pronounced enhancement using KPSA can be observed on the spectrum shown in fig. 6 (e), from which the interference related frequency components are completely eliminated, leaving only the FCF and its harmonics.
Interference suppression capability of the proposed STFD-KPSA method. In this section, the proposed method will be comprehensively evaluated in terms of interference suppression. In order to quantify the interference level of the signal mixture x (t), including the fault-induced pulses s (t) and the interference i (t), the signal-to-interference ratio may be defined as follows
Where T is the signal length, P s Is the power of the fault pulse, P i Is the power of the interference.
The detection performance of a method may be represented by the method using a three-dimensional map of the displayed FCF harmonic numbers. These two axes are related to two factors to be examined. In this case, these two factors are frequency and SIR, respectively. The dark level represents the level of energy, the darkest to brightest corresponding to the lowest to highest energy level. Fig. 7 shows a diagram of the detectability of the proposed method. The pulse signal is defined by equation (9), with an SNR of-7 dB, and the corresponding parameters are the same as those listed in Table 1.
The obtained detectability map shows that: the proposed method (a) is capable of handling signals with SIR as low as-33 dB, (b) for signals with SIR ranging from 0 to-25 dB, the proposed method can detect FCF and its third harmonic, and (c) detectability then deteriorates with decreasing SIR, it being seen that when SIR decreases from-25 dB to-33 dB, only FCF and the first harmonic are observed.
Therefore, the bearing fault feature extraction algorithm based on fractal dimension envelope demodulation provided by the invention is used for extracting bearing fault features by using STFD transformation on collected vibration signals according to the fractal dimension theory, and further reducing interference by using KPSA, so that the bearing fault features are extracted.
The present invention is also described in detail in this example in connection with experimental data:
the composite signal is represented by the equation: x (t) =s (t) +i (t) +n (t) (14)
Where s (t) is an analog pulse signal, i (t) represents vibration disturbance, and n (t) represents white gaussian noise. Using sinusoidal functionsMultiple disturbances from other electrical/mechanical components (e.g. unbalanced shaft and gear mesh) are simulated in this case eight disturbances are added, f i1 ...f i8 ＝1000Hz，500Hz，300Hz，150Hz，80Hz，30Hz，15Hz，8Hz，A i1 ...A i8 = 0.5,1,1.5,1,1,1,1,1.5, making SIR equal to-21 dB. In addition, background noise exists in the real vibration signal; therefore, gaussian white noise n (t) is added. The signal to noise ratio relative to the analog pulse signal is 7dB. Adding such a dry matterThe purpose of the disturbance is to simulate the real situation where disturbances and their harmonics may occur. A mixed signal x of 16000 sampled data points is plotted in fig. 8 (a), where the characteristics of the effect are severely masked by interference and noise. Fig. 8 (b) shows the hilbert envelope spectrum of the analog signal. As shown, the hilbert envelope spectrum is mainly determined by the frequency difference between the disturbances, and since the pulses are severely contaminated by disturbances and noise, frequency information related to FCF cannot be found. The proposed method is then applied, the results of which are shown in fig. 8 (c) (output of STFD method) and fig. 8 (d) (output of STFD-KPSA method). Spectral analysis was performed on the STFD sequence and the STFD-KPSA sequence, respectively, to obtain FIGS. 8 (e) and (f). It can be seen that the effect of the interference cannot be completely eliminated by using STFD alone, as shown in fig. 8 (e), in which the amplitude of the second frequency line associated with the interference difference is almost equivalent to the amplitude of the first FCF harmonic. This is because there are still multiple interference effects in the STFD representation. Fig. 8 (f) provides a clearer result in which FCF and its three harmonics are clearly identifiable because the use of the proposed KSPA algorithm further reduces the combined impact of interference. The above analysis shows that the demodulation result of STFD-KPSA is better than that of hilbert transform and STFD without KPSA shown in fig. 8 (b) and (e), respectively. Thus, it can be concluded that the proposed STFD-KPSA method is capable of extracting bearing failure characteristics and suppressing interference and noise without using pre-filtering, and that fig. 9, 10 and 11 are performance graphs of detecting side, bottom and inner ring failures using the failure detection method of the present invention, respectively, and that fig. 9, 10 and 11 (a) are original vibration signals; (b) a hilbert envelope spectrum of the original signal; (c) STFD-KPSA results of the original signal; (d) spectrum of STFD-KPSA results.
Referring to fig. 12, fig. 12 is a block diagram illustrating a device for extracting bearing fault characteristics according to an embodiment of the present invention; the specific apparatus may include:
the acquisition signal module 100 is used for acquiring a bearing vibration signal and converting the bearing vibration signal into a bearing vibration signal sequence;
the fractal dimension calculation module 200 is used for calculating fractal dimension values between adjacent bearing vibration signals by using a Katz method;
the extracting envelope module 300 is configured to track an impact position generated by a bearing fault between the vibration signals of each adjacent bearing according to the fractal dimension value, and extract an envelope curve of the impact position;
a calculation peak module 400 for calculating an average value and a standard deviation of each envelope, and calculating a peak value of the envelope signal based on the average value and the standard deviation;
and the interference suppression module 500 is configured to set a peak value of the envelope signal with a difference between the peak value and the standard deviation of the envelope signal being smaller than 1 to 1, and the remaining envelope signals are bearing fault characteristics.
The device for extracting bearing fault characteristics of the present embodiment is used to implement the foregoing method for extracting bearing fault characteristics, so that the specific implementation of the device for extracting bearing fault characteristics may be referred to the example portions of the foregoing method for extracting bearing fault characteristics, for example, the signal acquisition module 100, the fractal dimension calculation module 200, the envelope extraction module 300, and the interference suppression module 400, which are respectively used to implement steps S101, S102, S103, and S104 in the foregoing method for extracting bearing fault characteristics, so that the specific implementation thereof may refer to the description of the examples of the respective portions and will not be repeated herein.
The specific embodiment of the invention also provides equipment for extracting the bearing fault characteristics, which comprises the following steps: a memory for storing a computer program; and the processor is used for realizing the steps of the method for extracting the bearing fault characteristics when executing the computer program.
The specific embodiment of the invention also provides a computer readable storage medium, wherein the computer readable storage medium stores a computer program, and the computer program realizes the steps of the method for extracting the bearing fault characteristics when being executed by a processor.
In this specification, each embodiment is described in a progressive manner, and each embodiment is mainly described in a different point from other embodiments, so that the same or similar parts between the embodiments are referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant points refer to the description of the method section.
Those of skill would further appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative elements and steps are described above generally in terms of functionality in order to clearly illustrate the interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. The software modules may be disposed in Random Access Memory (RAM), memory, read Only Memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.
The method, the device, the equipment and the computer readable storage medium for extracting the bearing fault characteristics provided by the invention are described in detail. The principles and embodiments of the present invention have been described herein with reference to specific examples, the description of which is intended only to facilitate an understanding of the method of the present invention and its core ideas. It should be noted that it will be apparent to those skilled in the art that various modifications and adaptations of the invention can be made without departing from the principles of the invention and these modifications and adaptations are intended to be within the scope of the invention as defined in the following claims.
Claims (5)
1. A method of bearing fault signature extraction, comprising:
collecting a bearing vibration signal, and carrying out windowing interception on the bearing vibration signal;
calculating fractal dimension values of the intercepted bearing vibration signals by using a Katz method;
tracking the impact position generated by bearing faults in the bearing vibration signal according to the fractal dimension value, and extracting an envelope curve of the impact position; the method comprises the following steps: determining the width of a moving window according to the fractal dimension value; intercepting the bearing vibration signals from beginning to end by utilizing the moving window to obtain a plurality of signal segments; calculating the fractal dimension value of each bearing vibration section; splicing fractal dimension values of all signal segments to obtain the short-time fractal dimension sequence;
intercepting the bearing vibration signal from beginning to end by utilizing the moving window, and obtaining a plurality of signal segments comprises the following steps:
the matrix formed by the plurality of bearing vibration sections is as follows:
wherein sw is m For each signal segment, N win For window length, N win ＝int(αf s ) Int (-) is the integer part of the value, α is a preset constant, f s Is the sampling frequency of the waveform;
the step of splicing the fractal dimension values of all the signal segments to obtain the short-time fractal dimension sequence comprises the following steps:
calculating the fractal dimension value of each signal segment to obtain a short-time fractal dimension vector:
STFD＝[STFD(1),STFD(2),...,STFD(m),...,]；
converting it into the short-time fractal dimension sequence: STFD = STFD original -MIN+1；
Wherein the vector STFD original For the calculated signal STFD sequence, MIN is the sum STFD original The smallest component of the same length, 1 is the same as STFD original The unit vectors with the same length and STFD is the obtained STFD sequence;
calculating an average value and a standard deviation of each envelope, and calculating a peak value of an envelope signal based on the average value and the standard deviation;
setting the peak value of the envelope signal with the difference value of the peak value and the standard deviation smaller than 1 as 1, wherein the rest envelope signals are bearing fault characteristics;
setting the peak value of the envelope signal with the standard deviation difference less than 1 to be 1, wherein the rest envelope signals are bearing fault characteristics, and the method comprises the following steps:
the formula defining the peak search algorithm:
where μ is the average of the signal x (t), σ is the standard deviation of the signal x (t), E { } is the mathematical expectation operator, x (t) =s (t) +i (t) =a mp e -βt sinω r t+A i cosσ i t is; wherein A is mp For the amplitude, ω, of the vibration of the pulse r For the vibration frequency of the pulse A i To the amplitude of the disturbance, ω i Is the frequency of the disturbance;
calculating the standard deviation sigma and the peak value K of each signal;
judging whether the amplitude is greater than or equal to 1+sigma, if the amplitude is greater than or equal to 1+sigma, the amplitude is kept unchanged, and if the amplitude is less than 1+sigma, the amplitude is set to be 1;
judging whether the peak value difference of the two iterations meets a preset value, if so, stopping calculating the peak value, and if not, continuing the iterations;
stopping iteration until all signal calculation is completed, and obtaining the vibration fault characteristics of the bearing;
the peak search algorithm comprises the following calculation steps:
s71: initializing, let i=1,
s72: calculating the standard deviation sigma of the signal segment x (i) i And peak value K i ；
S73: judging whether the amplitude in the signal section is more than or equal to 1+sigma i ；
S74: if greater than or equal to 1+sigma i The amplitude remains unchanged; if less than 1+sigma i Let the amplitude set to 1;
s75: judging that i is more than or equal to N, and stopping iteration if the i is more than or equal to N;
s76: if the result is not true, the method comprises the steps of, then determine if |K is satisfied i -K i-1 And if the I is less than or equal to epsilon, stopping calculating K i -K i-1 Wherein ε is a threshold minimum constant;
s77: if not, i=i+1 is set to step S72.
2. The method of claim 1, wherein calculating the fractal dimension value of the intercepted bearing vibration signal using the Katz method comprises:
according to the bearing vibration signal s= { s 1 ,s 2 ,...,s N -calculating the euclidean distance between successive signals as:
wherein N is the total number of sampling points in the bearing vibration signal, s i Is any point in the vibration signal, s i The coordinates are (x) i ,y i ),i＝1,2,...,N；
According to the Euclidean distance between the continuous signals, calculating the plane range d of the waveform and the total length L of the curve, wherein the calculation formula is as follows:
normalizing the total length L of the curve and the planar extent d of the waveform by the average distance a between the successive signals and using the mathematical definition formula of KatzFDCalculating fractal dimension:
where k=n-1 is the number of steps in the curve, a=l/k.
3. A device for extracting bearing fault characteristics, comprising:
the signal acquisition module is used for acquiring bearing vibration signals and carrying out windowing interception on the bearing vibration signals;
the fractal dimension calculating module is used for calculating the fractal dimension value of the intercepted bearing vibration signal by using a Katz method;
the envelope extraction module is used for tracking the impact position generated by bearing faults in the bearing vibration signals according to the fractal dimension values and extracting an envelope of the impact position; determining the width of a moving window according to the fractal dimension value; intercepting the bearing vibration signals from beginning to end by utilizing the moving window to obtain a plurality of signal segments; calculating the fractal dimension value of each bearing vibration section; splicing fractal dimension values of all signal segments to obtain the short-time fractal dimension sequence;
intercepting the bearing vibration signal from beginning to end by utilizing the moving window, and obtaining a plurality of signal segments comprises the following steps:
the matrix formed by the plurality of bearing vibration sections is as follows:
wherein sw is m For each signal segment, N win For window length, N win ＝int(αf s ) Int (-) is the integer part of the value, α is a preset constant, f s Is the sampling frequency of the waveform;
the step of splicing the fractal dimension values of all the signal segments to obtain the short-time fractal dimension sequence comprises the following steps:
calculating the fractal dimension value of each signal segment to obtain a short-time fractal dimension vector:
STFD＝[STFD(1),STFD(2),...,STFD(m),...,]；
converting it into the short-time fractal dimension sequence: STFD = STFD original -MIN+1；
Wherein the vector STFD original For the calculated signal STFD sequence, MIN is the sum STFD original The smallest component of the same length, 1 is the same as STFD original The unit vectors with the same length and STFD is the obtained STFD sequence;
a calculation peak module for calculating an average value and a standard deviation of each envelope, and calculating a peak value of the envelope signal based on the average value and the standard deviation;
the interference suppression module is used for setting the peak value of the envelope signal with the peak value of which the standard deviation difference value is smaller than 1 to be 1, and the rest envelope signals are bearing fault characteristics; setting the peak value of the envelope signal with the standard deviation difference less than 1 to be 1, wherein the rest envelope signals are bearing fault characteristics, and the method comprises the following steps:
the formula defining the peak search algorithm:
where μ is the average of the signal x (t), σ is the standard deviation of the signal x (t), E { } is the mathematical expectation operator, x (t) =s (t) +i (t) =a mp e -βt sinω r t+A i cosω i t is; wherein A is mp For the amplitude, ω, of the vibration of the pulse r For the vibration frequency of the pulse A i To the amplitude of the disturbance, ω i Is the frequency of the disturbance;
calculating the standard deviation sigma and the peak value K of each signal;
judging whether the amplitude is greater than or equal to 1+sigma, if the amplitude is greater than or equal to 1+sigma, the amplitude is kept unchanged, and if the amplitude is less than 1+sigma, the amplitude is set to be 1;
judging whether the peak value difference of the two iterations meets a preset value, if so, stopping calculating the peak value, and if not, continuing the iterations;
stopping iteration until all signal calculation is completed, and obtaining the vibration fault characteristics of the bearing;
the peak search algorithm comprises the following calculation steps:
s71: initializing, let i=1,
s72: calculating the standard deviation sigma of the signal segment x (i) i And peak value K i ；
S73: judging whether the amplitude in the signal section is more than or equal to 1+sigma i ；
S74: if greater than or equal to 1+sigma i The amplitude remains unchanged; if less than 1+sigma i Let the amplitude set to 1;
s75: judging that i is more than or equal to N, and stopping iteration if the i is more than or equal to N;
s76: if the result is not true, the method comprises the steps of, then determine if |K is satisfied i -K i-1 And if the I is less than or equal to epsilon, stopping calculating K i -K i-1 Wherein ε is a threshold minimum constant;
s77: if not, i=i+1 is set to step S72.
4. An apparatus for bearing fault signature extraction, comprising:
a memory for storing a computer program;
a processor for implementing the steps of a method of bearing fault signature extraction as claimed in any one of claims 1 or 2 when executing said computer program.
5. A computer readable storage medium, characterized in that it has stored thereon a computer program which, when executed by a processor, implements the steps of a method of bearing fault feature extraction according to any of claims 1 or 2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211057305.7A CN115436058B (en) | 2022-08-30 | 2022-08-30 | Bearing fault feature extraction method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211057305.7A CN115436058B (en) | 2022-08-30 | 2022-08-30 | Bearing fault feature extraction method, device, equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115436058A CN115436058A (en) | 2022-12-06 |
| CN115436058B true CN115436058B (en) | 2023-10-03 |
Family
ID=84244733
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211057305.7A Active CN115436058B (en) | 2022-08-30 | 2022-08-30 | Bearing fault feature extraction method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115436058B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101251445A (en) * | 2008-04-16 | 2008-08-27 | 邓艾东 | Method for analysis of fractal characteristic of rotating machinery bump-scrape acoustic emission signal |
| CN110160767A (en) * | 2019-06-14 | 2019-08-23 | 安徽智寰科技有限公司 | Impulse period automatic identification and extracting method and system based on Envelope Analysis |
| CN114705426A (en) * | 2022-01-06 | 2022-07-05 | 杭州爱华仪器有限公司 | Early fault diagnosis method for rolling bearing |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| SE1000313A1 (en) * | 2010-03-30 | 2011-09-20 | Rubico Ab | Method for error detection of rolling bearings by increasing statistical asymmetry |
-
2022
- 2022-08-30 CN CN202211057305.7A patent/CN115436058B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101251445A (en) * | 2008-04-16 | 2008-08-27 | 邓艾东 | Method for analysis of fractal characteristic of rotating machinery bump-scrape acoustic emission signal |
| CN110160767A (en) * | 2019-06-14 | 2019-08-23 | 安徽智寰科技有限公司 | Impulse period automatic identification and extracting method and system based on Envelope Analysis |
| CN114705426A (en) * | 2022-01-06 | 2022-07-05 | 杭州爱华仪器有限公司 | Early fault diagnosis method for rolling bearing |
Non-Patent Citations (1)
| Title |
|---|
| 基于形态学分形的往复式压缩机气阀故障诊断方法;杨晓;中国优秀硕士学位论文全文数据库(第03期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115436058A (en) | 2022-12-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Cong et al. | Short-time matrix series based singular value decomposition for rolling bearing fault diagnosis | |
| CN109682601A (en) | The initial failure recognition methods of rolling bearing under a kind of variable speed operating condition | |
| Hu et al. | Bearing fault diagnosis based on an improved morphological filter | |
| He et al. | A joint adaptive wavelet filter and morphological signal processing method for weak mechanical impulse extraction | |
| RU2010120713A (en) | METHOD FOR DETECTING AND AUTOMATIC IDENTIFICATION OF DAMAGE TO ROLLING BEARINGS | |
| CN112101245A (en) | Short-time Fourier transform mechanical impact feature extraction method based on frequency domain window function | |
| RU2709414C1 (en) | Method of recording and analyzing acoustic emission signals in a system for diagnostic monitoring of production facilities | |
| CN111769810B (en) | Fluid mechanical modulation frequency extraction method based on energy kurtosis spectrum | |
| CN109932624B (en) | Cable partial discharge period narrow-band interference denoising method based on Gaussian scale space | |
| Liao et al. | An automatic filtering method based on an improved genetic algorithm—with application to rolling bearing fault signal extraction | |
| Bozchalooi et al. | Teager energy operator for multi-modulation extraction and its application for gearbox fault detection | |
| Jaber et al. | A simulation of non-stationary signal analysis using wavelet transform based on LabVIEW and Matlab | |
| Williams et al. | Helicopter transmission fault detection via time-frequency, scale and spectral methods | |
| Zhang et al. | Improved local cepstrum and its applications for gearbox and rolling bearing fault detection | |
| Zhen et al. | Rolling bearing fault diagnosis based on VMD reconstruction and DCS demodulation | |
| CN112098093A (en) | Bearing fault feature identification method and system | |
| Zheng et al. | Faults diagnosis of rolling bearings based on shift invariant K-singular value decomposition with sensitive atom nonlocal means enhancement | |
| CN114486263A (en) | Noise reduction and demodulation method for vibration signal of rolling bearing of rotary machine | |
| Xu et al. | Rolling bearing fault feature extraction via improved SSD and a singular-value energy autocorrelation coefficient spectrum | |
| Makowski et al. | Parametric time-frequency map and its processing for local damage detection in rotating machinery | |
| CN115436058B (en) | Bearing fault feature extraction method, device, equipment and storage medium | |
| Gong et al. | Fault detection for rolling element bearing based on repeated single-scale morphology and simplified sensitive factor algorithm | |
| Huang et al. | Short-time kurtogram for bearing fault feature extraction under time-varying speed conditions | |
| CN114061746B (en) | Repeated transient signal extraction method in rotary machinery fault diagnosis | |
| CN116519301A (en) | Bearing fault diagnosis method and system based on time-frequency envelope spectrum peak analysis |
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 |