Bearing fault diagnosis method based on adaptive chirp mode decomposition and order analysis

Abstract

The invention relates to the technical field of equipment monitoring, in particular to a bearing fault diagnosis method based on adaptive chirp mode decomposition and order analysis, and the method comprises the steps: firstly collecting and obtaining a vibration time sequence signal and a low-sampling-rate rotating speed time sequence signal in the operation process of a rotating machine; processing by adopting a parameter search adaptive chirp mode decomposition method, and acquiring a rotating speed component signal from the vibration data; processing the rotating speed component signal, firstly eliminating the amplitude influence, and then performing Hilbert transform on the rotating speed component signal without the amplitude influence to obtain an analytic signal of the rotating speed component signal; differentiating the phase of the obtained analytic signal to obtain an instantaneous frequency curve of the rotating speed component signal; performing angle resampling on the original vibration signal by using the instantaneous frequency curve to obtain an angular domain signal of the vibration signal; and finally, Fourier transform is performed on the angular domain signal to obtain an order spectrum of the angular domain signal, and whether a fault occurs is judged by using a corresponding order in the order spectrum.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment monitoring, and in particular to a bearing fault diagnosis method based on adaptive chirp modal decomposition and order analysis. Background Art

[0002] In recent years, with the further development of theoretical technologies such as mathematics, engineering, and computers, bearing fault diagnosis methods based on signal processing have emerged in an endless stream, providing a theoretical basis for bearing fault diagnosis. However, in actual operation, the bearing speed is generally not constant, but changes according to the operating status of the equipment. Therefore, the order tracking (OT) method has emerged to identify bearing faults under variable speed. The OT method resamples the signal at a fixed angle increment so that the newly sampled angle signal can eliminate the influence of the speed change, thereby realizing fault feature extraction and fault diagnosis. Order tracking technology is generally divided into hardware order tracking (HOT), computed order tracking (COT), and tacho-less order tracking (TLOT). The HOT method uses high-precision phase tracking hardware for real-time speed estimation, which is costly and complex to install. TLOT estimates the speed directly from the vibration signal or other collected signals and then performs order tracking. Signal coupling, equipment noise, initial parameter settings, etc. may affect the accuracy of the estimated speed. The COT technology uses a tachometer and vibration signals to perform order analysis, achieving a good balance between cost and accuracy, and is widely used in bearing fault diagnosis under variable speed.

[0003] The calculation of speed is the core issue of COT technology. Usually, COT needs to use a speed sensor to collect speed signals and convert them into speed curves. However, in general vibration monitoring situations, the sampling frequency of speed data is much lower than the sampling frequency of vibration signals, resulting in the speed data obtained by the tachometer being much lower than the vibration data, which makes it difficult to achieve high-precision order tracking. Summary of the invention

[0004] The purpose of the present invention is to provide a bearing fault diagnosis method based on adaptive chirp modal decomposition and order analysis. By utilizing parameter search adaptive chirp modal decomposition method, a speed curve with denser sample points is estimated from a small number of speed sample points combined with vibration signals, thereby achieving high-precision order tracking.

[0005] To achieve the above object, the present invention provides a bearing fault diagnosis method based on adaptive chirp modal decomposition and order analysis, comprising the following steps:

[0006] Step 1: Use a vibration acceleration sensor and a tachometer to collect signals to obtain a vibration time series signal and a speed time series signal;

[0007] Step 2: Input the vibration acceleration time series signal and the speed time series signal into the parameter search adaptive chirp mode decomposition algorithm to obtain the optimal speed component signal;

[0008] Step 3: Processing the optimal speed component signal, the processing process includes eliminating the amplitude modulation of the speed component, performing Hilbert transform on the speed component signal after the amplitude modulation is eliminated, and calculating the instantaneous frequency of the analytical signal;

[0009] Step 4: Convert the instantaneous frequency obtained into a speed curve, use the speed curve to resample the original vibration signal in the angular domain to obtain an equal-angle signal, perform Fourier transform on the equal-angle signal to obtain an order spectrum, and realize fault feature extraction and diagnosis.

[0010] Optionally, in step 1, a vibration acceleration sensor is arranged at the measuring point as needed, and a tachometer is arranged at the coupling end of the output shaft to synchronously collect vibration acceleration time series signal data and speed time series signal data.

[0011] Optionally, the parameter search adaptive chirp modal decomposition algorithm in step 2 uses sample entropy to calculate the optimal rotational speed component signal.

[0012] Optionally, the execution process of step 3 is specifically to use signal processing to remove the signal amplitude modulation to obtain a pure FM signal, then perform Hilbert transform on the obtained pure FM signal to obtain the analytical signal of the pure FM signal, and finally process the analytical signal to obtain the instantaneous frequency curve of the analytical signal, wherein the instantaneous frequency curve is a high-precision speed curve extracted from the vibration signal.

[0013] Optionally, the execution process of step 4 includes the following steps:

[0014] Step 4.1: Convert the instantaneous frequency sequence into a speed sequence;

[0015] Step 4.2: Perform order tracking calculation;

[0016] Step 4.3: Perform order spectrum analysis, compare the calculated order spectrum with the fault order, and perform fault diagnosis.

[0017] The present invention provides a bearing fault diagnosis method based on adaptive chirp modal decomposition and order analysis, wherein a vibration acceleration sensor is installed at a key position of the machine to be measured, and a tachometer is installed at a rotating shaft position, and a vibration time series signal and a low sampling rate speed time series signal during the operation of the rotating machine are collected and obtained; then, a parameter search adaptive chirp modal decomposition method is used to process the speed component signal from the above vibration data; then, the speed component signal is processed, firstly, the amplitude influence is eliminated, and then the speed component signal with the amplitude influence eliminated is subjected to Hilbert transformation to obtain an analytical signal of the speed component signal; the phase of the analytical signal obtained is differentiated to obtain an instantaneous frequency curve of the speed component signal, which is a high-precision speed curve; the original vibration signal is angle-resampled using the high-precision speed curve to obtain an angular domain signal of the vibration signal; finally, the angular domain signal is Fourier transformed to obtain an order spectrum of the angular domain signal, and the corresponding order in the order spectrum is used to determine whether a fault occurs. The present invention estimates a speed curve with denser sample points from a small number of speed sample points combined with a vibration signal to achieve high-precision order tracking. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0019] Figure 1 It is a schematic flow chart of the steps of a bearing fault diagnosis method based on adaptive chirp modal decomposition and order analysis of the present invention.

[0020] Figure 2 It is a schematic diagram of obtaining vibration signals and rotation speed signals according to a specific embodiment of the present invention.

[0021] Figure 3 It is the flow chart of parameter search adaptive chirp mode decomposition algorithm.

[0022] Figure 4 It is the optimal parameter search result of three bearing faults in the specific embodiment of the present invention.

[0023] Figure 5 It is a flow chart of obtaining the instantaneous frequency curve according to a specific embodiment of the present invention.

[0024] Figure 6 It is the angular domain signal obtained by calculating the three bearing faults in the specific embodiment of the present invention.

[0025] Figure 7It is an outer ring fault order diagram obtained by combining three methods in a specific embodiment of the present invention.

[0026] Figure 8 It is an inner ring fault order diagram obtained by combining three methods in a specific embodiment of the present invention.

[0027] Fig. 9 It is a rolling element fault order diagram obtained by combining three methods in a specific embodiment of the present invention. DETAILED DESCRIPTION

[0028] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and should not be construed as limiting the present invention.

[0029] See also Figure 1 The present invention provides a bearing fault diagnosis method based on adaptive chirp modal decomposition and order analysis, comprising the following steps:

[0030] S1: Use a vibration acceleration sensor and a tachometer to collect signals to obtain a vibration time series signal and a speed time series signal;

[0031] S2: Input the vibration acceleration time series signal and the speed time series signal into the parameter search adaptive chirp mode decomposition algorithm to obtain the optimal speed component signal;

[0032] S3: processing the optimal speed component signal, the processing process includes eliminating the amplitude modulation of the speed component, performing Hilbert transform on the speed component signal after the amplitude modulation is eliminated, and calculating the instantaneous frequency of the analytical signal;

[0033] S4: converting the instantaneous frequency obtained into a speed curve, using the speed curve to resample the original vibration signal in the angular domain to obtain an equal-angle signal, performing Fourier transform on the equal-angle signal to obtain an order spectrum, and realizing fault feature extraction and diagnosis.

[0034] The present invention aims to estimate a speed curve with denser sample points from a small number of speed sample points combined with vibration signals by using a parameter search adaptive chipmode decomposition (ACMD) method to achieve high-precision order tracking.

[0035] See also Figures 2 to 9 The present invention will be further described below in conjunction with specific embodiments:

[0036] This embodiment uses the subway train drive motor bearing as an example, and pre-sets three types of bearing faults, all of which are processed by wire cutting.

[0037] The preset faults are bearing inner ring fault, bearing outer ring fault, and bearing rolling element fault.

[0038] The vibration acceleration sensor is placed horizontally on the upper end of the bearing seat of the faulty bearing, and a speed sensor is used at the motor output shaft to obtain the time series of the speed change.

[0039] The sampling frequency of the vibration acceleration sensor used in this embodiment is 25600 Hz.

[0040] The sampling frequency of the speed sensor used in this embodiment is 1 Hz.

[0041] In this embodiment, the speed changes are increasing speed, constant speed, and decreasing speed. The speed change time is 30 seconds, which is 10 seconds for increasing speed, 10 seconds for constant speed, and 10 seconds for decreasing speed.

[0042] In this embodiment, the maximum rotation speed is set to 1500 revolutions per minute (1500 rpm).

[0043] The specific implementation steps are as follows:

[0044] Step S1 is vibration time series signal acquisition and speed time series signal acquisition;

[0045] Specifically, based on the arranged vibration acceleration sensor and tachometer, the vibration acceleration time series and speed time series generated when the train drive motor is running are obtained; the speed series is converted into a corresponding frequency curve according to the frequency of 1 Hz equivalent to 1 revolution per second;

[0046] Specifically, the vibration acceleration time series obtained is X(t); where t represents the tth sampling point of the vibration acceleration series;

[0047] Specifically, the obtained speed time series is T(t), where t represents the tth sampling point of the acceleration time series;

[0048] Specifically, the vibration signal and the speed signal obtained in this embodiment are as follows: Figure 2 As shown;

[0049] Specifically, T(t) is in revolutions per minute and is converted into a frequency curve F(t) according to Formula 1;

[0050]

[0051] Step S2, extracting the rotation speed component by using a parameter search adaptive chirp modal decomposition algorithm;

[0052] The flow chart of the parameter search adaptive chirp mode decomposition algorithm is as follows: Figure 3 As shown;

[0053] Step S2 includes the following execution steps:

[0054] S21 adaptive chirp mode decomposition algorithm initialization;

[0055] Specifically, there are three key parameters in the adaptive chirp modal decomposition that need to be set in advance, namely α, β and υ; α is a signal parameter that controls the smoothness of the output signal; β is a frequency parameter that controls the smoothness of the output instantaneous frequency curve; υ is the termination threshold that controls the number of iterations of the adaptive chirp modal decomposition algorithm;

[0056] Specifically, in this embodiment, the initial values ​​of α, β and υ are all set to 1e-15, the search interval is set to 1e-15 to 1e-1, and the search step length Δl is set to 1e-15;

[0057] S22 executes an adaptive chirp mode decomposition algorithm;

[0058] Specifically, the vibration acceleration time series X(t) and frequency curve F(t) obtained in the previous step are used as input quantities, and the parameter search adaptive chirp mode decomposition algorithm is introduced for calculation;

[0059] Specifically, the parameters that are changed in the parameter search adaptive chirp modal decomposition are α, β and υ, and the changing method of the above-mentioned changing parameters is carried out according to formula (2);

[0060]

[0061] Specifically, the rotation speed component is R m (t), where m represents the mth speed component;

[0062] Specifically, for the rotation speed component R m (t) Calculate the sample entropy value. The sample entropy value is calculated in a conventional manner and will not be described in detail herein.

[0063] Specifically, the samples are sorted according to their entropy values, and the speed component {R m (t)} min ;

[0064] Specifically, the optimal parameter search results for the three faults are as follows: Figure 4 As shown;

[0065] Specifically, the optimal speed component signal {R m (t)} min The expression is

[0066] Step S3 is a process of processing the optimal speed signal component and obtaining an instantaneous frequency curve.

[0067] Specifically, the process of obtaining the instantaneous frequency curve is as follows: Figure 5 As shown;

[0068] Specifically, the amplitude modulation effect of the optimal speed signal obtained above is eliminated, and the optimal speed component signal {R m (t)} min The expression is The amplitude A has been calculated in S2, and the amplitude influence is eliminated according to formula (3);

[0069]

[0070] Specifically, the optimal speed signal R after eliminating the amplitude effect is obtained. new (t) performing Hilbert transform to obtain an analytical signal;

[0071] Specifically, the instantaneous phase is easily obtained by analyzing the signal, and the instantaneous frequency is obtained by differentiating the instantaneous phase;

[0072] In step S4, speed extraction, order tracking and fault diagnosis are implemented, and the corresponding steps are as follows:

[0073] S41 converts the instantaneous frequency sequence into a speed sequence;

[0074] Specifically, according to the frequency of 1 Hz being equivalent to 1 revolution per second, the instantaneous frequency sequence obtained by S3 is converted into a rotation speed sequence, and the instantaneous frequency sequence is converted into a rotation speed sequence and expressed as r(t);

[0075] S42 performs order tracking calculation;

[0076] Specifically, f s1 For a known vibration signal, the frequency is adopted and the order tracks the maximum analysis order D max Calculate according to formula (4);

[0077] D max =f s1 / 2r(t)(4)

[0078] Specifically, the equal-angle sampling interval Δθ is calculated according to formula (5);

[0079] Δθ=π / D max (5)

[0080] Specifically, according to the instantaneous rotation speed at all sampling moments, the instantaneous phase of the reference axis is calculated according to formula (6):

[0081]

[0082] Specifically, the sampling time corresponding to the equal-angle resampling is solved according to formula (7);

[0083] T n (v) = Φ -1 (vΔθ)(7)

[0084] In formula (7),

[0085] In formula (7), Φ -1 for The inverse function of

[0086] Specifically, using the interpolation value T n The corresponding amplitude completes the acquisition of the angular domain signal;

[0087] Specifically, the angular domain signals calculated for the three faults are as follows: Figure 6 As shown;

[0088] S43 performs order spectrum analysis;

[0089] Specifically, the angular domain signal obtained in S42 is subjected to a fast Fourier transform to obtain an order spectrum;

[0090] Specifically, fault diagnosis is achieved based on the correspondence analysis between the calculated order spectrum and the fault order;

[0091] Specifically, in this embodiment, the outer ring fault order is 7.7, the inner ring fault order is 10.2, and the rolling element fault order is 14.5;

[0092] Furthermore, the present invention also uses another three method combinations to compare with the present invention to reflect the superiority of the method of the present invention;

[0093] Specifically, the final fault classification results are shown in Figure 7 , Figure 8 and Fig. 9 In the following, we will further explain the various methods in the figure:

[0094] The other three comparison method combinations used are short-time Fourier transform (STFT) and adaptive chirp modal decomposition (STFT+ACMD), synchronized squeezing transform (SST) and adaptive chirp modal decomposition (SST+ACMD), and synchronized extraction transform (SET) and adaptive chirp modal decomposition (SET+ACMD).

[0095] Specifically, STFT, SST and SET use time-frequency methods to extract the rotational speed and then import it into adaptive chirp modal decomposition calculation. It is difficult to obtain accurate rotational speed information in the case of noise, which leads to low accuracy of subsequent order analysis. The method of the present invention directly uses low sampling rate rotational speed signals for calculation, which overcomes this shortcoming. In addition, the designed parameter search adaptive chirp modal decomposition method can search for the optimal result, further improve the accuracy of rotational speed acquisition, enable high-precision order analysis, and thus obtain good fault diagnosis effect.

[0096] What is disclosed above is only a preferred embodiment of the present invention, and it certainly cannot be used to limit the scope of rights of the present invention. Ordinary technicians in this field can understand that all or part of the processes of the above embodiment and equivalent changes made according to the claims of the present invention still fall within the scope of the invention.

Claims

1. A bearing fault diagnosis method based on adaptive chirp modal decomposition and order analysis, characterized in that: The following steps are involved: Step 1: Use a vibration acceleration sensor and a tachometer to collect signals to obtain a vibration time series signal and a speed time series signal; Step 2: Input the vibration acceleration time series signal and the speed time series signal into the parameter search adaptive chirp mode decomposition algorithm to obtain the optimal speed component signal; Step 3: Processing the optimal speed component signal, the processing process includes eliminating the amplitude modulation of the speed component, performing Hilbert transform on the speed component signal after the amplitude modulation is eliminated, and calculating the instantaneous frequency of the analytical signal; Step 4: Convert the instantaneous frequency obtained into a speed curve, use the speed curve to resample the original vibration signal in the angular domain to obtain an equal-angle signal, perform Fourier transform on the equal-angle signal to obtain an order spectrum, and realize fault feature extraction and diagnosis.

2. The bearing fault diagnosis method based on adaptive chirp modal decomposition and order analysis according to claim 1, characterized in that: In step 1, a vibration acceleration sensor is arranged at the measuring point as needed, and a tachometer is arranged at the coupling end of the output shaft to synchronously collect vibration acceleration time series signal data and speed time series signal data.

3. The bearing fault diagnosis method based on adaptive chirp modal decomposition and order analysis according to claim 2, characterized in that: The parameter search adaptive chirp modal decomposition algorithm in step 2 uses sample entropy to calculate the optimal speed component signal.

4. The bearing fault diagnosis method based on adaptive chirp modal decomposition and order analysis according to claim 3 is characterized in that: The execution process of step 3 is specifically to use signal processing to remove the signal amplitude modulation to obtain a pure FM signal, then perform Hilbert transform on the obtained pure FM signal to obtain the analytical signal of the pure FM signal, and finally process the analytical signal to obtain the instantaneous frequency curve of the analytical signal, where the instantaneous frequency curve is a high-precision speed curve extracted from the vibration signal.

5. The bearing fault diagnosis method based on adaptive chirp modal decomposition and order analysis according to claim 4, characterized in that: The execution process of step 4 includes the following steps: Step 4.1: Convert the instantaneous frequency sequence into a speed sequence; Step 4.2: Perform order tracking calculation; Step 4.3: Perform order spectrum analysis, compare the calculated order spectrum with the fault order, and perform fault diagnosis.

Images