Rolling bearing fault detection method based on dual-optical encoder synchronized time difference

Abstract

This invention discloses a rolling bearing fault detection method based on synchronous time difference of dual optical encoders. The method acquires pulse signals from an external encoder and a motor encoder respectively, obtains the zero-crossing times of the two sets of pulse signals using the zero-crossing time method, and performs angle downsampling processing on the zero-crossing time of the motor encoder. A cross-correlation algorithm is used to compensate for the delay of the zero-crossing times of the external encoder and the downsampled zero-crossing times of the motor encoder, reconstructing the zero-crossing times of both as angle-time signals. The two reconstructed angle-time signals are then differentially processed to obtain the residual signal. The residual signal is biased and compensated, and then inverted to construct an instantaneous angular velocity signal. Order spectrum analysis is performed on the ω signal to extract bearing fault features. This invention can solve the problem of rolling bearing fault feature signals being masked by velocity trend components and quantization errors, enabling rolling bearing fault detection under variable speed conditions.

Description

Technical Field

[0001] This invention relates to a rolling bearing fault detection method based on synchronous time difference of dual optical encoders, belonging to the field of sensor technology and fault diagnosis technology. Background Technology

[0002] Time-varying speed conditions are common operating conditions for mechanical equipment, such as electric vehicles, wind turbines, and mine hoists. Rolling bearings, as crucial supporting components of rotating machinery, directly impact the operational accuracy, efficiency, and lifespan of the equipment. Therefore, fault diagnosis of rolling bearings under time-varying speed conditions has become a research hotspot in the field of fault diagnosis. Incremental optical encoders (IAS) are widely used in servo motors, industrial robots, and other fields. The IAS signals they measure contain a wealth of information. When a bearing fails, the contact stiffness between the rolling elements and raceways at the fault location changes, resulting in a regular change in the corresponding IAS signal. Compared to traditional vibration signals, IAS signals offer advantages such as low noise, direct correlation with machine dynamics, no need for periodic calibration, and short transmission paths. Therefore, bearing condition monitoring and fault diagnosis based on IAS signals are a hot topic in the field of fault diagnosis. However, under time-varying speed conditions, the fault components in the IAS signals of rolling bearings measured by encoders are often affected by averaging of velocity trend components and discretization errors, making fault feature extraction difficult. Although some research has been conducted on high-precision estimation of IAS signals, there are few reports on high-precision estimation of IAS signals under variable speed conditions, and no research has been conducted on the interference of velocity components on rolling bearing fault detection under variable speed conditions.

[0003] Against this background, this paper proposes a method for instantaneous angular velocity estimation and rolling bearing fault feature detection under variable speed conditions based on synchronous time difference of dual optical encoders. Using this method, the new IAS signal is not affected by the speed trend component and the interference of quantization error is reduced. In addition, this method does not rely excessively on the encoder resolution and is not limited by parameter settings. Summary of the Invention

[0004] In variable speed conditions, the components of interest in the IAS signal are often masked by the speed trend component. Furthermore, while directly removing the trend component from the original IAS signal can separate the speed trend to some extent, the signal amplitude still exhibits modulation, thus weakening the identifiability of fault characteristics. Since the angle between adjacent encoder pulses is always fixed, the dynamic characteristics of the IAS signal are entirely dominated by the change in Δt; that is, the IAS signal essentially depends only on the reciprocal of the time interval. Therefore, an IAS estimation method based on synchronous time difference is proposed and applied to fault detection in rolling bearings.

[0005] The present invention provides a method for instantaneous angular velocity estimation under variable speed conditions and rolling bearing fault feature detection based on the synchronous time difference of dual optical encoders, as follows:

[0006] Step 1: Collect the pulse signals from the external encoder and the motor encoder respectively, use the zero-crossing time method to obtain the zero-crossing time of the two sets of pulse signals, and perform angle downsampling processing on the zero-crossing time of the motor encoder.

[0007] Accurate detection of the rising edge of the optical encoder pulse signal is the core issue in IAS signal inversion. The fundamental problem is that the time resolution is difficult to synchronize with the rising / falling edge timing, leading to quantization errors in the acquired IAS signal. Increasing the sampling frequency of the high-speed counter increases hardware costs, while using zero-crossing time effectively improves sampling accuracy. The specific steps are as follows:

[0008] Assume the transition interval at the rising edge is [c(j), c(j+1)], and the corresponding voltage amplitude is [y(j), y(j+1)], where y(j)... <V th ,y(j+1)>V th V th = (y(j) + y(j+1)) / 2;

[0009] The zero-crossing time is calculated using linear interpolation, and the formula is as follows:

[0010]

[0011] In the formula, f c This indicates the sampling frequency of the high-speed counter.

[0012] The angle downsampling processing performed on the zero-crossing moment of the motor encoder can be represented as follows:

[0013]

[0014] In the formula, D is the downsampling factor; n = 1, D+1, 2D+1, ...;

[0015] The external encoder is installed at the end of the shaft adjacent to the bearing to be tested.

[0016] Step 2: Use the cross-correlation algorithm to perform delay compensation on the zero-crossing times of the external encoder and the downsampled zero-crossing times of the motor encoder, and reconstruct the zero-crossing times of the external encoder and the downsampled zero-crossing times of the motor encoder as angle-time signals; the specific steps are as follows:

[0017] (1) The misalignment time matrix T of the external encoder is obtained by the following formula. o ;

[0018] T o [i] = [T PT … PK-2 TP K-1 T]

[0019]

[0020] In the formula, T o Let P be the external encoder misalignment time matrix; T represents the zero-crossing time of the external encoder; P is the permutation matrix; K is the array length of T; i = 1, 2, ..., K;

[0021] (2) Use the Pearson correlation coefficient to estimate T o With T m The correlation between them is expressed as follows:

[0022]

[0023] In the formula, T m The zero-crossing time of the motor encoder; cov(T) o ,T m ) represents T o With T m The covariance; σ(·) is the standard deviation;

[0024] It is worth noting the correlation coefficient ρ i The larger T is o With T m The better the correlation between them, that is, the smaller the zero-crossing delay between the external encoder and the motor encoder.

[0025] (3) The maximum correlation coefficient MC is obtained through the following expression:

[0026] MC = max(ρ i )

[0027] In the formula, the maximum correlation coefficient MC is the maximum correlation coefficient among different misaligned arrays;

[0028] (4) Obtain the delay phase τ using the following formula:

[0029] τ = argmax(MC)

[0030] In the formula, argmax(·) represents the index parameter at which the function reaches its maximum value, that is, the value of τ when the maximum MC is reached;

[0031] (5) The zero-crossing time of the reconstructed external encoder and the zero-crossing time of the motor encoder after downsampling are angle-time signals, respectively, expressed as:

[0032]

[0033] In the formula, Δt oIt is the angle-time signal reconstructed by the external encoder, Δt m It is the angle-time signal reconstructed from the motor encoder; H o (·) represents the function indicating the zero-crossing time of the external encoder, H m (·) represents a function indicating the zero-crossing time of the motor encoder, T i Let K be the array length of T at the i-th zero-crossing moment; i = 1, 2, ..., K;

[0034] Step 3: Perform differential processing on the two sets of reconstructed angle-time signals to obtain the residual signal. Then, offset and compensate the residual signal and invert it to construct the instantaneous angular velocity signal. The specific steps are as follows:

[0035] (1) The remaining time signal Δt containing fault information is calculated using the following formula. s ;

[0036] Δt s =Δt o -Δt m

[0037] (2) The remaining time signal Δt is compensated by the bias coefficient a. s , to obtain the instantaneous angular velocity signal ω after removing the velocity trend component;

[0038]

[0039] In the formula: N is the number of encoder lines;

[0040] Step 4: Perform order spectrum analysis on the ω signal to extract bearing fault features.

[0041] Advantages and technical effects of the present invention:

[0042] 1. The IAS signal obtained by the method of this invention will not be affected by the velocity trend component and will reduce the interference of quantization error;

[0043] 2. This method does not overly rely on the encoder resolution and is not limited by parameter settings;

[0044] 3. The dual encoder synchronous time difference proposed in this invention can solve the problem that the rolling bearing fault characteristic signal is masked due to the speed trend component and the quantization error, and finally realize the rolling bearing fault detection under variable speed conditions. Attached Figure Description

[0045] Figure 1 This is a schematic diagram of the bearing testing bench used in Example 1;

[0046] Figure 2 For external encoders and motor encoders, the pulse signals are used.

[0047] Figure 3 for Figure 2 A magnified view of a portion of the image;

[0048] Figure 4 The zero-crossing moments of the external encoder and the reconstructed angle-time signal diagrams of the zero-crossing moments after downsampling of the motor encoder are shown.

[0049] Figure 5 The remaining time signal Δt s picture

[0050] Figure 6 ω diagram of instantaneous angular velocity signal

[0051] Figure 7 for Figure 6 A magnified view of a portion of the image;

[0052] Figure 8 The order spectrum of instantaneous angular velocity;

[0053] Figure 9 The data represents the traditional velocity differential experiment signal, where figure a is the instantaneous angular velocity diagram, figure b is a magnified view of figure a, figure c is the magnitude diagram of the fault impact at different speeds, and figure d is the order spectrum of the instantaneous angular velocity. Detailed Implementation

[0054] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Unless otherwise specified, the methods in this embodiment are conventional methods.

[0055] Example 1: Extraction of actual rolling bearing outer ring fault features using the method of the present invention

[0056] In this embodiment, a bearing testing platform is used, such as... Figure 1 As shown, a ZKT8025-002J-2500BZ3-5-24F incremental optical encoder (external encoder) is installed at the end of the shaft of the bearing to be tested on the experimental platform. The encoder line count N = 2500, and a sampling rate of 10 is used. 6 The PicoScope high-speed acquisition device acquires pulse signals from the external encoder and the motor encoder. The motor has a built-in K8025G-10000BM-K526 incremental optical encoder with a line count N' = 10000. The bearing type of this test bench is 6205-2RSJEMSKF(N b =9,E b =7.94, Ep =39.04, α=0), grooves with a width of approximately 0.8mm and a depth of approximately 0.28mm are machined on the outer ring of the bearing using wire cutting; the characteristic frequency f of the bearing outer ring failure is obtained by the following formula. reb It is 3.58×.

[0057]

[0058] In the formula, N b E represents the number of balls. b E represents the diameter of the ball bearing. p Where α is the pitch diameter of the rolling bearing, and α is the contact angle.

[0059] Step 1: Use the PicoScope high-speed acquisition device to acquire the pulse signals of the external encoder and the motor encoder respectively. The obtained signals are as follows: Figure 2 As shown;

[0060] The zero-crossing times of the two sets of pulse signals are obtained using the zero-crossing time method;

[0061] like Figure 3 As shown, assume the transition interval at the rising edge is [c(j), c(j+1)], and the corresponding voltage amplitude is [y(j), y(j+1)], and y(j)... <V th ,y(j+1)>V th V th = (y(j) + y(j+1)) / 2;

[0062] The pulse signal voltage amplitude range is 0-20V, and the threshold value is set to V. th =10;

[0063] The zero-crossing time t is calculated using linear interpolation. i The calculation formula is:

[0064]

[0065] In the formula: f c f represents the sampling frequency of the high-speed counter. c =10 6 V th =10, j=1,2,3…;

[0066] The zero-crossing moments of the motor encoder are sampled at an angle downsampling rate, and the corresponding moments can be represented as follows:

[0067]

[0068] In the formula, D is the downsampling factor, D = 4; n = 1, D+1, 2D+1, ...;

[0069] Step 2: Use the cross-correlation algorithm to perform delay compensation on the zero-crossing times of the external encoder and the downsampled zero-crossing times of the motor encoder, and reconstruct the zero-crossing times of the external encoder and the downsampled zero-crossing times of the motor encoder as angle-time signals; the specific steps are as follows:

[0070] (1) The misalignment time matrix T of the external encoder is obtained by the following formula. o ;

[0071] T o [i] = [T PT … P K-2 TP K-1 T]

[0072]

[0073] In the formula, T o Let P be the external encoder misalignment time matrix; T represents the zero-crossing time of the external encoder; P is the permutation matrix; K is the array length of T; i = 1, 2, ..., K;

[0074] (2) Use the Pearson correlation coefficient to estimate T o With T m The correlation between them is expressed as follows:

[0075]

[0076] In the formula, T m The zero-crossing time of the motor encoder; cov(T) o ,T m ) represents T o With T m The covariance; σ(·) is the standard deviation;

[0077] It is worth noting the correlation coefficient ρ i The larger T is o With T m The better the correlation between them, that is, the smaller the zero-crossing delay between the external encoder and the motor encoder.

[0078] (3) The maximum correlation coefficient MC is obtained through the following expression:

[0079] MC = max(ρ i )

[0080] In the formula, the maximum correlation coefficient MC is the maximum correlation coefficient among different misaligned arrays;

[0081] (4) Obtain the delay phase τ using the following formula:

[0082] τ = argmax(MC)

[0083] In the formula, argmax(·) represents the index parameter at which the function reaches its maximum value, that is, the value of τ when the maximum MC is reached;

[0084] (5) The zero-crossing time of the reconstructed external encoder and the zero-crossing time of the motor encoder after downsampling are angle-time signals, respectively, expressed as:

[0085]

[0086] In the formula, Δt o It is the angle-time signal reconstructed by the external encoder, Δt m It is the angle-time signal reconstructed from the motor encoder; H o (·) represents the function indicating the zero-crossing time of the external encoder, H m (·) represents a function indicating the zero-crossing time of the motor encoder, T i Let K be the array length of T at the i-th zero-crossing time; i = 1, 2, ..., K; the reconstructed signal is as follows: Figure 4 As shown.

[0087] Step 3: Perform differential processing on the two sets of reconstructed angle-time signals to obtain the residual signal. Then, offset-compensate the residual signal and invert it to construct the instantaneous angular velocity signal. The specific steps are as follows:

[0088] (1) The remaining time signal Δt containing fault information is calculated using the following formula. s ,like Figure 5 As shown in the figure, the fault impact can be effectively identified in the remaining time signal, but the impact amplitude is relatively small.

[0089] Δt s =Δt o -Δt m

[0090] (2) The remaining time signal Δt is compensated by the bias coefficient a. s To obtain the instantaneous angular velocity signal ω after removing the velocity trend component; such as Figure 6 As shown in the figure, the impact amplitude of ω is effectively increased. Figure 7 for Figure 6 The magnified view shows a peak value of 331.87.

[0091]

[0092] In the formula: N is the number of encoder lines;

[0093] Step 4: Perform order spectrum analysis on the ω signal to extract bearing fault features, such as... Figure 8 As shown in the figure, it can be seen from the order spectrum that the rolling bearing fault order of 3.6 can be effectively identified.

[0094] To verify the effectiveness of this method, the IAS obtained by the traditional velocity difference method is compared:

[0095]

[0096] Traditional velocity differential IAS such as Figure 9 As shown in Figure a, compared with the traditional velocity difference method, the fluctuation amplitude of the fault signal is larger, such as... Figure 9 As shown in b, the peak-to-peak value of the fault signal at low speeds is only 0.67, which is nearly 5000 times smaller than the peak-to-peak value of the IAS estimated by the method of this invention. Furthermore, the amplitude of the fault signal sampled using traditional speed differential is affected by the rotational speed, such as... Figure 9 As shown in Figure c, the higher the rotational speed, the greater the amplitude. From Figure 9 In the current method, the order of rolling bearing fault characteristics cannot be effectively identified. Therefore, the proposed method can effectively enhance the outer ring fault component of the rolling bearing in the IAS signal, thereby achieving rolling bearing fault characteristic identification.

Claims

1. A method for detecting a fault of a rolling bearing based on a dual-optical-encoder synchronous time-difference, characterized in that, The steps are as follows: (1) Collect the pulse signals of the external encoder and the motor encoder respectively, use the zero-crossing time method to obtain the zero-crossing time of the two sets of pulse signals, and perform angle downsampling processing on the zero-crossing time of the motor encoder. (2) The following formula is used to obtain the dislocation time matrix T of the external encoder o ; T o [i] = [T PT... P K-2 T P K-1 T] In the formula, T o Let P be the external encoder misalignment time matrix; T represents the zero-crossing time of the external encoder; P is the permutation matrix; K is the array length of T; i = 1, 2, ..., K; (3) Estimate T with Pearson correlation coefficient o correlation between T m , expressed as follows: where T m is the zero-crossing time of the motor encoder; cov(T o ,T m ) denotes the covariance of T o and T m ; and σ(·) is the standard deviation. (4) The maximum correlation coefficient MC is obtained through the following expression: MC = max(p i ) In the formula, the maximum correlation coefficient MC is the maximum correlation coefficient among different misaligned arrays; (5) The delay phase τ is obtained by the following formula: τ = argmax(MC) In the formula, argmax(·) represents the index parameter at which the function reaches its maximum value, that is, the value of τ when the maximum MC is reached; (6) The zero-crossing time of the reconstructed external encoder and the zero-crossing time of the motor encoder after downsampling are angle-time signals, respectively, expressed as: In the formula, Δt o It is the angle-time signal reconstructed by the external encoder, Δt m It is the angle-time signal reconstructed by the external encoder; H o (·) represents the function indicating the zero-crossing time of the external encoder, H m (·) represents a function indicating the zero-crossing time of the motor encoder, T i Let K be the array length of T at the i-th zero-crossing moment; i = 1, 2, ..., K; (7) The residual time signal Δt including the fault information is calculated using the following equation s ; Δt s = Δt o - Δt m (8) Compensate the residual time signal Δt by the bias coefficient a s Obtain the instantaneous angular velocity signal ω removing the speed trend component. In the formula: N is the number of encoder lines; (9) Perform order spectrum analysis on the ω signal to extract bearing fault characteristics.

2. The rolling bearing fault detection method based on dual optical encoder synchronous time difference according to claim 1, characterized in that, The specific operation of obtaining the rising edge times of the two sets of pulse signals using the zero-crossing time method in step (1) is as follows: (1) Assume the transition interval at the rising edge is [c(j), c(j+1)], and the corresponding voltage amplitude is [y(j), y(j+1)] and y(j) <V th ,y(j+1)>V th V th = (y(j) + y(j+1)) / 2; (2) The zero-crossing time t is calculated by linear interpolation i The calculation formula is: where: f c represents the high speed counter sampling frequency, j = 1, 2, 3,...

3. The dual optical encoder synchronous time-difference based rolling bearing fault detection method of claim 1, wherein: An external encoder is installed at the end of the shaft, which is close to the bearing to be tested.

Images