Anti-noise motor bearing fault adaptive detection method

Abstract

The invention belongs to the field of fault detection, and particularly relates to an anti-noise motor bearing fault adaptive detection method. The method comprises the following steps: firstly, determining mounting positions of a vibration sensor and a rotating speed sensor, and setting an initial value N0, a minimum value Nmin and a maximum value Nmax of the number N of synchronous sampling data points in each rotation of a motor rotating shaft, a rotation number L of the motor rotating shaft in a detection time window, and threshold values T1 and T2; acquiring N groups of periodic synchronous single-point sampling data Y; then calculating correlation coefficients of two adjacent groups of periodic synchronous single-point sampling data in the data Y to obtain data C, and calculating an average value M of absolute values of the data C; and finally, judging whether the motor bearing has a fault or not and realizing adaptive data acquisition. According to the method, correlation coefficients are adopted as fault features, the influence of disturbance noise on a detection result is restrained, the degradation state of the bearing is described reliably, the misdiagnosis rate and the missed diagnosis rate of fault detection are reduced, and then the reliability of motor bearing fault self-adaptive detection is improved; and meanwhile, an adaptive data acquisition strategy is introduced, so that the number N of synchronous sampling data points in each rotation of the motor rotating shaft can be adaptively determined, and the data acquisition, transmission and processing costs are reduced under the condition that the fault detection reliability is ensured.

Description

Technical Field

[0001] The invention belongs to the field of fault detection, and in particular relates to a noise-resistant motor bearing fault adaptive detection method in bearing fault detection technology. Background Art

[0002] Motors include electric motors and generators. Electric motors are common electrical devices that can convert electrical energy into mechanical energy, while generators are common electrical devices that can convert mechanical energy into electrical energy. At present, electric motors have been widely used in industries, energy, agriculture, transportation, household appliances and other fields, and are an indispensable source of power in modern industry and daily life. Motor bearings play a vital role in motors. They not only support the weight of the motor rotor and ensure the smooth rotation of the rotor, but also reduce friction and absorb vibration and shock generated by operation. Long-term operation of the motor will lead to the degradation of the motor bearings, and there is a certain risk of fatigue failure. Studies have shown that fatigue failure usually occurs in the inner ring, outer ring and rolling element of the bearing. In order to improve the working efficiency of the motor and avoid damage caused by bearing failure, it is necessary to perform fault detection on the motor bearings to improve the reliability of the motor operation.

[0003] At present, relevant researchers have proposed a variety of motor bearing fault detection technologies. In motor bearing fault detection, data acquisition and feature extraction are very important links. Using improper data acquisition methods or extracting inappropriate fault features will not only reduce the reliability of fault detection, but also produce erroneous detection results. Today's bearing fault detection methods often use fixed detection interval sampling or variable detection interval sampling. Fixed detection interval sampling will lead to data redundancy or data loss. Variable detection interval sampling refers to adaptively determining the detection interval according to the degradation state of the bearing. This method can overcome the shortcomings of fixed detection interval sampling to a certain extent. However, bearing faults may occur at any stage of its operation, not just in the fault detection stage. In this case, neither of the above two sampling methods can solve the problem of missed fault detection. In order to overcome the shortcomings of the above two sampling methods and reduce the cost of data acquisition, the invention patent with application number 2024115572115 discloses a real-time detection method for motor bearing faults based on adaptive data acquisition. Although this method can solve the problem of missed fault detection and reduce the amount of data collection, the fault features used are greatly affected by disturbance noise, that is, the anti-noise ability is poor and cannot effectively characterize the degradation state of the bearing. In practice, the disturbance noise of the signal is inevitable and changes with the change of the health state of the bearing. The changing disturbance noise seriously limits the reliability of the fault detection method.

[0004] In order to improve the anti-noise ability of the adaptive fault detection method, the present invention discloses a noise-resistant adaptive detection method for motor bearing faults. The method disclosed in the patent of the present invention introduces the correlation coefficient and uses it as a fault feature, thereby suppressing the influence of disturbance noise on the reliability of motor fault detection. The fault features used in the patent of the present invention can be well applied to various health states of bearings, and can also reliably characterize the degradation state of bearings, thereby providing a reference basis for adaptive data acquisition. Summary of the invention

[0005] Purpose of the invention: In response to the problems in the background technology, the present invention discloses a noise-resistant adaptive detection method for motor bearing faults, which introduces correlation coefficients and adaptive decisions, reduces the impact of disturbance noise on fault characteristics, and can adaptively determine the number of synchronously collected data points per revolution of the motor shaft, which not only ensures the reliability of bearing fault detection, but also reduces the amount of data collection, thereby effectively reducing the cost of data collection, transmission and processing.

[0006] Technical solution: The present invention provides a noise-resistant motor bearing fault adaptive detection method, comprising the following steps:

[0007] Step 1: Determine the installation positions of the vibration sensor and the speed sensor, and set the initial value N0 and the minimum value N of the number of synchronous sampling data points N per rotation of the motor shaft. min and the maximum value N max , the number of motor shaft revolutions L in a detection time window, and thresholds T1 and T2.

[0008] The vibration sensor and the speed sensor are installed at the pre-detection bearing and near the shaft, respectively, to measure the bearing vibration signal and the motor shaft speed signal. The initial value N0 of the number of synchronous sampling data per shaft revolution is set to Among them, round() is a rounding function, and the result is an integer. The thresholds T1 and T2 are set according to the statistical values ​​of the fault characteristics under the bearing non-fault and fault conditions.

[0009] Step 2: Obtain N groups of periodically synchronized single-point sampling data Y.

[0010] According to the vibration sensor, the speed sensor and the number of synchronous sampling data points N in each rotation of the motor shaft, N groups of periodic synchronous single-point sampling data Y are obtained. The data can be expressed as:

[0011]

[0012] Among them, Y i =[y(i)y(N+i)y(2N+i)…y((L-1)N+i)] represents the i-th group of periodic synchronous single-point sampling data.

[0013] Step 3: Calculate the correlation coefficient of two adjacent groups of periodic synchronous single-point sampling data in data Y to obtain data C, and calculate the average value M of the absolute value of data C.

[0014] According to the N groups of periodic synchronous single-point sampling data in data Y, N-1 correlation coefficients can be obtained, recorded as data C.

[0015] C=[r(1)r(2)r(3)…r(i)…r(N-1)],

[0016] Among them, two adjacent groups of periodic synchronous single-point sampling data Y i and Y i+1 The correlation coefficient r(i) can be expressed as:

[0017]

[0018] In the formula, cov(Y i ,Y i+1 ) are two sets of adjacent periodic synchronous single-point sampling data Y i and data Y i+1 The covariance of and Represents two adjacent groups of periodic synchronous single-point sampling data Y i and Y i+1 The standard deviation of .

[0019] Taking the absolute value of the data C and then calculating the average value M can be expressed as:

[0020]

[0021] Step 4: Determine whether the motor bearing is faulty and implement adaptive data acquisition.

[0022] If M is less than or equal to T1, the motor bearing is judged to be in a fault-free state, and the number of synchronous sampling data points N per motor shaft revolution is adaptively adjusted to N min , and return to step 2 to continue fault detection; if M is greater than T1 and less than T2, it is impossible to reliably determine whether the bearing is faulty. At this time, the number of synchronous sampling data points N per motor shaft revolution is set to And return to step 2 to continue fault detection; if M is greater than or equal to T2, it is considered that the motor bearing has failed.

[0023] Beneficial effects:

[0024] 1. The method of the present invention discloses a noise-resistant adaptive detection method for motor bearing faults, which overcomes the disadvantage that the existing real-time detection method for motor bearing faults is affected by disturbance noise, and can be applied to fault detection at various health stages of motor bearings.

[0025] 2. The method of the present invention adopts the correlation coefficient as the fault feature, suppresses the influence of disturbance noise on the detection result, reduces the misdiagnosis rate and missed diagnosis rate of fault detection, and improves the reliability of motor bearing fault detection.

[0026] 3. The method of the present invention introduces an adaptive data acquisition strategy, which can adaptively determine the number N of synchronously sampled data points per rotation of the motor shaft, thereby reducing the cost of data acquisition, transmission and processing while ensuring the reliability of fault detection.

[0027] 4. The method of the present invention has low computational cost and does not require a control unit with high processing speed, thereby reducing the cost of the fault detection system.

[0028] 5. The method of the present invention can be applied to the field of bearing fault detection of variable speed motors. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 A flow chart of the motor bearing adaptive fault detection method of the present invention;

[0030] Figure 2 is the probability density function and threshold of fault characteristics when there is no fault and when there is a fault;

[0031] Figure 3 This is the test result when the motor bearing is fault-free;

[0032] Figure 4 N is the number of synchronously sampled data points per shaft revolution when the motor bearing is fault-free;

[0033] Figure 5 This is the detection result when the motor bearing fails;

[0034] Figure 6 N is the number of synchronously sampled data points per shaft revolution when the motor bearing fails;

[0035] Figure 7 The following is a graph showing the average value of the standard deviation and the average value of the absolute value of the correlation coefficient as a function of noise. DETAILED DESCRIPTION

[0036] The implementation process of the present invention is described below in conjunction with the accompanying drawings.

[0037] The present invention discloses a motor bearing fault adaptive detection method based on correlation coefficient, such as Figure 1 As shown, the following steps are included:

[0038] Step 1: Determine the installation positions of the vibration sensor and the speed sensor, and set the initial value N0=51 and the minimum value N0 of the number of synchronous sampling data points N per rotation of the motor shaft. min =2, maximum value N max=100, the number of motor shaft revolutions in a detection time window L=10, threshold values ​​T1=0.287 and T2=0.320.

[0039] The vibration sensor and the speed sensor are installed at the pre-detection bearing and near the shaft, respectively, to measure the bearing vibration signal and the motor shaft speed signal. The initial value N0 of the number of synchronous sampling data per shaft revolution is set to Thresholds T1 and T2 are set according to the statistical values ​​of fault characteristics under the bearing fault-free and fault conditions, such as Figure 2 shown.

[0040] Step 2: Obtain N groups of periodically synchronized single-point sampling data Y.

[0041] According to the vibration sensor, the speed sensor and the number of synchronous sampling data points N in each rotation of the motor shaft, N groups of periodic synchronous single-point sampling data Y are obtained. The data can be expressed as:

[0042]

[0043] Among them, Y i =[y(i)y(N+i)y(2N+i)…y((L-1)N+i)] represents the i-th group of periodic synchronous single-point sampling data.

[0044] Step 3: Calculate the correlation coefficient of two adjacent groups of periodic synchronous single-point sampling data in data Y to obtain data C, and calculate the average value M of the absolute value of data C.

[0045] According to the N groups of periodic synchronous single-point sampling data in data Y, N-1 correlation coefficients can be obtained, recorded as data C.

[0046] C=[r(1)r(2)r(3)…r(i)…r(50)],

[0047] Among them, two adjacent groups of periodic synchronous single-point sampling data Y i and data Y i+1 The correlation coefficient r(i) can be expressed as:

[0048]

[0049] In the formula, cov(Y i ,Y i+1 ) are two sets of adjacent periodic synchronous single-point sampling data Y i and data Y i+1 The covariance of and Represents two adjacent groups of periodic synchronous single-point sampling data Y i and data Y i+1 The standard deviation of .

[0050] Taking the absolute value of the data C and then calculating the average value M can be expressed as:

[0051]

[0052] Step 4: Determine whether the motor bearing is faulty and implement adaptive data acquisition.

[0053] If the fault feature M is less than or equal to T1 = 0.287, the motor bearing is judged to be in a fault-free state, and the number of synchronous sampling data points N per motor shaft revolution is adaptively adjusted to N min =2, and return to step 2 to continue fault detection; if M is greater than T1 = 0.287 and less than T2 = 0.320, it is impossible to reliably determine whether a fault has occurred. At this time, the number of synchronous sampling data points N per motor shaft revolution is set to And return to step 2 to continue fault detection; if M is greater than or equal to T2 = 0.320, it is considered that the motor bearing has failed. In the entire fault detection process, in the same way, the number of synchronous sampling data points N per rotation of the motor shaft is adaptively adjusted to continuously detect whether the bearing fault occurs until the motor bearing is detected to be faulty.

[0054] Figure 3 The detection results when the motor bearing is fault-free are shown. It can be seen that the fault features M in the 30 time windows are all less than the threshold T2. Fault judgment is performed according to step 4, and the process returns to step 2 for adaptive data collection, and then the fault detection is continued. Figure 4 The number of synchronous sampling data points N in each rotation of the shaft when the motor bearing is fault-free is given. As can be seen from the figure, since the bearing is in a fault-free state, the adaptive synchronous acquisition data N in each rotation of the shaft for each fault detection is less than the maximum value of 100, which reduces the number of synchronous sampling data points N in the shaft, thereby reducing the cost of data acquisition, processing and transmission.

[0055] Figure 5 This is the detection result when the motor bearing fails. As can be seen from the figure, the fault feature of the first detection is greater than the threshold T2, which means that a fault has occurred. At this time, an alarm signal is issued or the motor operation is terminated and the fault detection process is stopped. Figure 6 The number N of synchronously sampled data points per rotation of the shaft when the motor bearing fails is given. Since the detected bearing is in a faulty state, the method of the present invention quickly detects the motor bearing failure.

[0056] The method of the present invention introduces the correlation coefficient and uses it as a fault feature, thereby suppressing the influence of disturbance noise on the detection result. Figure 7The figure is a graph showing the average value of the standard deviation and the fault characteristics of the method of the present invention as a function of noise. Compared with the average standard deviation, the fault characteristics of the method of the present invention are less susceptible to disturbance noise, that is, they have better noise resistance. The fault characteristics of the method of the present invention can be well applied to various health stages of bearings, and can also reliably reflect the degradation state of bearings, thereby providing a reference basis for adaptive data acquisition. In addition, the method of the present invention performs adaptive data acquisition based on the degradation state of the bearing, thereby reducing the amount of synchronously collected data per revolution of the shaft. While ensuring the reliability of fault detection, the cost of data acquisition, transmission and processing is reduced. Compared with traditional fault detection methods, the method determines the fault threshold based on the statistical results of the fault characteristics when the bearing has or does not have a fault, and compares it with the current fault characteristics to perform fault judgment and realize adaptive data acquisition. If the current fault characteristic M is less than or equal to T1, the bearing is judged to be in a fault-free state, and the number of synchronously sampled data points N per revolution of the motor shaft is set to the minimum value N min , and return to step 2 to continue fault detection; when the fault feature M is greater than T1 and less than T2, it is impossible to determine whether the bearing has a fault. At this time, N is set to And return to step 2 to continue fault detection; when the fault feature M is greater than or equal to the threshold T2, it is determined that the bearing is faulty.

[0057] The above embodiments are only for illustrating the technical concept and features of the present invention, and their purpose is to enable people familiar with the technology to understand the content of the present invention and implement it accordingly, and they cannot be used to limit the protection scope of the present invention. Any equivalent transformation or modification made according to the spirit of the present invention should be included in the protection scope of the present invention.

Claims

1. A noise-resistant motor bearing fault adaptive detection method, characterized in that: The following steps are involved: Step 1: Determine the installation positions of the vibration sensor and the speed sensor, and set the initial value N0 and the minimum value N of the number of synchronous sampling data points N per rotation of the motor shaft. min and the maximum value N max , the number of motor shaft revolutions L in a detection time window, and thresholds T1 and T2; Step 2: Obtain N groups of periodic synchronous single-point sampling data Y; Step 3: Calculate the correlation coefficient of two adjacent groups of periodic synchronous single-point sampling data in data Y to obtain data C, and calculate the average value M of the absolute value of data C; Step 4: Determine whether the motor bearing is faulty and implement adaptive data acquisition.

2. The noise-resistant motor bearing fault adaptive detection method according to claim 1, characterized in that: In the step 1, a vibration sensor and a rotation speed sensor are respectively installed at the pre-detection bearing and near the shaft to measure the bearing vibration signal and the motor shaft rotation speed signal.

3. The noise-resistant motor bearing fault adaptive detection method according to claim 1, characterized in that: In step 1, the initial value N0 of the number of synchronous sampling data per rotation of the shaft is set to Among them, round() is a rounding function, and the result is an integer.

4. The noise-resistant motor bearing fault adaptive detection method according to claim 1, characterized in that: In step 1, thresholds T1 and T2 are set according to the statistical values ​​of the fault characteristics of the bearing under normal and fault conditions.

5. The noise-resistant motor bearing fault adaptive detection method according to claim 1, characterized in that: In step 2, according to the vibration sensor, the speed sensor and the number of synchronous sampling data points N per rotation of the motor shaft, N groups of periodic synchronous single-point sampling data Y are obtained, which can be expressed as: Among them, Y i =[y(i) y(i+N) y(i+2N) … y((L-1)N+i)] represents the i-th group of periodic synchronous single-point sampling data.

6. The noise-resistant motor bearing fault adaptive detection method according to claim 1, characterized in that: In step 3, the correlation coefficient of two adjacent groups of periodic synchronous single-point sampling data in data Y is calculated to obtain data C = [r(1)r(2)r(3) ... r(i) ... r(N-1)], where two adjacent groups of periodic synchronous single-point sampling data Y i and Y i+1 Correlation coefficient In the formula, cov(Y i ,Y i+1 ) are two sets of adjacent periodic synchronous single-point sampling data Y i and data Y i+1 The covariance of and Represents two adjacent groups of periodic synchronous single-point sampling data Y i and Y i+1 The standard deviation of .

7. The noise-resistant motor bearing fault adaptive detection method according to claim 1, characterized in that: In step 3, the average value of the absolute value of data C is calculated 8. The noise-resistant motor bearing fault adaptive detection method according to claim 1, characterized in that: In step 4, M is used as a fault feature to compare with thresholds T1 and T2; if the fault feature M is less than or equal to T1, the motor bearing is determined to be in a fault-free state, and the number of synchronous sampling data points N per rotation of the motor shaft is adaptively adjusted to N min , and return to step 2 to continue fault detection; if the fault feature M is greater than T1 and less than T2, it is impossible to reliably determine whether a fault has occurred. At this time, the number of synchronous sampling data points N per motor shaft revolution is set to And return to step 2 to continue fault detection; if the fault feature M is greater than or equal to T2, it is considered that the motor bearing has failed.

Images