Remote detection method for early fault of bearing

Abstract

The invention discloses a bearing early-stage fault remote detection method. The method comprises the following steps that a vibration sensor is installed to measure bearing vibration data; the laser rotating speed sensor measures rotating speed data of the rotating shaft; the laser emitted by the laser rotating speed sensor is reflected by the sampling reflective strip, and the corresponding rotating shaft corner position during periodic synchronous single-point sampling vibration data acquisition is determined; setting a time window length; sampling a delay angle; a reference value; a threshold; acquiring periodic synchronization single-point sampling data in a specific time window; calculating a standard deviation and an accumulated value of the periodic synchronous single-point sampling data; and judging whether the time window has a fault or not, adjusting the sampling delay of the next time window, and carrying out fault detection of the next time window. According to the method, fault detection is carried out by changing the rotating angle of the rotating shaft during periodic synchronous single-point sampling in each time window, namely sampling delay and accumulating fault information of historical time windows, and the reliability of remote detection of early faults of the bearing is improved.

Description

Technical Field

[0001] The invention relates to the technical field of rotating machinery fault diagnosis, and in particular to a remote detection method for early bearing faults. Background Art

[0002] Bearing failure is a common fault in rotating machinery. According to incomplete statistics, 30% of rotating machinery failures are caused by bearing failures. In the field of motor applications, almost half of motor failures are caused by bearing failures. Once a bearing fails, it may trigger a chain reaction that causes the entire rotating machinery system to shut down, which may cause huge economic losses and adverse social impacts. Bearing early fault detection technology can detect it in the early stage of the fault, and then provide a basis for the next step of fault diagnosis or fault-tolerant control, thereby avoiding the fault and the harm it causes. Therefore, it is of great significance to detect early bearing faults in a timely and accurate manner.

[0003] Early bearing fault detection has attracted widespread attention from researchers and engineers. At present, in order to reliably detect early bearing faults and avoid their hazards, a variety of early bearing fault detection methods have been proposed. The invention patent with application number 201711375557.3 obtains synchronous data by resampling the signal, and then uses an improved bispectral model to calculate the resampled synchronous data, and detects and diagnoses early bearing faults based on the calculation results. Although this method can detect early faults, the amount of sampled data is large and the calculation cost is relatively high, resulting in increased detection cost and data storage cost.

[0004] After a bearing fault occurs, nonlinear vibration components such as bifurcation or chaos will appear in the bearing vibration response. Therefore, some methods based on bifurcation or chaos detection are also applied to bearing fault detection. The invention patent with application number CN201510297851.1 discloses an entropy-based bearing fault detection method, which requires the collection of a large number of bearing vibration signals and is not suitable for remote detection of bearing faults. In the field of rotating machinery fault diagnosis, the periodic synchronous single-point sampling technology synchronously collects a single amount of data in each rotation cycle of the rotating shaft. This sampling technology significantly reduces the amount of data in the vibration response, facilitates data transmission and storage, and can be applied to remote fault detection and diagnosis. Although the periodic synchronous single-point sampling technology obtains less sampling data and can be applied to remote fault diagnosis, this sampling technology cannot reliably detect bearing faults, such as oil film vortex faults in sliding bearings. Although the periodic synchronous multi-point sampling data can effectively characterize the bearing fault and avoid the shortcomings of the synchronous single-point sampling technology, the relatively large amount of data collected is not conducive to data transmission, thereby affecting the real-time nature of remote faults. Summary of the invention

[0005] Purpose of the invention: In order to overcome the shortcomings of the background technology, the present invention requests to disclose a method for remote detection of early bearing faults. The method performs fault detection by changing the shaft angle during periodic synchronous single-point sampling in each time window, that is, the sampling delay, and accumulating fault information in historical time windows, thereby improving the reliability of remote detection of early bearing faults.

[0006] Technical solution: The remote detection method for early bearing faults disclosed in the present invention comprises the following steps:

[0007] S1. Installation:

[0008] Vibration sensor, measuring bearing vibration data;

[0009] Laser speed sensor, measuring shaft speed data;

[0010] The sampling reflective strip reflects the laser emitted by the laser speed sensor to determine the corresponding shaft angular position during periodic synchronous single-point sampling vibration data collection;

[0011] Settings: time window length; sampling delay angle; reference value; threshold;

[0012] S2, combining the sampling delay angle to obtain periodic synchronous single-point sampling data within a specific time window;

[0013] S3, calculating the standard deviation and cumulative value of the above-mentioned periodic synchronous single-point sampling data;

[0014] S4. Use the accumulated value as a fault indicator to determine whether a fault occurs in the time window, adjust the sampling delay of the next time window, and perform fault detection in the next time window.

[0015] The vibration sensor in S1 is a vibration displacement sensor, a vibration velocity sensor or a vibration acceleration sensor.

[0016] Furthermore, the vibration sensor in S1 is arranged near the bearing to be detected; the laser speed sensor is arranged near the rotating shaft; the sampling reflective strip is arranged on the rotating shaft, and the rotating shaft angle of the setting position is set to zero radian angle.

[0017] Furthermore, the delay angle α in S1 is greater than zero and less than 2π; in the first time window, the sampling delay angle is not set, and in the subsequent time windows, the sampling delay angle is an integer multiple of the angle α; the reference value μ is the average value of the standard deviation of the periodic synchronous single-point sampling data when the bearing is fault-free; the threshold T is the preset fault indicator critical value.

[0018] Furthermore, S2 obtains the periodic synchronous single-point sampling data X in the kth time window k ={x k1 ,x k2,x k3 ,…,x kL}, where k = 1, 2, 3, ...; L is the time window length; the sampling delay angle of the kth time window is (k-1)a-2iπ, where i is an integer greater than or equal to zero, and its value should ensure 0≤(k-1)a-2iπ<2π.

[0019] Furthermore, in S3:

[0020] X k The standard deviation of k :

[0021]

[0022] Among them, μ k The periodic synchronous single-point sampling data X in the kth time window k The mean of

[0023] Furthermore, in S3: the CUSUM technique is used to calculate the standard deviation S of k time windows k ={s1,s2,s3,…,s k The cumulative value C k :

[0024] C k =C k-1 +(s k -μ)

[0025] Among them, C0 is 0, and μ is the reference value preset in S1.

[0026] Furthermore, in S4:

[0027] The accumulated value C k It is regarded as a fault indicator and compared with the preset threshold T. If the cumulative value C k If the value is greater than or equal to a preset threshold, it is determined that the bearing has experienced an early fault; otherwise, the bearing has not experienced an early fault, and fault detection continues for the k+1th time window;

[0028] Adjust the sampling delay angle of the k+1th time window to ka-2iπ, where i is an integer greater than or equal to zero, and its value should ensure that 0≤ka-2iπ<2π. According to the sampling delay angle of the k+1th time window, determine the sampling delay time of this window. in, is the period of each rotation of the shaft in the kth time window,

[0029] Among them, T k It is the time taken for the shaft to rotate L times in the kth time window.

[0030] Beneficial effects: Compared with the prior art, the advantages of the present invention are:

[0031] (1) One data is collected synchronously within one rotation cycle of the shaft, and the amount of data collected is relatively small, thereby reducing the cost of data transmission, processing and storage;

[0032] (2) It can collect periodic synchronous single-point sampling data at different shaft angular positions, overcome the shortcomings of traditional fault detection methods, and improve the reliability of fault detection;

[0033] (3) The CUSUM technology is used to accumulate fault information in the historical time window, further improving the reliability of fault detection;

[0034] (4) The method of the present invention is not affected by changes in the rotation speed of the rotating machinery. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a flow chart of the present invention;

[0036] Figure 2 is a fault-free simulation signal without noise;

[0037] Figure 3 Fault simulation signal without noise;

[0038] Figure 4 is a fault-free simulation signal containing noise;

[0039] Figure 5 is a fault simulation signal containing noise;

[0040] Figure 6 is the fault detection result of the method of the present invention;

[0041] Figure 7 This is the bearing fault detection result when there is no sampling delay. DETAILED DESCRIPTION

[0042] like Figure 1 The remote detection method for early bearing faults shown includes the following steps:

[0043] S1. Installation:

[0044] Vibration sensor, measuring bearing vibration data;

[0045] Laser speed sensor, measuring shaft speed data;

[0046] The sampling reflective strip reflects the laser emitted by the laser speed sensor to determine the corresponding shaft angular position during periodic synchronous single-point sampling vibration data collection;

[0047] Settings: time window length; sampling delay angle; reference value; threshold;

[0048] S2, combining the sampling delay angle to obtain periodic synchronous single-point sampling data within a specific time window;

[0049] S3, calculating the standard deviation and cumulative value of the above-mentioned periodic synchronous single-point sampling data;

[0050] S4. Use the accumulated value as a fault indicator to determine whether a fault occurs in the time window, adjust the sampling delay of the next time window, and perform fault detection in the next time window.

[0051] Specific:

[0052] The vibration sensor in S1 is installed near the bearing to be tested to measure the bearing vibration data. The vibration sensor can use a vibration displacement sensor, a vibration velocity sensor or a vibration acceleration sensor. The laser speed sensor is installed near the shaft to measure the shaft speed data. The sampling reflective strip needs to be pasted on the shaft and can reflect the laser emitted by the laser speed sensor. The sampling reflective strip is used to determine the corresponding shaft rotation angle position when collecting periodic synchronous single-point sampling vibration data; the pasting position of the sampling reflective strip on the shaft is set to the zero arc angle of the shaft.

[0053] The sampling delay angle α needs to be greater than zero and less than 2π. The time window length L is set to 10, and the sampling delay angle α is set to In the first time window, the sampling delay angle is not set, and the sampling delay angle can also be considered to be zero radian angle, while in subsequent time windows, the sampling delay angle is an integer multiple of α. The reference value μ is the average value of the standard deviation of the periodic synchronous single-point sampling data when the bearing is fault-free. The reference value μ is set to 2.35, and the threshold T is 0.65.

[0054] The periodic synchronous single-point sampling vibration data in S2 refers to the synchronous acquisition of a single sampling data within each rotation period of the shaft. With the assistance of the laser speed sensor, the vibration sensor is used to obtain the periodic synchronous single-point sampling vibration data X in the kth time window. k ={x k1 ,x k2 ,x k3 ,…,x kL}, where k = 1, 2, 3, ...; L is the time window length. The sampling delay angle of the kth time window is (k-1)a-2iπ, where i is an integer greater than or equal to zero, and its value should ensure that 0≤(k-1)a-2iπ<2π.

[0055] S3 calculates the periodic synchronous single-point sampling vibration data X in the kth time window k The standard deviation of k ;

[0056]

[0057] Among them, μ k The periodic synchronous single-point sampling data X in the kth time window k The mean of

[0058] Calculate the standard deviation S of k time windows k ={s1,s2,s3,…,s k The cumulative value C k ;

[0059] The CUSUM technique is used to calculate the cumulative value C of the standard deviation of k time windows. k ,

[0060] C k =C k-1 +(s k -μ)

[0061] Among them, C0 is 0, and μ is the reference value preset in S1.

[0062] In S4, it is determined whether a fault occurs, the sampling delay of the k+1th time window is adjusted, and fault detection of the k+1th time window is performed.

[0063] The accumulated value C k Compare with the preset threshold T. If the cumulative value C k If the value is greater than or equal to the preset threshold value T, the bearing is considered to have an early fault. Otherwise, the bearing is considered to have no early fault and the fault detection of the k+1th time window is continued. At this time, the sampling delay angle of the k+1th time window is adjusted to ka-2iπ, where i is an integer greater than or equal to zero and its value should ensure that 0≤ka-2iπ<2π. According to the sampling delay angle of the k+1th time window, the sampling delay time of this window is determined. in, is the period of each rotation of the shaft in the kth time window, Among them, T k It is the time taken by the shaft L to rotate in the kth time window.

[0064] This embodiment performs remote detection of the oil film vortex fault of the sliding bearing. The oil film vortex fault can also be called a half-speed vortex fault. The fault frequency of the oil film vortex fault is approximately half of the shaft speed. For a certain rotating machine, sliding bearings are used at both ends of the shaft for support. According to existing references, a sliding bearing simulation vibration signal is constructed and tested. The constructed fault simulation signal S F (t) and fault-free simulation signal S N (t) are as follows:

[0065] S F (t) = s1 + s2 + s3 + s4

[0066] S N (t) = s2 + s3 + s4

[0067] s1=4sin(10πt)

[0068] s2=10sin(20πt)

[0069] s3=4sin(40πt)

[0070] s4=2sin(60πt)

[0071] Where s1 is the fault component of the sliding bearing, and s2, s3 and s4 are the normal vibration components of the sliding bearing. N (t) and fault simulation signal S F (t) respectively as Figure 2 and 3 As shown in the figure. The asterisk and triangle in the figure represent two sets of periodic synchronous single-point sampling data at different shaft rotation angle positions. For the fault-free simulation signal, the two sets of periodic synchronous single-point sampling data are constants and do not contain fluctuation components. However, a set of periodic synchronous single-point sampling data of the fault simulation signal is no longer a constant, such as Figure 3 As shown by the asterisk in the figure. According to this group of periodic synchronous single-point sampling data, the sliding bearing fault can be identified, but according to the periodic synchronous single-point sampling data shown by the triangle, the sliding bearing fault cannot be identified. In order to solve this problem, the patent of the present invention proposes to adopt a delayed sampling strategy, that is, different time delay angles are used when acquiring periodic synchronous single-point sampling in each time window. In addition, the patent of the present invention proposes to use CUSUM technology to accumulate the fault information collected in each time window, so that early faults can be reliably detected.

[0072] In order to simulate the influence of disturbance noise, we set the signal-to-noise ratio of the above simulation signal to 10dB. The fault-free simulation signal and the fault simulation signal after adding noise are shown as follows: Figure 4 and 5 As shown in the figure. Due to the influence of noise, it is impossible to identify the bearing fault from the original data containing noise. The method of the present invention can detect the fault. The detection results of the noise-free and faulty signals are shown in the figure. Figure 6 As shown in the figure, the bearing fault is detected in the sixth time window. The method of the present invention adjusts the sampling delay in each time window, which effectively avoids the disadvantage of periodic synchronous single-point sampling in fault detection, that is, a certain set of periodic synchronous single-point sampling data may not be able to detect the bearing fault, for example Figure 3 The triangle in the middle represents periodically synchronized single-point sampling data. Figure 7 The detection results of this group of periodic synchronous single-point sampling data are shown. It can be seen from the figure that the bearing fault cannot be identified because the group of periodic synchronous single-point sampling data of the fault signal has the same characteristics as the periodic synchronous single-point sampling data of the fault-free signal.

[0073] In summary, the method of the present invention synchronously collects one data within one rotation of the shaft, and the amount of data collected is relatively small, thereby reducing the cost of data transmission and storage, which is conducive to remote fault detection; since the amount of data acquired within a time window is small and the data processing cost is low, the present invention can detect early bearing faults more quickly; the method of the present invention effectively avoids the disadvantage that a certain group of periodic synchronous single-point sampling data cannot identify faults by adjusting the sampling delay, thereby improving the reliability of early bearing fault detection.

Claims

1. A remote detection method for early bearing faults, characterized in that: The following steps are involved: S1. Installation: Vibration sensor, measuring bearing vibration data; Laser speed sensor, measuring shaft speed data; The sampling reflective strip reflects the laser emitted by the laser speed sensor to determine the corresponding shaft rotation angle position during periodic synchronous single-point sampling vibration data collection; Settings: time window length; sampling delay angle; reference value; threshold; S2, combining the sampling delay angle to obtain periodic synchronous single-point sampling data within a specific time window; S3, calculating the standard deviation and cumulative value of the above-mentioned periodic synchronous single-point sampling data; S4. Use the accumulated value as a fault indicator to determine whether a fault occurs in the time window, adjust the sampling delay of the next time window, and perform fault detection in the next time window.

2. The method for remotely detecting early bearing faults according to claim 1 is characterized in that: The vibration sensor in S1 is a vibration displacement sensor, a vibration velocity sensor, or a vibration acceleration sensor.

3. The method for remotely detecting early bearing faults according to claim 1 is characterized in that: The vibration sensor in S1 is arranged near the bearing to be detected; the laser speed sensor is arranged near the rotating shaft; the sampling reflective strip is arranged on the rotating shaft, and the rotating shaft angle at the setting position is set to zero radian angle.

4. The method for remotely detecting early bearing faults according to claim 1 is characterized in that: The delay angle α described in S1 is greater than zero and less than 2π; in the first time window, the sampling delay angle is not set, and in the subsequent time windows, the sampling delay angle is an integer multiple of the angle α; the reference value μ is the average value of the standard deviation of the periodic synchronous single-point sampling data when the bearing is fault-free; the threshold T is the preset fault indicator critical value.

5. The method for remotely detecting early bearing faults according to claim 1 is characterized in that: S2 obtains the periodic synchronous single-point sampling data X in the kth time window k ={x k1 ,x k2 ,x k3 ,…,x kL }, where k = 1, 2, 3, ...; L is the time window length; the sampling delay angle of the kth time window is (k-1)a-2iπ, where i is an integer greater than or equal to zero, and its value should ensure 0≤(k-1)a-2iπ<2π.

6. The method for remotely detecting early bearing faults according to claim 5, characterized in that: In S3: X k The standard deviation of k : Among them, μ k The periodic synchronous single-point sampling data X in the kth time window k The mean of 7. The method for remotely detecting early bearing faults according to claim 6, characterized in that: In S3: Use CUSUM technology to calculate the standard deviation S of k time windows k ={s1,s2,s3,…,s k The cumulative value C k : C k =C k-1 +(s k -μ) Among them, C0 is 0, and μ is the reference value preset in S1.

8. The method for remotely detecting early bearing faults according to claim 7, characterized in that: In S4: The accumulated value C k It is regarded as a fault indicator and compared with the preset threshold T. If the cumulative value C k If the value is greater than or equal to a preset threshold, it is determined that the bearing has experienced an early fault; otherwise, the bearing has not experienced an early fault, and fault detection continues for the k+1th time window; Adjust the sampling delay angle of the k+1th time window to ka-2iπ, where i is an integer greater than or equal to zero, and its value should ensure that 0≤ka-2iπ<2π. According to the sampling delay angle of the k+1th time window, determine the sampling delay time of this window in, is the period of each rotation of the shaft in the kth time window, Among them, T k It is the time taken for the shaft to rotate L times in the kth time window.

Images