A Method for Extracting Bearing Fault Acoustic Emission Events Based on Short-Time Autocorrelation

Abstract

The present invention discloses a method for extracting bearing fault acoustic emission events based on short-time autocorrelation, which includes the following steps: First, use a low-speed bearing fault simulation test bench to collect bearing fault acoustic emission signals, and the fault type is outer race fault. Then, perform frame-by-frame processing on the bearing fault acoustic emission signals, and calculate the short-time autocorrelation function of each frame of signals. Based on the short-time autocorrelation function, set a threshold to initially obtain the starting point and ending point of the fault acoustic emission events. Finally, search for the endpoints of the fault acoustic emission events according to the second derivative of the short-time autocorrelation function. It overcomes the problem of fixed extraction parameters for traditional acoustic emission events and the low accuracy of the extracted events. By parameterizing the waveform characteristics of the signals, it can more accurately extract bearing fault acoustic emission events.

Description

Technical Field

[0001] The present invention relates to a method for extracting bearing fault acoustic emission events based on short-time autocorrelation, which is a method for adaptively extracting acoustic emission events. Background Art

[0002] In the detection of acoustic emission events, the traditional method is to complete the extraction of events for continuous acoustic emission signals by setting fixed thresholds, pre-triggering, and wavelengths. It is difficult to achieve adaptive changes of parameters according to the fault transient pulse characteristics, and the extraction of events with fixed parameters is likely to cause missed detection of acoustic emission events and interference between two adjacent events.

[0003] During the extraction process of acoustic emission events, the events of acoustic emission signals have a large difference in amplitude relative to background noise. If the waveform characteristic parameters of acoustic emission signals are parameterized, the signal components of events are extracted adaptively. This extraction idea can automatically adjust the intercepted length according to the characteristics of events. Compared with the traditional extraction method with fixed parameters, short-time autocorrelation can effectively express the difference between acoustic emission events and background noise and the sample length occupied by events in continuous signals, and the accuracy of event extraction is higher. Summary of the Invention

[0004] The present invention provides a method for extracting bearing fault acoustic emission events based on short-time autocorrelation, which is used to solve the problems of fixed extraction parameters of traditional acoustic emission events and low precision of extracted events. The present invention uses a low-speed bearing fault simulation test bench to collect bearing fault acoustic emission signals, and the fault type is an outer race fault; the present invention performs frame division processing on the bearing fault acoustic emission signals and calculates the short-time autocorrelation function of each frame of signals; based on the short-time autocorrelation function, a threshold is set to initially obtain the start and end points of fault acoustic emission events; finally, the endpoints of fault acoustic emission events are searched according to the second derivative of the short-time autocorrelation function.

[0005] The technical solution adopted by the present invention is as follows: A method for extracting bearing fault acoustic emission events based on short-time autocorrelation, the method comprising the following steps:

[0006] (1) Use a low-speed bearing fault simulation test bench to collect bearing fault acoustic emission signals, and the fault type is an outer race fault;

[0007] (2) Perform frame division processing on the bearing fault acoustic emission signals and calculate the short-time autocorrelation function of each frame of signals;

[0008] (3) According to the short-time autocorrelation function of the signal, set a threshold to initially obtain the start and end points of fault acoustic emission events;

[0009] (4) Calculate the second derivative of the short-time autocorrelation function, search for the points where the second derivative function is greater than 0 near the starting point and the ending point as the endpoints of the fault acoustic emission event, and extract the fault acoustic emission event according to the indices of the endpoints.

[0010] As a further solution of the present invention, the data acquisition parameters in step (1) are: the sampling unit is V, the sampling rate is 1 MHz, the bearing outer ring fault size is 1 mm, and the rotational speed of the bearing operation is 4 r / min.

[0011] As a further solution of the present invention, in step (2), frame processing is performed on the bearing fault acoustic emission signal, and the processing process is as follows:

[0012] Perform windowing on the signal, the window function is a Hanning window, the width of the window function is the interval between the two peaks of the first pulse of the fault acoustic emission signal, and the sliding amount of each frame of the signal is half of the window function.

[0013] As a further solution of the present invention, in step (2), calculate the short-time autocorrelation function of each frame of the signal, and the specific process is as follows:

[0014] Calculate the autocorrelation function for the framed acoustic emission signal, generate the short-time autocorrelation function of the bearing fault acoustic emission signal, and perform normalization processing on the short-time autocorrelation function.

[0015] As a further solution of the present invention, in step (3), the threshold is set as the mean value of the short-time autocorrelation function of the bearing after normalization processing, and the intersection points of the threshold and the normalized short-time autocorrelation function can initially determine the starting point and the ending point of the bearing fault acoustic emission event.

[0016] As a further solution of the present invention, in step (4), search for the endpoints of the bearing fault acoustic emission event, and the specific process is as follows:

[0017] Calculate the second derivative of the normalized short-time autocorrelation function, index all the points where the second derivative is greater than 0, and compare these points with the starting point and the ending point of the already determined fault acoustic emission event. The point with the smallest index position difference is the endpoint of the fault event.

[0018] The beneficial effects of the present invention are as follows:

[0019] 1. The present invention uses a low-speed bearing fault simulation test bench to collect bearing fault acoustic emission signals, and the fault type is an outer ring fault; the present invention performs frame processing on the bearing fault acoustic emission signal and calculates the short-time autocorrelation function of each frame of the signal; based on the short-time autocorrelation function, a threshold is set to initially obtain the starting point and the ending point of the fault acoustic emission event. Finally, search for the endpoints of the fault acoustic emission event according to the second derivative of the short-time autocorrelation function;

[0020] 2. The present invention solves the problems of fixed parameter extraction for traditional acoustic emission events and low accuracy of the extracted events. By parameterizing the waveform characteristics of the signal, the present invention can more accurately extract bearing fault acoustic emission events. Description of the Drawings

[0021] Figure 1 It is the overall flowchart of the present invention;

[0022] Figure 2 It is a schematic diagram for determining the frame length and window function slip amount parameters of the present invention;

[0023] Figure 3 It is the time-domain waveform of the bearing fault acoustic emission signal including two fault events of the present invention;

[0024] Figure 4 It is the autocorrelation function of the fault signal after normalization of the present invention;

[0025] Figure 5 It is a schematic diagram for calculating the preliminary search for the endpoints of fault events of the present invention;

[0026] Figure 6 It is to determine the final endpoints according to the second derivative of the autocorrelation function after normalization of the present invention;

[0027] Figure 7 It is the fault event result extracted by the present invention;

[0028] Figure 8 It is the extraction result of the bearing fault event with a sampling length of 100 s of the present invention;

[0029] Figure 9 It is the extraction result of the fault event from 7.6 to 8.4 s of the present invention; Detailed Embodiment

[0030] The present invention will be further described below in conjunction with embodiments, but the protection scope of the present invention is not limited thereto:

[0031] Embodiment 1: As Figures 1-9 shown, a method for extracting bearing fault acoustic emission events based on short-time autocorrelation, the method includes the following steps:

[0032] (1) Use a low-speed bearing fault simulation test bench to collect bearing fault acoustic emission signals, the fault type is outer ring fault; the sampling unit is V, the sampling rate is 1 MHz, the size of the outer ring fault of the bearing is 1 mm, and the rotational speed of the bearing operation is 4 r / min;

[0033] (2) Frame the bearing fault acoustic emission signal and calculate the short-time autocorrelation function of each frame of the signal; in step (2), the process of framing the bearing fault acoustic emission signal is as follows: perform windowing on the signal, with the window function being the Hanning window, the width of the window function being the interval between the two peaks of the first pulse of the fault acoustic emission signal, and the sliding amount of each frame of the signal being half of the window function.

[0034] In step (2), the process of calculating the short-time autocorrelation function of each frame of the signal is as follows: calculate the autocorrelation function of the framed acoustic emission signal to generate the short-time autocorrelation function of the bearing fault acoustic emission signal, and normalize the short-time autocorrelation function.

[0035] (3) Set a threshold based on the short-time autocorrelation function of the signal to initially obtain the start and end points of the fault acoustic emission event; in step (3), the threshold is set as the mean value of the short-time autocorrelation function of the bearing after normalization, and the intersection points of the threshold and the normalized short-time autocorrelation function can be initially determined as the start and end points of the bearing fault acoustic emission event.

[0036] (4) Calculate the second derivative of the short-time autocorrelation function, search for the points where the second derivative function is greater than 0 near the start and end points as the end points of the fault acoustic emission event, and extract the fault acoustic emission event according to the indices of the end points. In step (4), the process of searching for the end points of the bearing fault acoustic emission event is as follows: calculate the second derivative of the normalized short-time autocorrelation function, index all the points where the second derivative is greater than 0, and compare these points with the start and end points of the already determined fault acoustic emission event. The point with the smallest index difference is the end point of the fault event.

[0037] Example 2: As Figures 1-9 shown, a method for extracting bearing fault acoustic emission events based on short-time autocorrelation, the overall process of this method is as Figure 1 shown, and the specific steps of the method are as follows:

[0038] Step 1: Use a low-speed bearing fault simulation test bench to collect bearing fault acoustic emission signals. The fault type is an outer ring fault, and the size of the bearing outer ring fault is 1 mm; the rotational speed of the bearing is 4 r / min; the sampling unit is V, and the sampling rate is 1 MHz.

[0039] Step 2: Frame the bearing fault acoustic emission signal and calculate the short-time autocorrelation function of each frame of the signal;

[0040] (1) The framing process is as follows:

[0041] A bearing fault acoustic emission signal with a length of N is x(n), and the i-th frame signal after frame splitting is y i (n), which can be expressed as:

[0042] yi y(n) = ω(n)x((i - 1)*inc + n), 1 ≤ n ≤ L, 1 ≤ i ≤ M

[0043] where ω(n) is a window function; y i (n) represents the value of each signal frame; n represents the number of samples in each data frame; L is the frame length and also the window width of the window function ω(n); inc is the sliding length of the window function;

[0044] M is the total number of frames and can be expressed as:

[0045]

[0046] The window function ω(n) is selected as the Hanning window, and the frame length L and the sliding length inc of the window function are selected as 300 and 150 sample points of the data. The specific process is shown in Figure 2 .

[0047] (2) The actual fault event endpoints are as Figure 3 shown. To accurately search for the endpoints, the autocorrelation function of each frame of the signal needs to be calculated. The calculation process is as follows:

[0048]

[0049] where R i (k) is the autocorrelation function of the i-th frame signal y i (n), and k is the delay amount. Generally, k = 1.

[0050] Step 3: Normalize the short-time autocorrelation function of the signal, as Figure 4 shown; and set a threshold to initially obtain the start and end points of the fault acoustic emission event, see Figure 5 ;

[0051] The normalization process is as follows:

[0052]

[0053] where R N represents the normalized short-time autocorrelation function, R represents the autocorrelation function generated in Step 2, R max and R min represent the maximum and minimum values of the autocorrelation function.

[0054] The threshold is set as follows:

[0055]

[0056] where th is the threshold, R i represents the value of the i-th normalized autocorrelation function, and n is the data sample length of the normalized autocorrelation function.

[0057] Step 4: Calculate the second derivative of the short-time autocorrelation function, and search for the points where the second derivative function is greater than 0 near the starting point and the ending point as the endpoints of the fault acoustic emission event, as Figure 6 shown; Extract the fault acoustic emission event according to the index of the endpoint, and the result is shown in Figure 7 .

[0058] Finally, taking a continuous data with a sampling length of 100 s as an example, the result of extracting the fault event is shown in Figure 8 ; The extraction of the fault event with local amplification from 7.6 s to 8.4 s is as Figure 9 shown, and the result shows that the method of the present invention can accurately identify the endpoints of two adjacent events and avoid the overlap of events.

[0059] The specific embodiments of the present invention have been described in detail above in conjunction with the accompanying drawings. However, the present invention is not limited to the above embodiments, and various changes can be made without departing from the spirit of the present invention within the scope of knowledge possessed by those of ordinary skill in the art.

Claims

1. A method for extracting bearing fault acoustic emission events based on short-time autocorrelation, characterized in that: The following steps are involved: (1) Use a low-speed bearing fault simulation test bench to collect bearing fault acoustic emission signals, and the fault type is outer ring fault; (2) Frame the bearing fault acoustic emission signal and calculate the short-time autocorrelation function of each frame signal; (3) According to the short-time autocorrelation function of the signal, a threshold is set to preliminarily obtain the starting point and end point of the fault acoustic emission event; (4) Calculate the second-order derivative of the short-time autocorrelation function, search for points near the starting point and the end point where the second-order derivative function is greater than 0 as the endpoints of the fault acoustic emission event, and extract the fault acoustic emission event according to the index of the endpoint.

2. The method for extracting bearing fault acoustic emission events based on short-time autocorrelation according to claim 1 is characterized in that: The data acquisition parameters in step (1) are: sampling unit is V, sampling rate is 1 MHz, bearing outer race fault size is 1 mm, and bearing running speed is 4 r / min.

3. The method for extracting bearing fault acoustic emission events based on short-time autocorrelation according to claim 1 is characterized in that: In step (2), the bearing fault acoustic emission signal is framed and processed as follows: The signal is windowed and the window function is a Hanning window. The width of the window function is the interval between the two peaks of the first pulse of the fault acoustic emission signal, and the slippage of each frame signal is half of the window function.

4. The method for extracting bearing fault acoustic emission events based on short-time autocorrelation according to claim 1 is characterized in that: In step (2), the short-time autocorrelation function of each frame signal is calculated. The specific process is: The autocorrelation function of the framed acoustic emission signal is calculated to generate the short-time autocorrelation function of the bearing fault acoustic emission signal, and the short-time autocorrelation function is normalized.

5. The method for extracting bearing fault acoustic emission events based on short-time autocorrelation according to claim 1 is characterized in that: The threshold value in step (3) is set as the mean value of the short-time autocorrelation function after normalization of the bearing. The intersection of the threshold value and the normalized short-time autocorrelation function can be preliminarily determined as the starting point and end point of the bearing fault acoustic emission event.

6. The method for extracting bearing fault acoustic emission events based on short-time autocorrelation according to claim 1 is characterized in that: In step (4), the endpoint of the bearing fault acoustic emission event is searched. The specific process is as follows: Calculate the second-order derivative of the normalized short-time autocorrelation function, index all points where the second-order derivative is greater than 0, and compare these points with the start and end points of the determined fault acoustic emission event. The point with the smallest index position difference is the endpoint of the fault event.

Images