A noise correlation-based abrasive grain signal extraction method

Abstract

The present application belongs to the field of oil liquid abrasive particle signal detection, and particularly relates to a kind of abrasive particle signal extraction method based on noise correlation, including installing inductive abrasive particle detection sensor, collection detection signal and reference signal;Detection signal and reference signal are filtered to obtain detection low frequency harmonic and reference low frequency harmonic, and are divided into L detection wave and L reference wave, the waveform similarity of detection wave and its corresponding reference wave is calculated;Waveform similarity not greater than waveform similarity threshold is screened out, and its corresponding fragment position is marked as suspected abrasive particle signal fragment position;Detection signal and reference signal are subtracted to carry out low-pass filtering and obtain preliminary noise reduction signal;And according to suspected abrasive particle signal fragment position, preprocessing fragment is extracted;The numerical characteristics of preprocessing fragment are extracted, and ideal abrasive particle signal is established;The fragment matching degree is calculated and threshold is set, the fragment below threshold is excluded, and the extraction and counting of abrasive particle signal are completed;The present application can stably segment abrasive particle signal.

Description

Technical Field

[0001] This invention belongs to the field of oil abrasive signal detection, specifically relating to a method for extracting abrasive signals based on noise correlation. Background Technology

[0002] The number of wear particles generated during the operation of mechanical equipment is a crucial indicator of its health, reflecting the internal wear condition and thus enabling the development of maintenance plans to prevent major safety accidents caused by abnormal wear. Oil wear particle detection sensors are widely used in fields such as wear detection of ship gearboxes and operational status assessment of aircraft engines. However, in practical applications, the induced signals generated by wear particles in the sensor are very weak and inevitably subject to random noise and vibration interference, causing distortion in the sensor's output wear particle signal and severely limiting the detection performance of inductive wear particle detection sensors.

[0003] Currently, abrasive particle detection typically utilizes signal processing algorithms, combined with abrasive particle signal models and identification indicators, to extract abrasive particle feature information. However, during processing, the sampled signal often needs to undergo low-pass filtering and harmonic cancellation before weighted analysis to determine the location of the abrasive particle signal and complete extraction and counting. Frequent processing, especially harmonic cancellation, can affect the waveform and amplitude of the abrasive particle signal, even destroying it. To address this, some researchers have proposed adding a reference coil near the detection coil to collect background noise, combined with signal processing techniques, to achieve more accurate abrasive particle signal identification and extraction. However, if the two coils are placed too close, a reverse, single-cycle sinusoidal signal will be generated in the reference coil when the abrasive particle passes through the detection area, affecting the algorithm's detection performance. If the two coils are placed slightly further apart, the correlation between the detection signal and the reference signal will weaken, hindering signal processing. Summary of the Invention

[0004] To address the aforementioned problems, this invention provides a method for extracting abrasive signals based on noise correlation. When the detection coil and the reference coil are close in distance, the position of the abrasive signal is determined based on signal correlation, thereby enabling the identification and counting of the abrasive signal.

[0005] The specific plan includes the following steps:

[0006] S1. Install an inductive abrasive particle detection sensor to collect detection signals and their corresponding reference signals;

[0007] S2. Select the cutoff frequency and perform low-pass filtering on the detection signal and the reference signal respectively to obtain the detection low-frequency harmonic and the reference low-frequency harmonic;

[0008] S3. Divide the detected low-frequency harmonics and the reference low-frequency harmonics into L segments of detected waves and L segments of reference waves, respectively, and calculate the waveform similarity between each segment of detected waves and its corresponding reference wave.

[0009] S4. Set a waveform similarity threshold, filter out waveform similarities not greater than the waveform similarity threshold, and mark the segment position corresponding to the waveform similarity as the suspected abrasive signal segment position;

[0010] S5. The result of subtracting the detection signal from the reference signal is low-pass filtered to obtain the preliminary noise-reduced signal;

[0011] S6. Extract the initial noise reduction signal based on the location of the suspected abrasive particle signal segment to obtain several preprocessed segments;

[0012] S7. Extract the numerical features of each preprocessed segment and establish an ideal abrasive grain signal;

[0013] S8. Using the Pearson correlation coefficient, calculate the matching degree between the preprocessed segment and the ideal abrasive signal, set a threshold, exclude segments below the threshold, and simultaneously complete the extraction and counting of abrasive signals.

[0014] Furthermore, step S1, installing the inductive abrasive particle detection sensor, includes: winding the detection coil and the reference coil onto the detection oil pipe and the reference oil pipe respectively, and connecting them to the amplifier; aligning and paralleling the detection oil pipe and the reference oil pipe in the center of the air gap of the magnetic pole, and ensuring that the detection coil and the reference coil do not directly contact each other when installing the detection oil pipe and the reference oil pipe, with a maximum center distance of 25mm between the detection coil and the reference coil; fixing an iron core below the air gap of the magnetic pole, winding the excitation coil around the iron core and connecting it to a regulated DC power supply.

[0015] Furthermore, in step S2, a cutoff frequency of 1.5 to 2 times the theoretical abrasive signal frequency is selected as the cutoff frequency for low-pass filtering.

[0016] Furthermore, in step S3, the lengths of the detection wave and the reference wave are 0.6 to 0.8 times the theoretical width of the abrasive grain signal.

[0017] Furthermore, the Pearson correlation coefficient is used to calculate the waveform similarity between each detected wave segment and its corresponding reference wave. If the waveform similarity is greater than the waveform similarity threshold, it is considered that the detected wave segment does not contain abrasive signal segments and is set to 0.

[0018] Furthermore, in step S5, a cutoff frequency of 2.5 to 3.5 times the theoretical abrasive signal frequency is selected as the cutoff frequency for low-pass filtering.

[0019] Furthermore, step S6, which processes the preliminary noise reduction signal based on the suspected abrasive particle signal segments, includes: extracting the portion corresponding to each suspected abrasive particle signal segment from the preliminary noise reduction signal and setting the remaining portion to zero.

[0020] The beneficial effects of this invention are:

[0021] A method for locating and segmenting suspected abrasive signal fragments based on the similarity of low-frequency harmonics in the detection and reference signals is proposed. Both signals are low-pass filtered at a low cutoff frequency (1.5–2 times the theoretical abrasive signal frequency) to reveal the waveform trend of the low-frequency harmonics. The processed low-frequency signal is then divided into L segments, each with a length of M, where M should be (0.6–0.8) times the width of the ideal abrasive signal to effectively prevent the abrasive signal from being segmented into too many fragments. The similarity between the corresponding detection and reference signal fragments is calculated using the Pearson correlation coefficient. If a fragment contains an abrasive signal, its correlation coefficient should be below a set threshold. Fragments with correlation coefficients higher than the threshold are filtered out to obtain the suspected abrasive signal fragments. Compared to conventional abrasive signal location and extraction methods, this method does not need to consider the impact on the reference signal when the detection coil and reference coil are too close. It can reliably segment the fragment containing the abrasive signal.

[0022] A detection signal filtering method based on the similarity of low-frequency harmonics between the detection signal and the reference signal is proposed. The detection signal is subtracted from the reference signal, and the result is then low-pass filtered (the cutoff frequency is 2.5–3.5 times the theoretical abrasive signal frequency). Compared to traditional methods that perform weighted filtering and other noise reduction processing after low-pass filtering and harmonic cancellation to remove high-frequency interference, this method better preserves the characteristics of the abrasive signal and avoids the impact of harmonic cancellation on the target signal. Attached Figure Description

[0023] Figure 1 This is a flowchart of the noise correlation-based abrasive signal extraction method of the present invention;

[0024] Figure 2 This is a schematic diagram of the sensor used in this invention;

[0025] Figure 3 This is a cross-sectional view of the sensor used in this invention;

[0026] Figure 4 This is a time-domain diagram of the detection signal and reference signal output by the acquisition card in an embodiment of the present invention;

[0027] Figure 5 This is a waveform trend diagram of the low-frequency signal after low-pass filtering according to an embodiment of the present invention;

[0028] Figure 6The result of the suspected abrasive particle signal fragment remaining after filtering out signal fragments higher than the waveform similarity threshold in an embodiment of the present invention is reflected in the preliminary noise reduction signal;

[0029] Figure 7 This is an abrasive signal diagram obtained by threshold filtering after calculating the matching degree between the embodiment of the present invention and the ideal abrasive signal;

[0030] Among them, 1-detection coil, 2-reference coil, 3-detection oil pipe, 4-reference oil pipe, 5-excitation coil, 6-magnetic pole, 7-iron core. Detailed Implementation

[0031] The technical solutions of 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 some embodiments of the present invention, and not all 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.

[0032] This invention proposes a method for extracting abrasive grain signals based on noise correlation, such as... Figure 1 As shown, it includes the following steps:

[0033] S1. Install an inductive abrasive particle detection sensor to collect the detection signal and a corresponding reference signal of the same length.

[0034] Specifically, in step S1, an inductive abrasive particle detection sensor is installed, the structure of which is as follows: Figure 2 , Figure 3 The following are included: the detection coil 1 and the reference coil 2 are wound around the detection oil pipe 3 and the reference oil pipe 4 respectively, and connected to the amplifier; the detection oil pipe 3 and the reference oil pipe 4 are aligned and placed parallel to each other in the center of the air gap of the magnetic pole 6, and an iron core 7 is fixed below the air gap of the magnetic pole 6. The excitation coil 5 is wound around the iron core 7 and connected to the regulated DC power supply.

[0035] Specifically, when installing the detection oil pipe 3 and the reference oil pipe 4, ensure that the detection coil 1 and the reference coil 2 do not come into direct contact, and the maximum center distance between the detection coil 1 and the reference coil 2 is 25mm.

[0036] Specifically, in one embodiment, the signal sampling frequency f is set. s =5000Hz, sampling time T=15s, the acquired detection signal and reference signal are as follows Figure 4 As shown.

[0037] S2. Select the cutoff frequency and perform low-pass filtering on the detection signal and the reference signal respectively to obtain the detection low-frequency harmonic and the reference low-frequency harmonic.

[0038] Specifically, this invention sets a relatively low cutoff frequency to perform low-pass filtering on both the detection signal and the reference signal, thereby revealing the waveform trends of their low-frequency harmonics. In this embodiment, the relatively low cutoff frequency is set to 120 Hz. Figure 4 The acquired signal is low-pass filtered to obtain, as shown Figure 5 The image shows the detected low-frequency harmonics and the reference low-frequency harmonics.

[0039] S3. Divide the detected low-frequency harmonics and the reference low-frequency harmonics into L segments of detection wave and L segments of reference wave, respectively, and calculate the waveform similarity between each segment of detection wave and its corresponding reference wave.

[0040] Specifically, assuming the lengths of both the detection signal and the reference signal are N>1, they are each divided into L segments of detection wave and L segments of reference wave, with each segment having a length of M, where N = L × M. The length M of the detection wave and the reference wave should be a times the theoretical width of the abrasive signal. The larger M is, the lower the probability of the abrasive signal being segmented, but a should not be less than 0.5 to prevent the abrasive signal from being divided into too many segments. Therefore, this invention sets the range of a to 0.6–0.8.

[0041] In this embodiment, the lengths of the detection signal and the reference signal are both 75000. The ideal abrasive signal width is about 63, so the length of each segment M is 50, and both signals are divided into L = 1500 parts.

[0042] S4. Set a waveform similarity threshold, filter out waveform similarities not greater than the waveform similarity threshold, and mark the segment position corresponding to the waveform similarity (i.e., the position of a segment of the detection wave used to calculate the waveform similarity in the detection signal) as the suspected abrasive signal segment position.

[0043] Specifically, the Pearson correlation coefficient is used to calculate the waveform similarity between each detected wave segment and its corresponding reference wave. If the waveform similarity is greater than the waveform similarity threshold, the detected wave segment is considered not to contain abrasive signal fragments, and is set to 0 and filtered out, thus achieving the localization and segmentation of suspected abrasive signal fragments. In this embodiment, the waveform similarity threshold is set to 0.9.

[0044] S5. The result of subtracting the detection signal from the reference signal is low-pass filtered to obtain the preliminary noise-reduced signal.

[0045] Specifically, the detection signal acquired in step S1 is subtracted from the reference signal, and the subtraction result is low-pass filtered to achieve preliminary noise reduction. In step S5, a cutoff frequency of 2.5 to 3.5 times the theoretical abrasive signal frequency is selected as the cutoff frequency for low-pass filtering. In this embodiment, the cutoff frequency is set to 160 Hz.

[0046] S6. Process the preliminary noise reduction signal based on the suspected abrasive particle signal segments to obtain several preprocessed segments.

[0047] Specifically, in the initial noise reduction signal, the portion corresponding to each suspected abrasive particle signal segment is extracted, and the remaining portion is set to zero. The processing result of this embodiment is as follows: Figure 6 As shown.

[0048] S7. Extract the numerical features of each preprocessed segment and establish an ideal abrasive signal.

[0049] S8. Using the Pearson correlation coefficient, calculate the matching degree between the preprocessed segment and its corresponding segment on the ideal abrasive grain signal, and set a threshold to exclude segments below the threshold. Simultaneously, complete the extraction and counting of the abrasive grain signal. The processing results of this embodiment are as follows: Figure 7 As shown.

[0050] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "setting," "connection," "fixing," "rotation," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal connection of two components or the interaction between two components. Unless otherwise explicitly limited, those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0051] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for extracting abrasive grain signals based on noise correlation, characterized in that, Includes the following steps: S1. Install an inductive abrasive particle detection sensor to collect detection signals and their corresponding reference signals; S2. Select the cutoff frequency and perform low-pass filtering on the detection signal and the reference signal respectively to obtain the detection low-frequency harmonic and the reference low-frequency harmonic; A low-pass filter is performed using a cutoff frequency that is 1.5 to 2 times the theoretical abrasive signal frequency. S3. Divide the detected low-frequency harmonics and the reference low-frequency harmonics into L segments of detection wave and L segments of reference wave, respectively, and calculate the waveform similarity between each segment of detection wave and its corresponding reference wave; the segment length of the detection wave and the reference wave is 0.6 to 0.8 times the theoretical abrasive signal width; S4. Set a waveform similarity threshold, filter out waveform similarities not greater than the waveform similarity threshold, and mark the segment position corresponding to the waveform similarity as the suspected abrasive signal segment position; S5. The result of subtracting the detection signal from the reference signal is low-pass filtered to obtain a preliminary noise-reduced signal; 2.5 to 3.5 times the theoretical abrasive signal frequency is selected as the cutoff frequency for low-pass filtering; S6. Extract the initial noise reduction signal based on the location of the suspected abrasive particle signal segment to obtain several preprocessed segments; S7. Extract the numerical features of each preprocessed segment and establish an ideal abrasive grain signal; S8. Using the Pearson correlation coefficient, calculate the matching degree between the preprocessed segment and the ideal abrasive signal, set a threshold, exclude segments below the threshold, and simultaneously complete the extraction and counting of abrasive signals.

2. The method for extracting abrasive signals based on noise correlation according to claim 1, characterized in that, Step S1 involves installing the inductive abrasive particle detection sensor, which includes: winding the detection coil and reference coil onto the detection oil pipe and reference oil pipe respectively, and connecting them to the amplifier; aligning and paralleling the detection oil pipe and reference oil pipe in the center of the air gap of the magnetic pole, ensuring that the detection coil and reference coil do not directly contact each other when installing the detection oil pipe and reference oil pipe, with a maximum center distance of 25mm between the detection coil and reference coil; fixing an iron core below the air gap of the magnetic pole, winding the excitation coil around the iron core and connecting it to a regulated DC power supply.

3. The method for extracting abrasive signals based on noise correlation according to claim 1, characterized in that, The Pearson correlation coefficient is used to calculate the waveform similarity between each detected wave segment and its corresponding reference wave. If the waveform similarity is greater than the waveform similarity threshold, it is considered that the detected wave segment does not contain abrasive signal segments and is set to 0.

4. The method for extracting abrasive signals based on noise correlation according to claim 1, characterized in that, Step S6 processes the preliminary noise reduction signal based on the suspected abrasive particle signal segments, including: extracting the portion corresponding to the position of each suspected abrasive particle signal segment from the preliminary noise reduction signal and setting the remaining portion to zero.

Images