Method for detecting and suppressing pulse interference in a mechanical vibration signal

Abstract

This invention relates to the field of signal processing technology, and more particularly to a method for detecting and suppressing pulse interference in mechanical vibration signals. The method includes: acquiring the original vibration signal of a mechanical device and extracting the envelope signal; calculating a local pulse factor sequence and a local variance sequence on the envelope signal using a sliding window, and then weighted and fused the normalized versions of the two to obtain a joint detection index sequence; constructing an adaptive detection threshold based on the median and absolute median difference of the joint detection index sequence, and using this threshold to segment the joint detection index sequence to obtain pulse interference intervals; for sampling points within the pulse interference interval, constructing a truncation interval based on the quantiles of the local window, and replacing sampling points with amplitudes exceeding the truncation interval with the mean of the upper and lower quartiles of the local window to obtain the suppressed vibration signal. This invention can effectively detect and suppress random pulse interference in mechanical vibration signals while preserving true fault characteristics.

Description

Technical Field

[0001] This invention relates to the field of signal processing technology, and in particular to a method for detecting and suppressing pulse interference in mechanical vibration signals. Background Technology

[0002] As a core transmission component in large mechanical equipment such as wind power generators and tunneling machines, the operating status of the gearbox directly affects the safety and service life of the entire machine. Vibration signal-based fault diagnosis technology, due to its advantages such as strong real-time performance and ease of online monitoring, has become a primary means of mechanical condition monitoring and fault early warning. Theoretically, vibration signals from a gearbox in a healthy state contain key fault information such as gear wear, broken teeth, and shaft imbalance.

[0003] However, in actual industrial sites such as coal mining and tunnel excavation, mechanical equipment often operates under harsh conditions of high load and strong impact. This results in a large number of non-stationary, high-energy random pulse interferences mixed in with the vibration signals picked up by sensors. The sources of this interference are complex: the start-up and shutdown of nearby equipment and material impacts can introduce broadband impacts; internal load changes and speed fluctuations can also induce transient responses. These pulse interferences manifest as amplitude abrupt changes and short durations in the time domain; in the frequency domain, they raise the full-band noise floor, generate false spectral peaks, and severely contaminate the original waveform, becoming a major cause of misdiagnosis of mechanical equipment. To suppress pulse interference, researchers have developed various technical approaches. Early methods used kurtosis, peak factor, or dynamic thresholds of "mean ± n times standard deviation" for pulse identification. However, these methods are extremely sensitive to outliers; a single large-amplitude pulse can raise the mean and standard deviation, leading to excessively high thresholds and missed detection of small-amplitude pulses. More importantly, such methods have difficulty distinguishing between random disturbances and periodic transient impacts caused by gear tooth breakage or bearing failure. They are very likely to misjudge fault impacts as disturbances and reject them, resulting in irreversible information loss.

[0004] Another approach involves signal processing techniques for suppression, such as wavelet thresholding or median filtering. Wavelet thresholding can remove Gaussian noise, but it suffers from difficulties in choosing the threshold function and the wavelet basis relies on experience. Median filtering has some suppression effect on isolated pulses, but it is ineffective for pulse trains composed of multiple consecutive points and easily smooths out the transient fault leading edge, leading to waveform distortion. Adaptive filtering algorithms can extract fault features, but their computational complexity is high. Amplitude truncation methods, while simple to implement, directly cut off the pulse amplitude, introducing drastic waveform changes and disrupting the signal's phase continuity and amplitude accuracy. Single-feature methods relying solely on local variance for pulse detection are easily confused with amplitude modulation during equipment speed changes, resulting in a high false detection rate. Regarding threshold determination strategies, traditional methods often use a combination of mean and standard deviation, but their sensitivity to outliers limits their effectiveness. The absolute median deviation, as a robust scaling estimator, has been applied in signal processing, providing a useful reference for constructing detection thresholds resistant to pulse interference.

[0005] In summary, existing pulse interference suppression technologies generally suffer from problems such as difficulty in balancing detection sensitivity and specificity, and insufficient protection of true fault characteristics when dealing with strong pulse interference with large amplitude variations and random locations. How to accurately suppress random pulses while preserving the true fault impact characteristics in vibration signals to the greatest extent possible under strong background interference is a technical bottleneck that urgently needs to be overcome in the field of mechanical fault diagnosis. Summary of the Invention

[0006] To overcome the above shortcomings, this invention provides a method for detecting and suppressing pulse interference in mechanical vibration signals. It aims to improve the existing methods' problems of difficulty in balancing sensitivity and specificity in detecting random pulse interference in mechanical vibration signals under strong pulse interference environments, easy misjudging of real fault impacts as interference and rejecting them, and poor adaptability due to the reliance on empirically set threshold parameters.

[0007] This invention provides the following technical solution: a method for detecting and suppressing pulse interference in mechanical vibration signals, comprising the following steps: S1. Collect the original vibration signal of the mechanical equipment, and perform envelope transformation on the original vibration signal to extract the envelope signal; S2. Calculate the local impulse factor sequence and local variance sequence on the envelope signal using a sliding window; S3. The normalized local pulse factor sequence and the normalized local variance sequence are weighted and fused to obtain a joint detection index sequence; S4. Construct an adaptive detection threshold based on the median and absolute median difference of the joint detection index sequence, and use the threshold to segment the joint detection index sequence to obtain the impulse interference interval; S5. For the sampling points within the pulse interference interval, construct a cutoff interval using the quantiles of the local window; S6. Replace the sampling points whose amplitude exceeds the cutoff interval with the mean of the upper and lower quartiles of the local window to obtain the suppressed vibration signal.

[0008] Preferably, in step S1, the step of extracting the envelope signal specifically includes: The original vibration signal is subjected to Hilbert transform to obtain an analytical signal; The magnitude of the analytic signal is calculated to obtain the envelope signal.

[0009] Preferably, in step S2, the steps of calculating the local impulse factor sequence and the local variance sequence specifically include: Set the length of the sliding window; Align the starting point of the sliding window with the starting point of the envelope signal, and extract the data segment within the current window; Calculate the ratio of the maximum value to the mean absolute value of the data segment within the current window, and use it as the local impulse factor value at the center of the current window. Calculate the variance of the data segment within the current window, and use it as the local variance value at the center position of the current window; The sliding window is moved backward by one sampling point, and the above extraction and calculation steps are repeated until the sliding window has traversed the entire envelope signal to obtain the local impulse factor sequence and the local variance sequence.

[0010] Preferably, in step S3, the step of obtaining the joint detection index sequence specifically includes: Find the minimum and maximum values ​​of the local pulse factor sequence, and perform minimum-maximum normalization on the local pulse factor sequence to obtain the normalized pulse factor sequence. Find the minimum and maximum values ​​of the local variance sequence, and perform min-max normalization on the local variance sequence to obtain the normalized variance sequence; Set the pulse factor weight and variance weight; The normalized pulse factor sequence is multiplied by the pulse factor weight, the normalized variance sequence is multiplied by the variance weight, and the products of the two are added together to obtain the joint detection index sequence.

[0011] Preferably, in step S4, the step of constructing the adaptive detection threshold specifically includes: All elements in the joint detection index sequence are sorted in ascending order of numerical value, and the value of the element in the middle position after sorting is taken as the median. Calculate the absolute difference between each element in the joint detection index sequence and the median to obtain the absolute deviation sequence, and then take the median of the absolute deviation sequence as the absolute median difference; The adaptive detection threshold is obtained by adding the median and the absolute median difference multiplied by a preset multiplier.

[0012] Preferably, in step S4, the step of segmenting the joint detection index sequence to obtain the impulse interference interval specifically includes: Each element in the joint detection index sequence is compared with the adaptive detection threshold. Elements that are greater than or equal to the threshold are marked as pulse candidate points, and elements that are less than the threshold are marked as non-pulse points. The position in the joint detection index sequence from a non-pulse point to a pulse candidate point is taken as the pulse start point, and the position from a pulse candidate point to a non-pulse point is taken as the pulse end point. The interval defined by each pair of adjacent pulse start points and pulse end points is taken as a pulse interference interval.

[0013] Preferably, in step S5, the step of constructing a cutoff interval using the quantiles of a local window for sampling points within the pulse interference interval specifically includes: For each sampling point within the pulse interference interval, a local window centered on that sampling point is determined; Calculate the lower quartile and upper quartile of all sampling points within the local window; Calculate the difference between the upper quartile and the lower quartile, and use it as the interquartile range; The lower truncation limit is the product of the lower quartile minus the preset truncation coefficient and the interquartile range. The upper quartile plus the product of the preset truncation coefficient and the interquartile range is used as the upper truncation limit, and the lower truncation limit and the upper truncation limit together constitute the truncation interval.

[0014] Preferably, in step S6, the step of obtaining the suppressed vibration signal specifically includes: For each sampling point within the pulse interference interval, obtain its amplitude and the upper and lower quartiles of the local window corresponding to that sampling point; Determine whether the amplitude of the sampling point is less than the lower cutoff limit or greater than the upper cutoff limit; If the judgment result is yes, then the amplitude of the sampling point is replaced with the arithmetic mean of the lower quartile and the upper quartile of the local window; If the result is negative, the amplitude of that sampling point remains unchanged; After traversing all sampling points within the pulse interference range, the signal after replacement processing is used as the suppressed vibration signal.

[0015] The present invention has the following beneficial effects: 1. In this invention, pulses are comprehensively identified from two dimensions, impact intensity and energy fluctuation, by jointly detecting pulse factor and variance, overcoming the limitations of a single indicator and significantly reducing the probability of false detection and missed detection.

[0016] 2. In this invention, the threshold is constructed using the median and the absolute median difference, which is insensitive to large-amplitude pulses in the signal and can maintain stable detection performance under strong pulse interference environment, thus solving the problem that traditional thresholds are easily skewed.

[0017] 3. In this invention, detection is performed in the envelope domain. The envelope signal extracted using Hilbert envelope transform can highlight the impulse characteristics of the pulse, suppress the interference of the carrier component, and further improve the detection reliability. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating a method for detecting and suppressing pulse interference in mechanical vibration signals proposed in this invention. Figure 2 This is a schematic diagram of the time-domain waveform of the original signal in the method for detecting and suppressing pulse interference in mechanical vibration signals proposed in this invention. Figure 3 This is a schematic diagram comparing the envelope signal and the original signal in a method for detecting and suppressing pulse interference in mechanical vibration signals proposed in this invention. Figure 4 This is a comparative schematic diagram of the pulse factor curve, variance curve, and joint detection index curve of the method for detecting and suppressing pulse interference in mechanical vibration signals proposed in this invention. Figure 5 This is a schematic diagram comparing the signals before and after pulse suppression in the method for detecting and suppressing pulse interference in mechanical vibration signals proposed in this invention. Detailed Implementation

[0019] 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 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.

[0020] This invention provides a method for detecting and suppressing pulse interference in mechanical vibration signals, such as... Figure 1 As shown, it includes the following steps: S1. Collect the original vibration signal of the mechanical equipment, perform envelope transformation on the original vibration signal, and extract the envelope signal.

[0021] Furthermore, in S1, the step of extracting the envelope signal specifically includes: The original vibration signal is subjected to Hilbert transform to obtain the analytic signal; Calculate the magnitude of the analytic signal to obtain the envelope signal.

[0022] Specifically, the raw vibration signal during the operation of the mechanical equipment is collected and denoted as x(n), where n = 1, 2, ..., N, and N is the signal length. The time-domain waveform of the raw signal is as follows: Figure 2 As shown. The Hilbert envelope transform is performed on the original vibration signal x(n) to extract the envelope signal. The specific process is as follows: The original vibration signal x(n) is subjected to a Hilbert transform to obtain the analytic signal z(n). The formula for calculating the Hilbert transform is: ; in, The expression represents the Hilbert transform operator, and * represents the convolution operation. This is derived from the original vibration signal x(n) and its Hilbert transform result. Together they constitute the analytic signal z(n), whose expression is: ; Where j is the imaginary unit. The magnitude of the analytic signal z(n) is calculated to obtain the envelope signal. The calculation formula is: ; Through the above transformation, the high-frequency carrier component in the original vibration signal is suppressed, and the pulse impact characteristics are highlighted, resulting in an envelope signal. Used for subsequent processing. A comparison diagram of the original signal and the envelope signal is shown below. Figure 4 As shown, the top image represents the original signal, and the bottom image represents the envelope signal. From... Figure 4 It can be seen that the envelope signal can effectively extract the impact envelope of the original signal, highlight the pulse characteristics, and lay the foundation for subsequent detection.

[0023] S2. Calculate the local impulse factor sequence and local variance sequence on the envelope signal using a sliding window.

[0024] Furthermore, in S2, the steps for calculating the local impulse factor sequence and the local variance sequence specifically include: Set the length of the sliding window; Align the starting point of the sliding window with the starting point of the envelope signal, and extract the data segment within the current window; Calculate the ratio of the maximum value to the mean absolute value of the data segment within the current window, and use it as the local impulse factor value at the center of the current window. Calculate the variance of the data segment within the current window, and use it as the local variance value at the center position of the current window; Move the sliding window back one sampling point and repeat the above extraction and calculation steps until the sliding window has traversed the entire envelope signal to obtain the local impulse factor sequence and the local variance sequence.

[0025] Specifically, based on the envelope signal obtained in step S1 The local impulse factor sequence and local variance sequence are calculated using the sliding window technique. The specific process is as follows: The length of the sliding window is set to L. The starting point of the sliding window is then compared with the envelope signal. Align with the starting point, extract the data segment within the current window, and denot it as... , where j is the starting position of the window.

[0026] For the data segment within the current window Calculate the ratio of its maximum value to its absolute mean, and use this ratio as the local impulse factor value at the current window center position. Local impulse factor The calculation formula is: ; in, The maximum value of the data within the current window. It is the average of the absolute values ​​of the data within the current window. This indicates the index of the current window's center position within the envelope signal.

[0027] For data segments within the same current window Calculate its variance, and use it as the local variance value for the current window center position. Local variance The calculation formula is: ; in, The number of data points within the window, i.e., the window length L. This is the arithmetic mean of the data within the current window.

[0028] After completing the calculation of the current window, the sliding window is moved back one sampling point, that is, the new window starts at j+1. The above steps of extracting data segments, calculating local impulse factor values ​​and calculating local variance values ​​are repeated until the sliding window has traversed the entire envelope signal.

[0029] After the traversal is completed, the local impulse factor sequence is obtained. and local variance sequences The lengths of both sequences are equal to the length N of the envelope signal, where each of the two ends of the sequence has a length equal to the length N of the envelope signal. The values ​​at each point are the results of the half-window calculation.

[0030] S3. The normalized local pulse factor sequence and the normalized local variance sequence are weighted and fused to obtain the joint detection index sequence.

[0031] Furthermore, in S3, the steps for obtaining the joint detection index sequence specifically include: Find the minimum and maximum values ​​of the local pulse factor sequence, and perform minimum-maximum normalization on the local pulse factor sequence to obtain the normalized pulse factor sequence. Find the minimum and maximum values ​​of the local variance sequence, and perform min-max normalization on the local variance sequence to obtain the normalized variance sequence. Set the pulse factor weight and variance weight; The normalized pulse factor sequence is multiplied by the pulse factor weight, the normalized variance sequence is multiplied by the variance weight, and the products of the two are added together to obtain the joint detection index sequence.

[0032] Specifically, based on the local pulse factor sequence obtained in step S2 and local variance sequences After normalization, the data are weighted and fused to obtain a joint detection index sequence. The specific process is as follows: First, the local pulse factor sequence is processed... Perform min-max normalization to identify the local impulse factor sequence. minimum value and maximum value Normalized pulse factor sequence The calculation formula is: ; in, It is a very small positive number, and its range is 1. to This is used to prevent the denominator from being zero.

[0033] Secondly, for local variance sequences Perform min-max normalization to identify the local variance sequence. minimum value and maximum value Normalized variance sequence The calculation formula is: ; in, It is also a very small positive number.

[0034] Then, set the pulse factor weights. and variance weight The two weights satisfy + =1. Finally, normalize the pulse factor sequence. With pulse factor weight Multiplying the normalized variance sequence Variance weights Multiply them, then add the products together to obtain the joint detection index sequence. The calculation formula is: ; Through the above weighted fusion, the joint detection indicator sequence It simultaneously reflects information in two dimensions: local pulse impact intensity and local energy fluctuation. Its length is the same as that of the local pulse factor sequence and the local variance sequence, both being the length N of the envelope signal.

[0035] Figure 3 Pulse factor curves are shown variance curve and joint detection index curve A comparison. From Figure 4 It can be seen that: It is sensitive to large-amplitude impulses, but it also responds to some small fluctuations. It is sensitive to sudden energy changes, but less responsive to single pulse points. sharp; It combines the advantages of both, retaining the Sensitivity to pulses, and also incorporates The ability to characterize energy fluctuations is more prominent in the pulse region, while fluctuations in the non-pulse region are effectively suppressed.

[0036] It combines the advantages of both, retaining the Sensitivity to pulses, and also incorporates The ability to characterize energy fluctuations is more prominent in the pulse region, while fluctuations in the non-pulse region are effectively suppressed. S4. Construct an adaptive detection threshold based on the median and absolute median difference of the joint detection index sequence, and use this threshold to segment the joint detection index sequence to obtain the impulse interference interval.

[0037] Furthermore, in S4, the steps for constructing the adaptive detection threshold specifically include: All elements in the joint detection index sequence are sorted in ascending order of numerical value, and the value of the element in the middle position after sorting is taken as the median. Calculate the absolute difference between each element in the joint detection index sequence and the median to obtain the absolute deviation sequence, and then take the median of the absolute deviation sequence as the absolute median difference; The adaptive detection threshold is obtained by adding the median and the absolute median difference multiplied by a preset multiplier.

[0038] Furthermore, in S4, the step of segmenting the joint detection index sequence to obtain the impulse interference interval specifically includes: Each element in the joint detection index sequence is compared with an adaptive detection threshold. Elements that are greater than or equal to the threshold are marked as pulse candidate points, and elements that are less than the threshold are marked as non-pulse points. The position in the joint detection index sequence where a non-pulse point jumps to a pulse candidate point is taken as the pulse start point, and the position where a pulse candidate point jumps to a non-pulse point is taken as the pulse end point. The interval defined by the start and end points of each pair of adjacent pulses is considered as a pulse interference interval.

[0039] Specifically, the joint detection index sequence obtained in step S3 First, an adaptive detection threshold is constructed, and then this threshold is used to segment the joint detection index sequence to obtain the impulse interference interval. The specific process is as follows: the joint detection index sequence is... All elements are sorted in ascending order of value, and the value of the middle element in the sorted sequence is taken as the median. If the sequence length is odd, the median is the value of the middle element; if the sequence length is even, the median is the arithmetic mean of the two middle elements. Calculate the joint detection index sequence. Each element and the median The absolute difference is used to obtain the absolute deviation sequence, and then the median of this absolute deviation sequence is taken as the absolute median difference. The calculation formula is: ; median With absolute median The adaptive detection threshold T is obtained by multiplying the values ​​by a preset multiplier and then summing them. The calculation formula is as follows: ; Using an adaptive detection threshold T to analyze the joint detection index sequence Perform binarization segmentation. Separate the joint detection index sequence. Each element in the joint detection index sequence is compared with an adaptive detection threshold T. Elements greater than or equal to the threshold T are marked as pulse candidate points, and elements less than the threshold T are marked as non-pulse points. The position in the joint detection index sequence where a non-pulse point transitions into a pulse candidate point is taken as the pulse start point. The position where a pulse candidate point transitions into a non-pulse point is taken as the pulse end point. The starting point of each group of adjacent pulses. and pulse end point The defined interval is considered as a pulse interference interval. All pulse interference intervals constitute a pulse position matrix, where each row corresponds to a pulse interference interval, the first column is the starting point position, and the second column is the ending point position.

[0040] S5. For sampling points within the pulse interference interval, construct the cutoff interval using the quantiles of the local window.

[0041] Furthermore, in S5, the step of constructing a cutoff interval based on the quantiles of the local window for sampling points within the impulse interference interval specifically includes: For each sampling point within the pulse interference interval, determine a local window centered on that sampling point; Calculate the lower and upper quartiles of all sampling points within the local window; Calculate the difference between the upper quartile and the lower quartile, and use it as the interquartile range; The lower cutoff limit is obtained by subtracting the product of the preset cutoff coefficient and the interquartile range from the following quartiles; The product of the above quartiles, the preset truncation coefficient, and the interquartile range is used as the upper truncation limit. The lower truncation limit and the upper truncation limit together constitute the truncation interval.

[0042] Specifically, based on the pulse interference interval obtained in step S4, for each sampling point within the pulse interference interval, a quantile truncation method is used to construct a truncation interval. The specific process is as follows: For each sampling point within the pulse interference interval, a local window is determined with that sampling point as the center and a preset length as the window radius. Let the index of the current sampling point be i, and the one-sided length of the local window be W, then the range of the local window is... The number of data points within the window is 2W+1. Calculate the lower quartiles of all sampled points within this local window. and upper quartiles Lower quartiles The upper quartile is the value at the 25th percentile after the data within the local window is sorted in ascending order. The value is located at the 75th percentile. Calculate the upper quartile. and lower quartile The difference is used as the interquartile range. The calculation formula is: ; Based on the current quartiles Subtract the preset cutoff coefficient c and interquartile range The product of the products is used as the lower cutoff limit. The above quartiles Add the preset cutoff coefficient c and interquartile range The product of the products is used as the upper limit of the cutoff. The calculation formula is: ; ; For each sampling point within the pulse interference interval, repeat the above process to construct its corresponding cutoff interval.

[0043] S6. Replace the sampling points whose amplitude exceeds the cutoff interval with the mean of the upper and lower quartiles of the local window to obtain the suppressed vibration signal.

[0044] Furthermore, in S6, the step of obtaining the suppressed vibration signal specifically includes: For each sampling point within the pulse interference interval, obtain its amplitude and the upper and lower quartiles of the local window corresponding to that sampling point; Determine whether the amplitude of the sampling point is less than the lower cutoff limit or greater than the upper cutoff limit; If the judgment result is yes, then the amplitude of the sampling point is replaced with the arithmetic mean of the lower quartile and the upper quartile of the local window; If the result is negative, the amplitude of that sampling point remains unchanged; After traversing all sampling points within the pulse interference range, the signal after replacement processing is used as the suppressed vibration signal.

[0045] Specifically, based on the pulse interference interval obtained in step S4 and the truncation interval constructed in step S5, the sampling points within the pulse interference interval are judged and replaced point by point to obtain the suppressed vibration signal. The specific process is as follows: For each sampling point within the pulse interference interval, the original amplitude x(i) of the sampling point and the lower quartile of the local window corresponding to the sampling point are obtained. and upper quartiles And obtain the lower cutoff limit constructed in step S5. and cutoff upper limit Determine whether the amplitude at the sampling point is less than the lower cutoff limit. or greater than the cutoff limit If the judgment result is yes, that is... or If the sampling point is determined to be a pulse interference point, its amplitude is replaced with the lower quartile of the local window. and the upper quartile The arithmetic mean of the substitution value is calculated using the following formula: ; If the judgment result is negative, that is If the sampling point is normal, its amplitude is kept constant. ; After traversing all sampling points within the pulse interference interval, all replaced sampling points are combined with the original sampling points outside the interval in their original order to obtain the suppressed vibration signal. A comparison of the signals before and after pulse suppression is shown below. Figure 5 As shown. From Figure 5 It can be seen that the large amplitude pulses in the original signal are effectively suppressed, and the background trend and detailed features of the signal are well preserved, proving the effectiveness of the invention.

[0046] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for detecting and suppressing pulse interference in mechanical vibration signals, characterized in that, Includes the following steps: S1. Collect the original vibration signal of the mechanical equipment, and perform envelope transformation on the original vibration signal to extract the envelope signal; S2. Calculate the local impulse factor sequence and local variance sequence on the envelope signal using a sliding window; S3. The normalized local pulse factor sequence and the normalized local variance sequence are weighted and fused to obtain a joint detection index sequence; S4. Construct an adaptive detection threshold based on the median and absolute median difference of the joint detection index sequence, and use the threshold to segment the joint detection index sequence to obtain the impulse interference interval; S5. For the sampling points within the pulse interference interval, construct a cutoff interval using the quantiles of the local window; S6. Replace the sampling points whose amplitude exceeds the cutoff interval with the mean of the upper and lower quartiles of the local window to obtain the suppressed vibration signal.

2. The method for detecting and suppressing pulse interference in mechanical vibration signals according to claim 1, characterized in that, In step S1, the step of extracting the envelope signal specifically includes: The original vibration signal is subjected to Hilbert transform to obtain an analytical signal; The magnitude of the analytic signal is calculated to obtain the envelope signal.

3. The method for detecting and suppressing pulse interference in mechanical vibration signals according to claim 1, characterized in that, In S2, the steps of calculating the local impulse factor sequence and the local variance sequence specifically include: Set the length of the sliding window; Align the starting point of the sliding window with the starting point of the envelope signal, and extract the data segment within the current window; Calculate the ratio of the maximum value to the mean absolute value of the data segment within the current window, and use it as the local impulse factor value at the center of the current window. Calculate the variance of the data segment within the current window, and use it as the local variance value at the center position of the current window; The sliding window is moved backward by one sampling point, and the above extraction and calculation steps are repeated until the sliding window has traversed the entire envelope signal to obtain the local impulse factor sequence and the local variance sequence.

4. The method for detecting and suppressing pulse interference in mechanical vibration signals according to claim 1, characterized in that, In step S3, the step of obtaining the joint detection index sequence specifically includes: Find the minimum and maximum values ​​of the local pulse factor sequence, and perform minimum-maximum normalization on the local pulse factor sequence to obtain the normalized pulse factor sequence. Find the minimum and maximum values ​​of the local variance sequence, and perform min-max normalization on the local variance sequence to obtain the normalized variance sequence; Set the pulse factor weight and variance weight; The normalized pulse factor sequence is multiplied by the pulse factor weight, the normalized variance sequence is multiplied by the variance weight, and the products of the two are added together to obtain the joint detection index sequence.

5. The method for detecting and suppressing pulse interference in mechanical vibration signals according to claim 1, characterized in that, In step S4, the step of constructing the adaptive detection threshold specifically includes: All elements in the joint detection index sequence are sorted in ascending order of numerical value, and the value of the element in the middle position after sorting is taken as the median. Calculate the absolute difference between each element in the joint detection index sequence and the median to obtain the absolute deviation sequence, and then take the median of the absolute deviation sequence as the absolute median difference; The adaptive detection threshold is obtained by adding the median and the absolute median difference multiplied by a preset multiplier.

6. The method for detecting and suppressing pulse interference in mechanical vibration signals according to claim 1, characterized in that, In step S4, the step of segmenting the joint detection index sequence to obtain the impulse interference interval specifically includes: Each element in the joint detection index sequence is compared with the adaptive detection threshold. Elements that are greater than or equal to the threshold are marked as pulse candidate points, and elements that are less than the threshold are marked as non-pulse points. The position in the joint detection index sequence from a non-pulse point to a pulse candidate point is taken as the pulse start point, and the position from a pulse candidate point to a non-pulse point is taken as the pulse end point. The interval defined by each pair of adjacent pulse start points and pulse end points is taken as a pulse interference interval.

7. The method for detecting and suppressing pulse interference in mechanical vibration signals according to claim 1, characterized in that, In step S5, the step of constructing a cutoff interval using the quantiles of a local window for sampling points within the pulse interference interval specifically includes: For each sampling point within the pulse interference interval, a local window centered on that sampling point is determined; Calculate the lower quartile and upper quartile of all sampling points within the local window; Calculate the difference between the upper quartile and the lower quartile, and use it as the interquartile range; The lower truncation limit is the product of the lower quartile minus the preset truncation coefficient and the interquartile range. The upper quartile plus the product of the preset truncation coefficient and the interquartile range is used as the upper truncation limit, and the lower truncation limit and the upper truncation limit together constitute the truncation interval.

8. The method for detecting and suppressing pulse interference in mechanical vibration signals according to claim 7, characterized in that, In step S6, the step of obtaining the suppressed vibration signal specifically includes: For each sampling point within the pulse interference interval, obtain its amplitude and the upper and lower quartiles of the local window corresponding to that sampling point; Determine whether the amplitude of the sampling point is less than the lower cutoff limit or greater than the upper cutoff limit; If the judgment result is yes, then the amplitude of the sampling point is replaced with the arithmetic mean of the lower quartile and the upper quartile of the local window; If the result is negative, the amplitude of that sampling point remains unchanged; After traversing all sampling points within the pulse interference range, the signal after replacement processing is used as the suppressed vibration signal.

Images