Oscilloscope signal data processing method and system for bearing detection

Abstract

This invention relates to the field of data processing, and more particularly to an oscilloscope signal data processing method and system for bearing testing. The method involves acquiring bearing vibration data using an oscilloscope, extracting data from a first time moment for frequency domain analysis, obtaining its power spectrum, and dividing it into frequencies to obtain a first frequency sequence. Similarly, a second frequency sequence is obtained from the data at a second time moment. Aligning the two frequency sequences yields frequency matching pairs. Utilizing the frequency variations of neighboring frequencies within a single matching pair, the noise figure of each frequency in the matching pair is calculated, resulting in the noise figure for any frequency in the first frequency sequence. Based on these noise figures, a first mean sequence is constructed. The distribution consistency of each frequency point in the first mean sequence is obtained, and the segmentation effectiveness of each frequency point is calculated. Combined with a noise segmentation threshold, oscilloscope signal processing is completed. This invention denoises bearing vibration data acquired by an oscilloscope, improving the accuracy of the denoised oscilloscope signal data.

Description

Technical Field

[0001] This invention relates to the field of data processing, and more particularly to an oscilloscope signal data processing method and system for bearing testing. Background Technology

[0002] In bearing condition monitoring and fault diagnosis, the oscilloscope is a key device for acquiring vibration signals. The raw vibration signals contain rich information about the equipment's operating status, especially the subtle characteristics of early bearing faults. However, the acquired signals inevitably contain various background noises, with high-frequency noise, characterized by strong randomness and wide energy distribution, being particularly prominent. This high-frequency noise not only obscures the characteristic frequencies characterizing bearing damage but also masks sideband information generated by modulation effects, severely interfering with the identification and extraction of effective fault features and reducing the accuracy of condition monitoring and early warning capabilities.

[0003] Traditional signal denoising methods, such as fixed-threshold filtering, face significant limitations when processing complex vibration signals. Fixed thresholds are difficult to adapt to the non-stationary characteristics of signals and the dynamic changes in noise intensity. In order to suppress high-frequency noise, fixed thresholds often indiscriminately filter out all frequency components above a certain threshold. Summary of the Invention

[0004] To address the problem of adaptively identifying and eliminating noise frequencies in the processing of bearing vibration signals acquired by oscilloscopes, this invention provides solutions in the following aspects.

[0005] In a first aspect, embodiments of the present invention provide an oscilloscope signal data processing method for bearing testing, the method comprising the following steps:

[0006] Acquire bearing vibration data collected by an oscilloscope, obtain the data at the first moment, perform frequency domain analysis on the data at the first moment to obtain the power spectrum at the first moment, divide the power spectrum at the first moment into frequencies to obtain the first frequency sequence. Similarly, acquire the data at the second moment, and obtain the second frequency sequence of the data at the second moment by following the process of obtaining the first frequency sequence from the data at the first moment.

[0007] Align the first frequency sequence and the second frequency sequence to obtain frequency-aligned matching pairs. Calculate the noise figure of each frequency in a single matching pair by utilizing the frequency changes of neighboring frequencies in the single matching pair. Obtain the noise figure of any frequency in the first frequency sequence from the noise figure of any frequency in each matching pair.

[0008] The noise figure distribution is obtained based on the first frequency sequence. The distribution consistency of each frequency in the first frequency sequence is calculated. The segmentation effectiveness of each frequency in the first frequency sequence is calculated using the distribution consistency of each frequency in the first frequency sequence. The noise segmentation threshold is obtained based on the segmentation effectiveness of each frequency in the first frequency sequence. The oscilloscope signal processing is completed based on the noise segmentation threshold.

[0009] Preferably, the first moment data includes: acquiring bearing vibration signals using the vibration signal acquisition probe of an oscilloscope, converting the bearing vibration signals into oscilloscope signal data using analog-to-digital conversion; obtaining the latest moment as the first moment, and obtaining oscilloscope signal data within a certain time interval from the first moment as the first moment data.

[0010] Preferably, the first frequency sequence includes: acquiring the power spectrum of the data at the first time moment; applying a mean filtering algorithm to filter the power values ​​in the power spectrum to obtain a filtered power value sequence; applying a peak point detection algorithm to detect peak points in the filtered power value sequence to obtain the number of peak points; using the number of peak points as the preset number of Gaussian functions to be fitted in the Gaussian mixture model; obtaining multiple Gaussian functions of the filtered power value sequence through the Gaussian mixture model algorithm; acquiring the frequency corresponding to the mean of each Gaussian function in the data at the first time moment; and sorting the filtered power value sequence to obtain the first frequency sequence.

[0011] Preferably, the noise figure of any frequency in the first frequency sequence includes: obtaining the previous time as the second time, and obtaining the second time data; obtaining the second frequency sequence of the second time data, and obtaining the first frequency sequence and the second frequency sequence; using the DTW algorithm to obtain the registration result between the first frequency sequence and the second frequency sequence, wherein one registration result is a matching pair; obtaining the KL divergence value between the Gaussian function corresponding to the frequency in the first frequency sequence and the Gaussian function corresponding to the frequency in the second frequency sequence in any matching pair; obtaining the mean of the KL divergence value corresponding to each frequency in the first frequency sequence; obtaining the weighted average of the mean of the KL divergence values ​​corresponding to all frequencies in the first frequency sequence; obtaining the ratio of the mean of the KL divergence value corresponding to each frequency in the first frequency sequence to the weighted average of the mean of the KL divergence values ​​corresponding to all frequencies; and combining the power value of that frequency in the first frequency sequence with the negative correlation mapping to obtain the noise figure of each frequency in the first frequency sequence.

[0012] Preferably, the segmentation effectiveness of each frequency in the first frequency sequence includes: sorting the noise coefficients corresponding to each frequency in the first frequency sequence in ascending order to obtain a sorted noise coefficient sequence; obtaining all frequencies in the first frequency sequence lower than a single frequency, obtaining the sorted noise coefficient sequences corresponding to each single frequency at each moment within a certain time period, and combining them into a final noise coefficient sequence; using an envelope acquisition method to obtain the upper and lower envelopes of the final noise coefficient sequence, and using the Pearson correlation coefficient to calculate the Pearson coefficient value between the upper and lower envelopes as the distribution consistency of all frequencies lower than a single frequency in the first frequency sequence; similarly, the distribution consistency of all frequencies greater than or equal to a single frequency in the first frequency sequence can be obtained; calculating the difference between the distribution consistency of all frequencies greater than or equal to a single frequency in the first frequency sequence and the distribution consistency of all frequencies less than a single frequency in the first frequency sequence, and combining this with the distribution consistency of all frequencies less than a single frequency in the first frequency sequence, to obtain the segmentation effectiveness of frequencies greater than or equal to a single frequency in the first frequency sequence; similarly, the segmentation effectiveness of each frequency in the first frequency sequence can be obtained.

[0013] Preferably, the step of completing oscilloscope signal processing based on the noise segmentation threshold includes: obtaining the frequency with the maximum value of segmentation effectiveness in the first frequency sequence as the adaptive frequency segmentation threshold, setting the power value corresponding to the frequency in the first moment data that is higher than the adaptive frequency segmentation threshold to 0, retaining the effective frequency in the obtained vibration data, and realizing the processing of oscilloscope vibration signal.

[0014] Preferably, obtaining the registration result between the first frequency sequence and the second frequency sequence includes: using the cumulative distance matrix acquisition process in the DTW algorithm to obtain the cumulative distance matrix between the first frequency sequence and the second frequency sequence, and filling the cumulative distance matrix; after filling the cumulative distance matrix, obtaining the registration result between the first frequency sequence and the second frequency sequence through the registration path process in the DTW algorithm.

[0015] Secondly, the present invention provides an oscilloscope signal data processing system for bearing testing, characterized in that it includes: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned oscilloscope signal data processing method for bearing testing is implemented.

[0016] The beneficial effects of the embodiments of the present invention compared with the prior art are as follows:

[0017] 1. By calculating the noise figure, the abnormal fluctuation of frequency points relative to the overall sequence is highlighted and low-power points are given a higher suspicion of noise. The dual mechanism of noise figure improves the identification accuracy of distinguishing low-power effective signals from real noise.

[0018] 2. By analyzing the characteristics of the regions on both sides of each frequency point, the overall pattern of low-frequency noise figure changes over multiple time intervals is evaluated to determine whether it remains stable and consistent. Simultaneously, based on the different fluctuation characteristics of the high-frequency band's variation pattern, the segmentation quality of each frequency point is assessed. This enables the automatic locking of the segmentation frequency that best distinguishes between stable low-frequency signals and high-frequency fluctuating noise, improving the accuracy of noise identification and removal in bearing vibration data acquired by the oscilloscope. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart of an oscilloscope signal data processing method for bearing testing provided in an embodiment of the present invention.

[0021] Figure 2 This is a system flowchart of an oscilloscope signal data processing system for bearing testing provided in an embodiment of the present invention. Detailed Implementation

[0022] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0023] It should be noted that the terms "first," "second," etc., used in the specification and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this invention. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this invention.

[0024] To illustrate the technical solution of the present invention, specific embodiments are described below.

[0025] See Figure 1 This is a flowchart of an oscilloscope signal data processing method for bearing testing provided in an embodiment of the present invention. Figure 1 As shown, the method may include:

[0026] Step S1: Acquire bearing vibration data collected by oscilloscope, acquire data at the first moment, perform frequency domain analysis on the data at the first moment to obtain the power spectrum at the first moment, divide the power spectrum at the first moment into frequencies to obtain the first frequency sequence. Similarly, acquire data at the second moment, and obtain the second frequency sequence of the data at the second moment by following the process of acquiring the first frequency sequence from the data at the first moment.

[0027] An oscilloscope with vibration signal detection function is selected, and the vibration signal acquisition probe of the oscilloscope is brought into contact with the bearing to acquire the bearing vibration signal. The acquired bearing vibration signal is converted into the bearing vibration signal of the bearing under test by the analog-to-digital converter of the oscilloscope, which is used as the oscilloscope signal data. The oscilloscope signal data is timing data.

[0028] When performing adaptive threshold adjustment on oscilloscope signal data, the adaptive threshold is a frequency threshold. Therefore, frequency domain analysis is performed on the oscilloscope signal data to identify noise frequencies. By targeting the noise frequencies, the erroneous noise reduction of valuable bearing vibration signals during oscilloscope noise reduction processing can be reduced.

[0029] To identify noise frequencies from the frequency spectrum, frequency information is first extracted from the oscilloscope signal data. Time-series data is required for frequency domain analysis, necessitating the acquisition of the current time, which is designated as the first time. The time interval between the current and first time moments is then determined. At a given moment, the oscilloscope signal data acquired during the time interval from that moment to the first moment is considered the first moment data. Take 0.2 seconds to accumulate experience points.

[0030] The short-time Fourier transform algorithm is used to obtain the time-frequency conversion result of the first-time data. The power spectrum of the time-frequency conversion result in the first-time data is calculated by statistical methods. In the power spectrum, the horizontal axis is the frequency value and the vertical axis is the power value corresponding to each frequency. After obtaining the power spectrum of the first-time data, the power values ​​are sorted in ascending order of the corresponding frequency to obtain ordered power values.

[0031] After obtaining the power spectrum of the first moment data, since the frequencies of the low power values ​​in the power spectrum are the vibration frequencies with extremely weak responses, in order to reduce the interference of the frequencies of the low power values ​​in the power spectrum on the subsequent adaptive selection of the noise reduction threshold, the mean filtering algorithm is used to filter the sorted power values ​​to obtain the filtered power value sequence.

[0032] If a single frequency has a high power value, it indicates that the vibration frequency has a significant response intensity, which will cause the neighboring frequencies to also have high power values. Therefore, a Gaussian function is selected to fit the neighboring frequencies with a Gaussian distribution, and a single Gaussian function is used as the frequency distribution of a single frequency in the power spectrum.

[0033] Since the power value sequence in the filtered power spectrum contains multiple frequency values, in order to obtain the Gaussian distribution of the neighboring frequencies corresponding to each frequency in the power spectrum, the Gaussian mixture model algorithm is used to fit the frequency distribution in the power spectrum with multiple Gaussian functions. However, the Gaussian mixture model algorithm requires the number of Gaussian functions to be fitted to be preset in the use of the algorithm.

[0034] To obtain the pre-set number of Gaussian functions to be fitted, a peak point detection algorithm is used to detect peak points in the filtered power value sequence to obtain the number of peak points. The number of peak points is used as the pre-set number of Gaussian functions to be fitted in the Gaussian mixture model. Then, multiple Gaussian functions in the power value sequence in the filtered power spectrum can be obtained through the Gaussian mixture model algorithm. Each Gaussian function contains two parameters: mean and variance. A single Gaussian function represents the frequency distribution characteristics of the neighboring frequencies, and the mean of the corresponding Gaussian function is the mean of the frequency distribution of the neighboring frequencies.

[0035] The frequency corresponding to the mean of each Gaussian function obtained in the data at the first time point can be obtained. The frequencies corresponding to the mean of each Gaussian function obtained in the data at the first time point are sorted in descending order to obtain the frequency sequence corresponding to the mean of the Gaussian function after sorting, which is denoted as the first frequency sequence.

[0036] Let the previous time point be the second time point. Similarly, we can obtain the frequency sequence corresponding to the mean of the Gaussian function after sorting the data at the second time point, which is denoted as the second frequency sequence.

[0037] Step S2: Align the first frequency sequence and the second frequency sequence to obtain frequency-aligned matching pairs. Calculate the noise figure of each frequency in a single matching pair using the frequency changes of neighboring frequencies in the matching pair. Obtain the noise figure of any frequency in the first frequency sequence from the noise figure of any frequency in each matching pair.

[0038] The first and second frequency sequences are obtained from the bearing vibration information at two consecutive moments. Therefore, the effective vibration signal in the two frequency sequences should be approximately the same. However, noise is random and unstable high-frequency information. Therefore, after aligning and registering the two frequency sequences, noise detection is performed based on the distribution changes between the two frequency sequences.

[0039] To align and register two frequency sequences, the cumulative distance matrix between the first and second frequency sequences is obtained using the cumulative distance matrix calculation method in the DTW algorithm. The cumulative distance matrix is ​​then filled. After filling the cumulative distance matrix, the registration result between the first and second frequency sequences is obtained through the registration path process in the DTW (Dynamic Time Warping) algorithm. A registration result consists of a single frequency in the first frequency sequence and a single frequency in the second frequency sequence. A registration result is denoted as a matching pair.

[0040] Because the DTW algorithm can adapt to the length difference between two frequency sequences when aligning data, it will result in one-to-many and one-to-one frequency correspondences in the matching pairs. Since the second time step is the time step before the first time step, it is necessary to perform noise reduction processing on the data at the first time step and calculate the change in the single frequency distribution at the first time step.

[0041] Since the two frequencies in any matching pair are the alignment results from the first frequency sequence and the second frequency sequence, obtaining the difference of the Gaussian function corresponding to the two frequencies in the matching pair can represent the change in the difference between the neighboring frequencies of the two frequencies in the matching pair.

[0042] Obtain the KL divergence value between the Gaussian function corresponding to the frequency in the first frequency sequence and the Gaussian function corresponding to the frequency in the second frequency sequence in any matching pair, which represents the change in the difference between the neighboring frequencies of the two frequencies in the matching pair.

[0043] The larger the KL divergence value, the greater the difference in the distribution of neighboring frequencies between the two frequencies, indicating a significant change in the neighboring frequencies. If the frequencies of the two frequencies are high but the power values ​​are low, they are likely noise frequencies, because noise frequencies are random high-frequency noise and have low power values ​​due to their low proportion in the signal composition. The calculation of the KL divergence value between two Gaussian functions using the KL divergence calculation formula is well-known and will not be elaborated further.

[0044] Because there is a one-to-many relationship in the matched pairs, the first frequency sequence... The frequency will exist in multiple matching pairs, thus the first frequency sequence will have a higher frequency. The frequency has at least one KL divergence value, thus obtaining the first frequency sequence. The mean of the KL divergence values ​​corresponding to the frequencies is .

[0045] The weighted average of the B values ​​for all frequencies in the first frequency sequence can be calculated. This is to demonstrate the degree of variation in the B value of a single frequency in the first frequency sequence relative to the B values ​​of all frequencies.

[0046] The reason for using a weighted average instead of a direct mean calculation is that the noise power is relatively low, causing significant changes in neighboring frequencies compared to the normal frequency. Therefore, the power values ​​of all frequencies in the first frequency sequence are normalized using a normalization algorithm, and the reciprocal of the normalized power value is used as the weight for a weighted average, resulting in the... The calculation process for the weighted average is a well-known technique and will not be described in detail here.

[0047] Obtain the first frequency sequence Frequency power value , obtain reciprocal If Then let reciprocal To prevent the denominator from being zero, calculate the sum of the reciprocals of the power values ​​of all frequencies in the first frequency sequence. Thus, the first frequency sequence is obtained. Frequency weight .

[0048] The first frequency sequence can be obtained. Noise figure of frequency

[0049]

[0050] in, For the first frequency sequence, the first... The frequency corresponds to the mean of the KL divergence values. Since there is a one-to-many relationship in the matching pairs, the first frequency in the first frequency sequence... The frequency will exist in multiple matching pairs, thus the first frequency sequence will have a higher frequency. The frequency has at least one KL divergence value, thus obtaining the first frequency sequence. The mean of the KL divergence values ​​corresponding to the frequencies is .

[0051] The weighted average of the B values ​​for all frequencies in the first frequency sequence. The weighted average is used because the noise frequency and its neighboring frequencies vary significantly, while the normal bearing vibration frequency and its neighboring frequencies vary less significantly. Compared to ratio .

[0052] The larger the value, the higher the value of the first frequency sequence. The average value of the KL divergence corresponding to the matching of a frequency is greater than the average value of the KL divergence corresponding to the matching of each frequency in the first frequency sequence. It has a relatively large frequency and neighboring frequency variation, and thus has a higher probability of belonging to noise frequency.

[0053] If the first frequency sequence is the first frequency sequence Frequency power value The smaller the value, the higher the probability that it belongs to noise. It is an exponential function, which acts as an index on the first frequency sequence. The effect of negative correlation mapping of power values ​​at frequency.

[0054] Step S3: Obtain the noise figure distribution based on the first frequency sequence, calculate the distribution consistency of each frequency in the first frequency sequence, calculate the segmentation effectiveness of each frequency in the first frequency sequence using the distribution consistency of each frequency in the first frequency sequence, obtain the noise segmentation threshold based on the segmentation effectiveness of each frequency in the first frequency sequence, and complete the oscilloscope signal processing based on the noise segmentation threshold.

[0055] After obtaining the noise figure of a single frequency in the first frequency sequence, the frequencies can be divided according to the magnitude of the noise figure of different frequencies in the first frequency sequence. The frequencies with a large noise proportion and their neighboring frequencies are classified as noise frequencies, so that the bearing vibration information acquired by the oscilloscope has less interference information.

[0056] To achieve adaptive frequency threshold partitioning, it is necessary to calculate the partitioning validity for each frequency in the first frequency sequence, and then complete the frequency partitioning based on the partitioning validity.

[0057] First, in the first frequency sequence, the noise figures corresponding to each frequency are sorted in ascending order to obtain the sorted noise figure sequence. Since the frequency of bearing vibration noise is relatively high compared to normal bearing vibration, and since the noise frequency distribution is unstable, the noise figure sequence corresponding to continuous time has a high noise figure and the continuous distribution changes have large differences.

[0058] When measuring the changes in the noise sequence at consecutive moments, since the vibration frequency of the bearing is quite stable when it is working normally, while the noise frequency corresponding to the noise generated by the bearing vibration is unstable, it is necessary to obtain the consistency of the frequency and the distribution of adjacent frequencies at consecutive moments.

[0059] The first frequency sequence is less than the first The specific calculation process for ensuring the uniformity of frequency distribution across all frequencies is as follows:

[0060] Obtain the frequency less than the first frequency in the first frequency sequence All frequencies of the frequency, obtain those frequencies at the 1st Time to the The noise figure sequences corresponding to each time point can be used to obtain 10 noise figure sequences, which are combined into the final noise figure sequence. Here, 9 is a hyperparameter that can be adjusted by the implementer according to the specific implementation scenario.

[0061] Using the envelope calculation method, the upper and lower envelopes of the final noise figure sequence are obtained. The Pearson correlation coefficient is then used to calculate the Pearson coefficient value between the upper and lower envelopes, which is used as the value of the first frequency sequence less than the first... Consistent distribution of all frequencies .

[0062] The Pearson coefficient value indicates the similarity of the fluctuation trends of the upper and lower envelopes of the final noise figure sequence. A larger Pearson coefficient value indicates a more similar fluctuation trend between the upper and lower envelopes. In the first frequency sequence, values ​​smaller than the second... The higher the uniformity of the distribution of all frequencies.

[0063] Similarly, it can be concluded that in the first frequency sequence, frequencies greater than or equal to the first... Consistent distribution of all frequencies .

[0064] The first frequency sequence The effectiveness of frequency segmentation is:

[0065]

[0066] For the first frequency sequence, the first... Effectiveness of frequency segmentation.

[0067] For the first frequency sequence less than the first The distribution of all frequencies is consistent.

[0068] For the first frequency sequence, greater than or equal to the first The distribution of all frequencies is consistent.

[0069] It is an exponential function with base e.

[0070] if Higher at the same time and The greater the difference, the more effective the segmentation. Since the Pearson coefficient can be negative, this leads to... There exist negative values ​​such that and After multiplication The values ​​become disordered, making it impossible to effectively identify noise in the bearing's oscilloscope signal.

[0071] use Exponential function pairs Implement data mapping, It is a function that is always greater than 0 and is monotonically increasing.

[0072] Calculate the segmentation effectiveness of each frequency in the first frequency sequence, and obtain the maximum value of the segmentation effectiveness of each frequency in the first frequency sequence as the adaptive frequency segmentation threshold.

[0073] By setting the power value corresponding to the frequency higher than the adaptive frequency segmentation threshold in the first moment data to 0, noise frequencies are eliminated, and the effective frequencies in the vibration data are retained, thus achieving noise reduction processing of the bearing vibration signal acquired by the oscilloscope.

[0074] This invention also provides an oscilloscope signal data processing system for bearing monitoring. For example... Figure 2 As shown, the system includes a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement the oscilloscope signal data processing method for bearing monitoring according to the first aspect of the present invention. The system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface, the setup and functions of which are known in the art and will not be described further here.

[0075] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. An oscilloscope signal data processing method for bearing testing, characterized in that, The oscilloscope signal data processing method for bearing testing includes: Acquire bearing vibration data collected by an oscilloscope, obtain the data at the first moment, perform frequency domain analysis on the data at the first moment to obtain the power spectrum at the first moment, divide the power spectrum at the first moment into frequencies to obtain the first frequency sequence. Similarly, acquire the data at the second moment, and obtain the second frequency sequence of the data at the second moment by following the process of obtaining the first frequency sequence from the data at the first moment. Align the first frequency sequence and the second frequency sequence to obtain frequency-aligned matching pairs. Calculate the noise figure of each frequency in a single matching pair by utilizing the frequency changes of neighboring frequencies in the single matching pair. Obtain the noise figure of any frequency in the first frequency sequence from the noise figure of any frequency in each matching pair. The noise figure distribution is obtained based on the first frequency sequence. The distribution consistency of each frequency in the first frequency sequence is calculated. The segmentation effectiveness of each frequency in the first frequency sequence is calculated using the distribution consistency of each frequency in the first frequency sequence. The segmentation effectiveness is used to determine the adaptive frequency segmentation threshold. The noise segmentation threshold is obtained based on the segmentation effectiveness of each frequency in the first frequency sequence. The oscilloscope signal processing is completed based on the noise segmentation threshold. The noise figure for any frequency in the first frequency sequence includes: The previous time step is used as the second time step to obtain the data for the second time step. The second frequency sequence of the data at the second time point is obtained to obtain the first frequency sequence and the second frequency sequence. The DTW algorithm is used to obtain the registration result between the first frequency sequence and the second frequency sequence, where one registration result is a matching pair. Obtain the KL divergence value between the Gaussian function corresponding to the frequency in the first frequency sequence and the Gaussian function corresponding to the frequency in the second frequency sequence for any matching pair, and obtain the mean of the KL divergence value corresponding to each frequency in the first frequency sequence. Obtain the weighted average of the mean values ​​of KL divergence values ​​corresponding to all frequencies in the first frequency sequence, obtain the ratio of the mean value of KL divergence values ​​corresponding to each frequency in the first frequency sequence to the weighted average of the mean values ​​of KL divergence values ​​corresponding to all frequencies, and combine it with the power value of that frequency in the first frequency sequence with negative correlation mapping to obtain the noise figure of each frequency in the first frequency sequence.

2. The oscilloscope signal data processing method for bearing testing according to claim 1, characterized in that, The data at the first moment includes: The vibration signal acquisition probe of the oscilloscope is used to acquire the bearing vibration signal, and the bearing vibration signal is converted from analog to oscilloscope signal data. The latest time is taken as the first time, and the oscilloscope signal data within a certain time interval from the first time is taken as the first time data.

3. The oscilloscope signal data processing method for bearing testing according to claim 1, characterized in that, The first frequency sequence includes: The power spectrum of the data at the first moment is obtained, and the power values ​​in the power spectrum are filtered by the mean filtering algorithm to obtain the filtered power value sequence. Peak point detection algorithm is used to detect peak points in the filtered power value sequence to obtain the number of peak points. The number of peak points is used as the preset number of Gaussian functions to be fitted in Gaussian mixture model. Multiple Gaussian functions of the filtered power value sequence are obtained through Gaussian mixture model algorithm. Obtain the frequency corresponding to the mean of each Gaussian function in the data at the first moment, sort the filtered power value sequence, and obtain the first frequency sequence.

4. The oscilloscope signal data processing method for bearing testing according to claim 1, characterized in that, The segmentation validity of each frequency in the first frequency sequence includes: In the first frequency sequence, the noise figures corresponding to each frequency are sorted in ascending order to obtain the sorted noise figure sequence. Obtain all frequencies less than a single frequency in the first frequency sequence, obtain the sorted noise figure sequence corresponding to each time point of a single frequency within a certain time period, and combine them into the final noise figure sequence. Using the envelope acquisition method, the upper and lower envelopes of the final noise figure sequence are obtained. The Pearson correlation coefficient is used to calculate the Pearson coefficient value between the upper and lower envelopes, which serves as the distribution consistency of all frequencies less than a single frequency in the first frequency sequence. Similarly, it can be seen that the distribution of all frequencies greater than or equal to a single frequency in the first frequency sequence is consistent; Calculate the difference between the distribution consistency of all frequencies greater than or equal to a single frequency in the first frequency sequence and the distribution consistency of all frequencies less than a single frequency in the first frequency sequence. Combine this with the distribution consistency of all frequencies less than a single frequency in the first frequency sequence to obtain the segmentation effectiveness of frequencies greater than or equal to a single frequency in the first frequency sequence. Similarly, we can obtain the segmentation validity of each frequency in the first frequency sequence.

5. The oscilloscope signal data processing method for bearing testing according to claim 1, characterized in that, The step of performing oscilloscope signal processing based on the noise segmentation threshold includes: The frequency with the maximum effective frequency segmentation value in the first frequency sequence is obtained as the adaptive frequency segmentation threshold. The power value corresponding to the frequency in the first moment data that is higher than the adaptive frequency segmentation threshold is set to 0, and the effective frequency in the obtained vibration data is retained, thereby realizing the processing of the oscilloscope vibration signal.

6. The oscilloscope signal data processing method for bearing testing according to claim 1, characterized in that, The process of obtaining the registration result between the first frequency sequence and the second frequency sequence includes: By utilizing the cumulative distance matrix acquisition process in the DTW algorithm, the cumulative distance matrix between the first frequency sequence and the second frequency sequence is obtained and filled. After filling the cumulative distance matrix, the registration result between the first frequency sequence and the second frequency sequence is obtained through the registration path process in the DTW algorithm.

7. An oscilloscope signal data processing system for bearing testing, characterized in that, include: A processor and a memory, the memory storing computer program instructions that, when executed by the processor, implement an oscilloscope signal data processing method for bearing detection according to any one of claims 1-6.

Images