Rolling bearing state online monitoring method and system based on stress wave signal analysis

Abstract

The invention discloses a rolling bearing state on-line monitoring method and system based on stress wave signal analysis, and relates to the technical field of rolling bearing state monitoring. Time periods are dynamically calibrated based on the rotating speed, waveband sets are screened through multi-cycle data comparison, signal fluctuation interference is eliminated by using the principle of minimization of the coincidence rate, and the state of a rolling bearing is monitored. It is ensured that the selected wave band completely represents the single-cycle operation characteristics of the bearing, and the error is reduced by 40% or above compared with a traditional fixed window interception method; through coincidence degree verification of a to-be-analyzed wave band, a waveform combination with the maximum characteristic difference is preferentially extracted, information limitation of a single wave band is avoided, time domain characteristic changes of different fault stages are covered, the early fault characteristic extraction integrity is improved by 65%, dominant frequency bands of all wave bands are dynamically locked, and through a double-wave-band characteristic frequency band cross validation mechanism, the fault detection accuracy is improved. The method not only can capture common fault frequency domain characteristics (energy extreme value is taken at the same frequency band), but also can identify asymmetric faults, and the diagnosis accuracy reaches 92.3%.

Description

Technical Field

[0001] The present invention relates to the technical field of rolling bearing condition monitoring, and particularly to an online rolling bearing condition monitoring method and system based on stress wave signal analysis. Background Art

[0002] Rolling bearings, as key components of rotary motion, are widely used in various high-end equipment, such as aero-engines, high-speed railways, wind power, etc. However, due to harsh service environments (long-term loads, friction, impacts, etc.), rolling bearings may develop faults such as fatigue, wear, cracks, and deformation. If the above-mentioned rolling bearing faults cannot be identified early and preventive maintenance cannot be carried out, it will have a significant impact on the healthy service of high-end equipment, and even cause irreparable catastrophic accidents.

[0003] Currently, in order to ensure the service safety of rolling bearings in high-end equipment, most methods adopt regular disassembly and use non-destructive testing technologies such as visual inspection, listening, or ultrasonic inspection. However, there are many inconveniences in actual engineering: (1) The faults of rolling bearings in high-end equipment are random. If a regular disassembly strategy is adopted, it is difficult to detect faults in the early stage in a timely manner; (2) It is difficult to disassemble some rolling bearings in high-end equipment after assembly, that is, the workload and difficulty of regular disassembly are large. Considering that the equipment needs to be shut down during the disassembly process, it is often difficult to implement in actual engineering; (3) Non-destructive testing technologies such as visual inspection, listening, or ultrasonic inspection cannot achieve online monitoring of the health status of rolling bearings in high-end equipment. Therefore, it is still necessary to propose new methods and technologies for online condition monitoring of rolling bearings in high-end equipment.

[0004] Existing condition monitoring methods for rolling bearings in high-end equipment are mainly based on vibration signal analysis, that is, by collecting and analyzing the acceleration signals of rolling bearings, and then realizing online monitoring of the health status. However, it should be noted that vibration signals are not sensitive to early damage of rolling bearings and are significantly affected by environmental noise. Therefore, there are also great limitations in practical engineering applications. In recent years, rolling bearing fault detection methods based on stress wave signals have gradually attracted the attention of the academic and industrial communities. However, it should be noted that most current rolling bearing fault monitoring methods based on stress wave signals focus on low-frequency characteristics, so they are greatly affected by noise, which will significantly reduce the monitoring effect. At the same time, most of them rely on machine learning and deep learning technologies, that is, a large number of fault stress wave signals are required for training. Therefore, it is also difficult to apply them to actual engineering.

[0005] Therefore, it is an urgent problem for those skilled in the art to propose an online rolling bearing condition monitoring method and system that takes into account the actual needs of high-end equipment health management and safety operation and maintenance, and realizes fault diagnosis and identification of rolling bearings in high-end equipment. Summary of the Invention

[0006] Aiming at the deficiencies of the existing technology, a method and system for on-line monitoring of the state of rolling bearings of high-end equipment based on the high-frequency characteristics of stress wave signals are invented to realize the fault diagnosis and identification of rolling bearings of high-end equipment.

[0007] To achieve the above objectives, the present invention is realized through the following technical solutions: An on-line monitoring method for the state of rolling bearings based on stress wave signal analysis, including the following steps:

[0008] Place stress wave monitoring sensors at designated positions of the rolling bearings and monitor the stress waves generated during the rotation of the rolling bearings. From the monitored stress waves, select a band set for a single operating cycle. The specific method is as follows:

[0009] During the stress wave monitoring process, based on the current rotational speed of the rolling bearing, confirm the characteristic time t1 associated with one rotation of the rolling bearing, and keep the current rotational speed and continue to rotate N weeks, where N is a preset value. Confirm the stress waves associated with N weeks from the monitoring data;

[0010] From the confirmed stress waves, calibrate the initial time to 0, and record the band between 0 - t1 as the band to be determined;

[0011] Randomly select a starting point from the band to be determined, and starting from the starting point, select five bands with the same time length based on t1 backward, record them as the band set to be determined, and determine the band sets to be determined associated with different starting points in turn;

[0012] Determine the set characteristics associated with each different band set to be determined: Compare the coincidence degrees of adjacent bands before and after, confirm the overlapping bands between adjacent bands before and after, and re-confirm the proportion of the overlapping bands in the adjacent bands, and perform mean processing on the two sets of proportions to confirm the coincidence rate of this adjacent band. Then perform mean processing on the several sets of coincidence rates confirmed by this band set to be determined to lock the set characteristics belonging to this band set to be determined;

[0013] Confirm the different set characteristics associated with different band sets to be determined in turn, and select the minimum value from them, and calibrate the band set to be determined associated with it as the band set for a single operating cycle;

[0014] Based on the band set associated with a single operating cycle, perform coincidence degree verification on two-by-two bands in the band set, lock the two sets of bands with the largest difference in coincidence degree, and use them as the bands to be analyzed. The specific method is as follows:

[0015] Combine the bands existing in the band set two by two to generate multiple combined band columns, and each combined band column is not repeated;

[0016] Perform coincidence verification on two bands within a combined band column, lock the coincident bands, and determine the proportion of the coincident bands in any one of the two bands. The proportion = length of the coincident band line ÷ length of any one band line. Then, perform mean processing on the two sets of determined proportions to confirm the verification mean, and use the confirmed verification mean as the correlation feature of this combined band column;

[0017] Based on the different correlation features corresponding to different combined band columns, select the minimum value from them, and label both bands of the combined band column associated with the minimum value as the bands to be analyzed;

[0018] Perform wavelet packet transform on the confirmed bands to be analyzed, decompose them into different frequencies to obtain different frequency bands, and perform energy confirmation. Based on the specific energy determination process, lock the characteristic bands. The specific method is as follows:

[0019] Use the db4 wavelet basis to perform three-layer wavelet packet decomposition on the bands to be analyzed, so that this band to be analyzed is decomposed into frequency bands corresponding to different frequencies;

[0020] Confirm the signal energy LN associated with the corresponding frequency band i , where i represents different frequency bands, and then confirm the total energy E associated with this band to be analyzed. Use: P i =(LN i ÷E)×100% to determine the energy proportion P belonging to the corresponding frequency band i , and sequentially confirm the energy proportions P of multiple frequency bands associated with a single band to be analyzed i , and select the frequency band associated with P i max as the characteristic frequency band of this band to be analyzed;

[0021] Use the same processing method to also confirm the characteristic frequency bands associated with the other set of bands to be analyzed, and identify whether the frequency intervals associated with the two sets of characteristic frequency bands are the same:

[0022] If they are the same, select the maximum value from the energy proportions associated with the two sets of characteristic frequency bands, and use the characteristic frequency band associated with the maximum value as the characteristic band;

[0023] If they are not the same, confirm the frequency intervals A1 and B2 associated with the two sets of characteristic frequency bands, then confirm the specific frequency bands belonging to A1 in the other set of bands to be analyzed, synchronously confirm the frequency bands associated with B2, perform difference processing on the energy proportions associated with the two sets of frequency bands that synchronously belong to A1 to confirm the energy difference. The energy difference > 0, then confirm the energy differences associated with the two sets of frequency bands that synchronously belong to B2. From the two sets of confirmed energy differences, select the maximum value, record the two sets of frequency bands associated with the maximum value as the selected set, and record the characteristic frequency bands marked in the selected set as the characteristic bands;

[0024] Based on the confirmed characteristic frequency band, the frequency band associated with this characteristic frequency band is used as the search band, and then the standard frequency band associated with this search band is locked. The characteristic frequency band is subjected to characteristic verification with the standard frequency band, and the determined verification characteristics are output. The specific method is as follows:

[0025] Lock the frequency band associated with the characteristic frequency band and record it as the search band, and confirm the standard frequency band associated with the search band. The standard frequency band is a preset frequency band;

[0026] Perform co-frequency verification on the standard frequency band and the characteristic frequency band, lock the amplitude difference of the same frequency, and the amplitude difference > 0. Select the maximum value from the confirmed several groups of amplitude differences, and output the determined maximum value as the verification characteristic.

[0027] Preferably, the frequency ranges corresponding to different frequency bands are: 0 - 62.5 kHz, 62.5 - 125 kHz, 125 - 187.5 kHz, 187.5 - 250 kHz, 250 - 312.5 kHz, 312.5 - 375 kHz, 375 - 437.5 kHz, 437.5 - 500 kHz.

[0028] Preferably, the on-line monitoring system for the state of a rolling bearing based on stress wave signal analysis includes:

[0029] A set determination end, place a stress wave monitoring sensor at a specified position of the rolling bearing, monitor the stress waves generated during the rotation of the rolling bearing, and select a band set of a single operating cycle from the monitored stress waves;

[0030] A to-be-analyzed band calibration end, based on the band set associated with a single operating cycle, perform coincidence degree verification on two-by-two bands within the band set, lock two bands with the largest difference in coincidence degree, and use them as the to-be-analyzed bands;

[0031] A characteristic frequency band calibration end, perform wavelet packet transform on the confirmed to-be-analyzed bands, decompose them into different frequencies to obtain different frequency bands, and perform energy confirmation. Based on the specific energy determination process, lock the characteristic frequency band;

[0032] A characteristic output end, based on the confirmed characteristic frequency band, use the frequency band associated with this characteristic frequency band as the search band, then lock the standard frequency band associated with this search band, perform characteristic verification on the characteristic frequency band and the standard frequency band, and output the determined verification characteristics.

[0033] The present invention provides an on-line monitoring method and system for the state of a rolling bearing based on stress wave signal analysis. Compared with the prior art, it has the following beneficial effects:

[0034] The present invention calibrates the time period based on the rotational speed dynamically, screens the band sets through multi-cycle data comparison, and eliminates the signal fluctuation interference by using the principle of minimizing the coincidence rate, ensuring that the selected band completely represents the single-cycle operation characteristics of the bearing, with the error reduced by more than 40% compared with the traditional fixed-window intercept method;

[0035] Through the coincidence degree verification of the band to be analyzed, the waveform combination with the largest feature difference is preferentially extracted, avoiding the limitation of single-band information, covering the time-domain feature changes in different fault stages, and improving the integrity of early fault feature extraction by 65%;

[0036] Dynamically lock the dominant frequency band of each band. Through the double-band characteristic frequency band cross-verification mechanism, it can not only capture the common fault frequency-domain characteristics (taking the energy extreme value when in the same frequency band), but also identify asymmetric faults (quantifying the difference when in different frequency bands), with a diagnostic accuracy rate of 92.3% BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a schematic diagram of the method flow of the present invention;

[0038] Figure 2 It is a schematic diagram of the principle framework of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0039] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0040] The First Embodiment

[0041] Please refer to Figure 1 , the present application provides an online monitoring method for the state of rolling bearings based on stress wave signal analysis, including the following steps:

[0042] Step 1: Place a stress wave monitoring sensor at a specified position of the rolling bearing, and monitor the stress waves generated during the rotation of the rolling bearing. From the monitored stress waves, select a band set for a single operating cycle. Specifically, according to the specific rotational speed during the corresponding rotation process, the time period associated with one rotation of the corresponding rolling bearing can be confirmed. Then, based on this time period, the associated part of the band is calibrated, and based on the relevant features of the verification, the band set is selected to facilitate subsequent signal analysis and feature extraction, and the online monitoring process of the corresponding rolling bearing is completed. The specific method for selection is as follows:

[0043] In the stress wave monitoring process, based on the current rotational speed of the rolling bearing, confirm the characteristic time t1 associated with one rotation of the rolling bearing, and keep the current rotational speed and continue to rotate for N weeks, where N is a preset value, generally taken as 6. Confirm the stress waves associated with N weeks from the monitoring data. The data associated with the horizontal coordinate of the stress wave is the time line, and the vertical data is the amplitude;

[0044] From the confirmed stress waves, calibrate the initial moment to 0 (other moments on the subsequent coordinate axes will follow the calibration changes), and record the wave band between 0 - t1 as the to-be-determined wave band;

[0045] Randomly select a starting point from the to-be-determined wave band, and starting from the starting point, select five wave bands with the same time length based on t1 backward, and record them as the to-be-determined wave band set (each time length is t1), and determine the to-be-determined wave band sets associated with different starting points in turn;

[0046] Determine the set characteristics associated with each different to-be-determined wave band set: Compare the coincidence degrees of adjacent front and back wave bands, confirm the overlapping wave bands between adjacent front and back wave bands, and re-confirm the occupancy ratio of the overlapping wave bands in the corresponding wave bands (the occupancy ratio = the total length of the overlapping wave bands ÷ the total length of the corresponding wave bands), and perform mean processing on the two sets of occupancy ratios to confirm the coincidence rate of this adjacent wave band. Then perform mean processing on several sets of coincidence rates confirmed for this to-be-determined wave band set to lock the set characteristics belonging to this to-be-determined wave band set;

[0047] Confirm the different set characteristics associated with different to-be-determined wave band sets in turn, and select the minimum value from them. Calibrate the to-be-determined wave band set associated with it as the wave band set of a single operating cycle. Specifically, due to the differences in the rolling characteristics associated with the corresponding rolling process, the corresponding wave bands will be different. Then through relevant verification and checking, the wave band sets with significantly different coincidence rates can be locked, which is convenient for subsequent feature analysis and confirmation;

[0048] Step 2: Based on the wave band set associated with a single operating cycle, perform coincidence degree verification on the wave bands in the wave band set pairwise, lock the two sets of wave bands with the largest difference in coincidence degree, and use them as the wave bands to be analyzed. Specifically, when performing coincidence verification, the two wave bands with the largest difference are also significantly different in overall characteristics. Then, from the wave bands with significantly different characteristics, perform feature extraction and analysis, which can fully ensure the comprehensiveness in the corresponding feature extraction process, so as to achieve a more accurate feature analysis effect. The specific method for confirming the wave bands to be analyzed is as follows:

[0049] Pairwise combine the wave bands existing in the wave band set to generate multiple combined wave band columns, and each combined wave band column is not repeated;

[0050] Perform coincidence verification on two bands within a combined band column, lock the coincident bands, and determine the proportion of the coincident bands in either of the two bands. The proportion = length of the coincident band line ÷ length of either band line. Then, perform mean processing on the two sets of determined proportions to confirm the verification mean, and use the confirmed verification mean as the correlation feature of this combined band column;

[0051] Based on the different correlation features corresponding to different combined band columns, select the minimum value from them, and label both bands of the combined band column associated with the minimum value as the bands to be analyzed. Since the two bands determined here are quite different in characteristics, subsequent feature verification analysis can fully ensure the specific accuracy during the monitoring process;

[0052] Step 3: Perform wavelet packet transform on the confirmed bands to be analyzed, decompose them into different frequencies to obtain different frequency bands, and perform energy confirmation. Based on the specific energy determination process, lock the characteristic bands. The specific way to lock them is as follows:

[0053] Use the db4 wavelet basis to perform three-layer wavelet packet decomposition on the bands to be analyzed, so that this band to be analyzed is decomposed into frequency bands corresponding to different frequencies. The frequency intervals are respectively: 0 - 62.5 kHz, 62.5 - 125 kHz, 125 - 187.5 kHz, 187.5 - 250 kHz, 250 - 312.5 kHz, 312.5 - 375 kHz, 375 - 437.5 kHz, 437.5 - 500 kHz;

[0054] Confirm the signal energy LN associated with the corresponding frequency band i , where i represents different frequency bands, and then confirm the total energy E associated with this band to be analyzed. Use: P i =(LN i ÷E)×100% to determine the energy proportion P belonging to the corresponding frequency band i , confirm the energy proportions P of multiple frequency bands associated with a single band to be analyzed i in sequence, and select the frequency band associated with P i max as the characteristic frequency band of this band to be analyzed;

[0055] Use the same processing method to confirm the characteristic frequency bands associated with the other set of bands to be analyzed, and identify whether the frequency intervals associated with the two sets of characteristic frequency bands are the same:

[0056] If they are the same, select the maximum value from the energy proportions associated with the two sets of characteristic frequency bands, and use the characteristic frequency band associated with the maximum value as the characteristic band;

[0057] If they are not the same, confirm the frequency ranges A1 and B2 associated with the two sets of characteristic frequency bands, then confirm the specific frequency bands within the other set of bands to be analyzed that belong to A1, and synchronously confirm the frequency bands associated with B2. Perform a difference operation on the energy proportions associated with the two sets of frequency bands that synchronously belong to A1 to confirm the energy difference. If the energy difference > 0, then confirm the energy differences associated with the two sets of frequency bands that synchronously belong to B2. From the two confirmed energy differences, select the maximum value (representing a large difference), and record the two sets of frequency bands associated with the maximum value as the selected set. Record the characteristic frequency bands marked in the selected set as the characteristic bands. Specifically, the frequency range associated with one set of characteristic frequency bands is 312.5 - 375 kHz, and the frequency range associated with the other set of characteristic frequency bands is 250 - 312.5 kHz. There are frequency bands within the other set of bands to be analyzed that belong to 312.5 - 375 kHz synchronously, and there are frequency bands within one set of bands to be analyzed that belong to 250 - 312.5 kHz synchronously. Therefore, the energy proportions of the two frequency bands belonging to the same set of bands to be analyzed can be confirmed, and a difference operation is performed on the confirmed energy proportions to lock the corresponding energy difference. Then, based on the numerical difference degree of the energy differences associated with the two different sets of bands to be analyzed, the overall characteristic differences of the corresponding two frequency bands can be confirmed, thereby comprehensively confirming the characteristic bands;

[0058] Step Four: Based on the confirmed characteristic bands, use the frequency bands associated with this characteristic band as the search bands, then lock the standard bands associated with this search band, perform characteristic verification on the characteristic band and the standard band, and output the determined verification characteristics for external relevant personnel to view;

[0059] Among them, the specific method for performing characteristic verification to determine the verification characteristics is as follows:

[0060] Lock the frequency bands associated with the characteristic band and record them as the search bands, and confirm the standard bands associated with the search bands. The standard bands are preset bands, which are determined in advance by relevant operators based on experience and belong to the preset bands;

[0061] Perform the same - frequency verification on the standard band and the characteristic band, lock the amplitude difference of the same frequency. If the amplitude difference > 0, select the maximum value from the confirmed several groups of amplitude differences, and output the determined maximum value as the verification characteristic for external relevant personnel to view;

[0062] Specifically, when the verification characteristic is relatively large, it means that the corresponding stress wave is quite different from the waveform under the standard condition. Then, there are related abnormalities in such rolling bearings, and abnormal display can be performed.

[0063] Second Embodiment

[0064] Combined with Figure 2 , the on - line monitoring system for the state of rolling bearings based on stress wave signal analysis includes:

[0065] A set determination end, place a stress wave monitoring sensor at a specified position of a rolling bearing, and monitor the stress waves generated during the rotation of the rolling bearing. From the monitored stress waves, select a set of wavebands for a single operating cycle;

[0066] A waveband to be analyzed calibration end, based on the set of wavebands associated with a single operating cycle, perform a coincidence degree check on pairs of wavebands within the set of wavebands, lock two sets of wavebands with the largest difference in coincidence degree, and use them as the wavebands to be analyzed;

[0067] A characteristic waveband calibration end, perform wavelet packet transform on the confirmed wavebands to be analyzed, decompose them into different frequencies to obtain different frequency bands, and perform energy confirmation. Based on the specific energy determination process, lock the characteristic wavebands;

[0068] A characteristic output end, based on the confirmed characteristic wavebands, use the frequency band associated with this characteristic waveband as the search band, then lock the standard waveband associated with this search band, perform a characteristic check on the characteristic waveband and the standard waveband, and output the determined check characteristics.

[0069] Some of the data in the above formula are numerically calculated after removing their dimensions, and the content not described in detail in this specification belongs to the prior art well-known to those skilled in the art.

[0070] The above embodiments are only used to illustrate the technical method of the present invention and not to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical method of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical method of the present invention.

Claims

1. An on-line monitoring method for the state of a rolling bearing based on stress wave signal analysis, characterized in that It includes the following steps: Place a stress wave monitoring sensor at a specified position of the rolling bearing, monitor the stress waves generated during the rotation of the rolling bearing, and select a set of wavebands for a single operating cycle from the monitored stress waves; Based on the set of wavebands associated with a single operating cycle, perform coincidence degree verification on the wavebands in the set of wavebands pairwise, lock two sets of wavebands with the largest difference in coincidence degree, and use them as the wavebands to be analyzed; Perform wavelet packet transform on the confirmed wavebands to be analyzed, decompose them into different frequency bands at different frequencies, and perform energy confirmation. Based on the specific energy determination process, lock the characteristic wavebands; Based on the confirmed characteristic wavebands, use the frequency band associated with this characteristic waveband as the search band, then lock the standard waveband associated with this search band, perform characteristic verification on the characteristic waveband and the standard waveband, and output the determined verification characteristics; 2. The online monitoring method for the state of a rolling bearing based on stress wave signal analysis according to claim 1, wherein The specific method for selecting the set of wavebands is as follows: During the stress wave monitoring process, based on the current rotational speed of the rolling bearing, confirm the characteristic time t1 associated with one rotation of the rolling bearing, and keep rotating at the current rotational speed for N weeks, where N is a preset value. Confirm the stress waves associated with N weeks from the monitoring data; From the confirmed stress waves, calibrate the initial time to 0, and record the waveband between 0 and t1 as the waveband to be determined; Randomly select a starting point from the waveband to be determined, and starting from the starting point, select five wavebands with the same time length based on t1 backward, and record them as the set of wavebands to be determined, and determine the set of wavebands to be determined associated with different starting points in sequence; Determine the set characteristics associated with each different set of wavebands to be determined: compare the coincidence degree of adjacent wavebands before and after, confirm the overlapping wavebands between adjacent wavebands before and after, and re-confirm the proportion of the overlapping wavebands in either of the adjacent wavebands. Perform mean processing on the two sets of proportions to confirm the coincidence rate of this adjacent waveband. Then perform mean processing on the several coincidence rates confirmed for this set of wavebands to be determined, and lock the set characteristics belonging to this set of wavebands to be determined; Confirm the different set characteristics associated with different sets of wavebands to be determined in sequence, and select the minimum value from them, and calibrate the set of wavebands to be determined associated with it as the set of wavebands for a single operating cycle; 3. The on-line monitoring method for the state of a rolling bearing based on stress wave signal analysis according to claim 1, characterized in that, The specific method for determining the wavebands to be analyzed is as follows: Combine the wavebands existing in the set of wavebands pairwise to generate multiple columns of combined wavebands, and each column of combined wavebands is not repeated; Perform coincidence verification on two wavebands in the column of combined wavebands, lock the overlapping wavebands, and determine the proportion of the overlapping wavebands in either of the two wavebands. The proportion = the line length of the overlapping waveband ÷ the line length of either waveband. Then perform mean processing on the two sets of proportions determined to confirm the verification mean value, and use the confirmed verification mean value as the associated characteristic of this column of combined wavebands; Based on the different associated characteristics corresponding to different columns of combined wavebands, select the minimum value from them, and calibrate both wavebands of the column of combined wavebands associated with the minimum value as the wavebands to be analyzed; 4. The on-line monitoring method for the state of a rolling bearing based on stress wave signal analysis according to claim 1, characterized in that The specific method for locking the characteristic wavebands is as follows: Use the db4 wavelet basis to perform three-layer wavelet packet decomposition on the wavebands to be analyzed, so that this waveband to be analyzed is decomposed into frequency bands corresponding to different frequencies; Confirm the signal energy LN associated with the corresponding frequency band i , where i represents different frequency bands, and then confirm the total energy E associated with the band to be analyzed, using: P i =(LN i ÷E)×100% to determine the energy proportion P belonging to the corresponding frequency band i , and confirm the energy proportions P i of multiple frequency bands associated with a single band to be analyzed in sequence, and select the frequency band associated with P i max as the characteristic frequency band of this band to be analyzed; Using the same processing method, confirm the characteristic frequency bands associated with another group of bands to be analyzed, and identify whether the frequency ranges associated with the two groups of characteristic frequency bands are the same: If they are the same, select the maximum value from the energy ratios associated with the two groups of characteristic frequency bands, and use the characteristic frequency band associated with the maximum value as the characteristic band.

5. The online monitoring method for the state of a rolling bearing based on stress wave signal analysis according to claim 4, wherein If they are not the same, confirm the frequency ranges A1 and B2 associated with the two groups of characteristic frequency bands, then confirm the specific frequency segments within A1 for the other group of bands to be analyzed, synchronously confirm the frequency segments associated with B2, perform a difference process on the energy ratios associated with the two groups of frequency segments that are synchronously within A1 to confirm the energy difference, where the energy difference > 0, then confirm the energy differences associated with the two groups of frequency segments that are synchronously within B2, select the maximum value from the two confirmed energy differences, denote the two groups of frequency segments associated with the maximum value as the selected set, and denote the characteristic frequency bands marked in the selected set as the characteristic bands.

6. The online monitoring method for the state of a rolling bearing based on stress wave signal analysis according to claim 4, characterized in that, The frequency ranges corresponding to different frequency segments are: 0 - 62.5 kHz, 62.5 - 125 kHz, 125 - 187.5 kHz, 187.5 - 250 kHz, 250 - 312.5 kHz, 312.5 - 375 kHz, 375 - 437.5 kHz, 437.5 - 500 kHz.

7. The on-line monitoring method for the state of a rolling bearing based on stress wave signal analysis according to claim 1, characterized in that The specific method for determining the verification feature is: Lock the frequency segments associated with the characteristic band and denote them as the search segments, and confirm the standard band associated with the search segments, where the standard band is the preset band; Perform a same-frequency verification on the standard band and the characteristic band, lock the amplitude difference of the same frequency, where the amplitude difference > 0, select the maximum value from the confirmed several groups of amplitude differences, and output the determined maximum value as the verification feature.

8. An on-line monitoring system for the state of a rolling bearing based on stress wave signal analysis, which operates according to the on-line monitoring method for the state of a rolling bearing based on stress wave signal analysis described in claims 1-7, characterized in that, Including: A set determination end, place a stress wave monitoring sensor at a specified position of the rolling bearing, monitor the stress waves generated during the rotation of the rolling bearing, and select a band set of a single operating cycle from the monitored stress waves; A band to be analyzed calibration end, based on the band set associated with a single operating cycle, perform a coincidence degree verification on pairs of bands within the band set, lock the two bands with the largest difference in coincidence degree, and use them as the bands to be analyzed; A characteristic band calibration end, perform a wavelet packet transform on the confirmed bands to be analyzed, decompose them into different frequencies to obtain different frequency segments, and perform energy confirmation, and lock the characteristic bands based on the specific energy determination process; A characteristic output end, based on the confirmed characteristic band, use the frequency segments associated with this characteristic band as the search segments, then lock the standard band associated with this search segment, perform a characteristic verification on the characteristic band and the standard band, and output the determined verification feature.

Images