A method, device, equipment and medium for monitoring abnormal vibration of bearing

Abstract

The present invention belongs to the field of bearing vibration monitoring technology, and in particular, relates to a monitoring method, device, equipment and medium for abnormal bearing vibration, wherein the method comprises: obtaining historical bearing monitoring data as a training data set; extracting feature data of abnormal bearing vibration in the training data set, and constructing a correlation between the feature data and the training data set; performing memory training on a training bearing abnormal vibration supervision model based on the training data set to obtain a bearing abnormal vibration supervision model; obtaining bearing vibration monitoring data in real time, and analyzing the bearing vibration monitoring data through the bearing abnormal vibration supervision model, and issuing an early warning signal when the bearing vibration monitoring data is greater than or equal to a monitoring threshold. The present invention monitors the bearing vibration state to accurately evaluate the bearing operating condition, and provides technical support for design and maintenance personnel.

Description

Technical Field

[0001] The present invention relates to the technical field of bearing vibration monitoring, and in particular to a method, device, equipment and medium for monitoring abnormal bearing vibration. Background Art

[0002] Bearings are important supporting components in rotating equipment. Under long-term operation, local wear faults such as pitting and peeling will occur on their surfaces, leading to degradation of bearing performance. In particular, early wear of bearings in large equipment is generally not easy to detect. If the equipment continues to be put into use, it will affect production efficiency and pose a major safety hazard. In recent years, vibration monitoring has been the mainstream method for bearing fault diagnosis. However, when a bearing fails, its vibration signal characteristics often show non-Gaussian and non-stationary randomness, and the vibration form of the unit transmission system is a complex multi-rigid body coupled vibration, with high operating noise and a large number of interference signals, which makes the accuracy of the bearing operating status assessment low.

[0003] Therefore, how to accurately evaluate the operating condition of bearings is a problem that needs to be solved at present. Summary of the invention

[0004] The present invention provides a monitoring method, device, equipment and medium for abnormal vibration of a bearing, aiming to accurately evaluate the operating condition of the bearing and achieve accurate control of the abnormal vibration condition of the bearing.

[0005] In order to achieve the above-mentioned object of the invention, the first aspect of the present invention provides a method for monitoring abnormal vibration of a bearing, the method comprising:

[0006] Obtain historical bearing monitoring data as a training data set;

[0007] Extracting characteristic data of abnormal bearing vibration in the training data set, and constructing an association relationship between the characteristic data and the training data set;

[0008] Performing memory training on the bearing abnormal vibration supervision model to be trained based on the training data set to obtain the bearing abnormal vibration supervision model;

[0009] The bearing vibration monitoring data is acquired in real time, and the bearing vibration monitoring data is analyzed by the bearing abnormal vibration supervision model. When the bearing vibration monitoring data is greater than or equal to the monitoring threshold, an early warning signal is issued.

[0010] Further, if the historical bearing monitoring data is bearing position monitoring data, the acquiring of the historical bearing monitoring data as a training data set includes:

[0011] Obtain the force data of radial force and axial force on the bearing when it is installed on the equipment;

[0012] Determine the load limit point of the bearing according to the force data;

[0013] A position monitoring data set of the load-limiting force point under different acceleration conditions is obtained, and the position monitoring data set is a training data set.

[0014] Further, if the historical bearing monitoring data is electrostatic monitoring data of the bearing, the step of obtaining the historical bearing monitoring data as a training data set includes:

[0015] Obtain the charge data of the bearing under different acceleration conditions;

[0016] Obtain the frequency data of bearing rotation under different acceleration conditions;

[0017] Converting the charge data into a voltage signal according to Laplace transform;

[0018] The frequency data and the voltage signal are transformed and matched with each other through Fourier transformation to obtain frequency domain features;

[0019] The noise in the frequency domain features is filtered out according to preset screening rules to obtain effective frequency domain features, and the effective frequency domain features are a training data set.

[0020] Furthermore, the extracting of characteristic data of abnormal bearing vibration in the training data set includes:

[0021] Reading abnormal vibration data of the bearing including the cage, rolling element, outer ring and inner ring in the training data set when the bearing has abnormal vibration;

[0022] Calculate the peak acceleration of the bearing rotation under different acceleration conditions;

[0023] Calculating the fault frequencies of the abnormal vibration data according to the acceleration peak values ​​to obtain fault characteristic frequencies including a cage fault characteristic frequency, a rolling element fault characteristic frequency, an outer ring fault characteristic frequency, and an inner ring fault characteristic frequency;

[0024] Compare the fault characteristic frequency with a preset fault threshold, and if the fault characteristic frequency is less than the preset fault threshold, record it as first abnormal characteristic data; and,

[0025] If the fault characteristic frequency is greater than or equal to the preset fault threshold, it is recorded as second abnormal characteristic data.

[0026] Furthermore, the constructing of the association relationship between the feature data and the training data set includes:

[0027] Filtering out first training data from the training data set according to the first abnormal feature data, establishing a data link between the first abnormal feature data and the first training data, and adding a first identification mark to obtain first training data;

[0028] The second training data is screened out from the training data set according to the second abnormal feature data, and a data link is established between the second abnormal feature data and the second training data, and a second identification mark is added to obtain second associated data.

[0029] Furthermore, the memory training of the abnormal vibration supervision model of the bearing to be trained based on the training data set includes:

[0030] Inputting the training data set into the bearing abnormal vibration supervision model to be trained;

[0031] In the bearing abnormal vibration supervision model to be trained, the first training data is used to train the bearing abnormal vibration supervision model to be trained for the recognition performance of the first-level damage caused by the bearing abnormal vibration, and the training results of the first training data are grouped into a first group; and,

[0032] Using the second training data to train the bearing abnormal vibration supervision model to be trained to identify the secondary damage caused by the bearing abnormal vibration, and grouping the training results of the second training data into a second group until the bearing abnormal vibration supervision model to be trained converges to obtain an initial bearing abnormal vibration supervision model;

[0033] Verifying the recognition result of the initial bearing abnormal vibration supervision model on the damage caused by the bearing abnormal vibration based on the first grouping and the second grouping;

[0034] If the recognition result is the same as the training results of the first group and the second group, it is determined that the abnormal vibration supervision model of the bearing to be trained has completed training, and the abnormal vibration supervision model of the bearing is obtained.

[0035] Furthermore, the real-time acquisition of bearing vibration monitoring data and the analysis of the bearing vibration monitoring data by the bearing abnormal vibration supervision model, when the bearing vibration monitoring data is greater than or equal to a monitoring threshold, an early warning signal is issued, including:

[0036] Collect bearing vibration monitoring data in real time and read bearing abnormal vibration supervision model;

[0037] The bearing vibration monitoring data is analyzed by the bearing abnormal vibration supervision model, and if the bearing vibration monitoring data is greater than or equal to a monitoring threshold, an early warning signal is issued;

[0038] If the bearing vibration monitoring data is less than the monitoring threshold, the bearing vibration monitoring data is updated to the training data set to train and update the bearing abnormal vibration supervision model to obtain a modified bearing abnormal vibration supervision model.

[0039] The present application also provides a monitoring device for abnormal bearing vibration, the device comprising:

[0040] An acquisition module, used for acquiring historical bearing monitoring data as a training data set;

[0041] An extraction module, used to extract characteristic data of abnormal bearing vibration in the training data set, and to construct an association relationship between the characteristic data and the training data set;

[0042] A training module, used for performing memory training on the bearing abnormal vibration supervision model to be trained based on the training data set to obtain the bearing abnormal vibration supervision model;

[0043] The monitoring module is used to obtain bearing vibration monitoring data in real time, and analyze the bearing vibration monitoring data through the bearing abnormal vibration supervision model, and issue an early warning signal when the bearing vibration monitoring data is greater than or equal to a monitoring threshold.

[0044] The present application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any of the above-mentioned methods for monitoring abnormal vibration of a bearing when executing the computer program.

[0045] The present application also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the method for monitoring abnormal vibration of a bearing described in any one of the above items are implemented.

[0046] Beneficial effect: In the present application, historical bearing monitoring data is obtained as a training data set; then the characteristic data of the bearing abnormal vibration in the training data set is extracted, and the association relationship between the characteristic data and the training data set is constructed; then, based on the training data set, the bearing abnormal vibration supervision model to be trained is memory trained, and then the bearing abnormal vibration supervision model is obtained. The bearing abnormal vibration supervision model has the performance of identifying the data generated by the bearing abnormal vibration. When the bearing vibration monitoring data is obtained in real time, the bearing vibration monitoring data can be analyzed by the bearing abnormal vibration supervision model. If the bearing vibration monitoring data is greater than or equal to the monitoring threshold, an early warning signal is issued to remind relevant technical personnel to inspect the bearing to ensure the normal operation of the bearing, thereby achieving accurate control of the bearing abnormal vibration condition. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 This is a flow chart of an embodiment of a method for monitoring abnormal vibration of a bearing of the present application;

[0048] Figure 2 This is a schematic structural diagram of an embodiment of a monitoring device for abnormal bearing vibration of the present application;

[0049] Figure 3 This is a schematic block diagram of the structure of an embodiment of a computer device of the present application.

[0050] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0052] It will be understood by those skilled in the art that, unless expressly stated, the singular forms "a", "an", "above", and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present invention refers to the presence of features, integers, steps, operations, elements, modules, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, modules, components, and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be an intermediate element. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The term "and / or" used herein includes all or any module and all combinations of one or more associated listed items.

[0053] Those skilled in the art will understand that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as those generally understood by those skilled in the art in the field to which the present invention belongs. It should also be understood that terms such as those defined in common dictionaries should be understood to have meanings consistent with the meanings in the context of the prior art, and will not be interpreted with idealized or overly formal meanings unless specifically defined as here.

[0054] Reference Figure 1 The embodiment of the present invention provides a method for monitoring abnormal vibration of a bearing, comprising the following steps S1-S4:

[0055] S1: Obtain historical bearing monitoring data as a training data set.

[0056] In this embodiment, when a bearing fails, abnormal vibration will occur. In order to accurately control the bearing state, the health state of the bearing can be controlled by monitoring the bearing vibration. Specifically, in order to accurately analyze the bearing vibration data, a bearing abnormal vibration supervision model can be used to analyze the bearing vibration data. Prior to this, the bearing abnormal vibration supervision model needs to be trained to enable it to have analysis performance. In order to train the bearing abnormal vibration supervision model, a data set for training can be obtained first. The data set is the historical monitoring data of the bearing. The historical monitoring data includes vibration data when the bearing rotates normally and abnormal data of abnormal vibration when the bearing is damaged or fails.

[0057] S2: extracting feature data of abnormal bearing vibration in the training data set, and constructing an association relationship between the feature data and the training data set.

[0058] After obtaining the historical monitoring data of the bearing as a training data set, the characteristic data of the abnormal vibration of the bearing in the training data set is extracted, and the correlation relationship between the characteristic data and the training data set is constructed. Specifically, the abnormal vibration data of the bearing including the cage, rolling element, outer ring and inner ring in the training data set when the bearing vibrates abnormally is read; and the acceleration peak value of the bearing rotation under different acceleration conditions is calculated; the fault frequency of the abnormal vibration data is then calculated using the acceleration peak value; in order to distinguish the fault frequency size when the bearing vibrates abnormally, the fault characteristic frequency is also compared with the preset fault threshold to distinguish the abnormal data characteristics corresponding to the fault degree size when the bearing vibrates abnormally, and then clarify the classification of bearing damage; the classified abnormal data is also associated with the characteristics so that the bearing fault signal achieves common frequency, that is, in the case of bearing vibration, the monitoring data and the bearing status characteristics match each other, and the health status of the bearing is intuitively displayed to users and related personnel.

[0059] S3: Memory training is performed on the bearing abnormal vibration supervision model to be trained based on the training data set to obtain the bearing abnormal vibration supervision model.

[0060] After extracting the characteristic data of the abnormal vibration of the bearing in the training data set and constructing the association relationship between the characteristic data and the training data set, the bearing abnormal vibration supervision model to be trained is memory-trained based on the training data set, so that the bearing abnormal vibration supervision model has the performance of identifying the abnormal vibration of the bearing. Specifically, the training data set is input into the bearing abnormal vibration supervision model to be trained; in the bearing abnormal vibration supervision model to be trained, the training data is used to train the bearing abnormal vibration supervision model to be trained on the abnormal data when the bearing abnormal vibration causes damage, until the bearing abnormal vibration supervision model to be trained converges, and the initial bearing abnormal vibration supervision model is obtained. In addition, in order to ensure the accuracy of the model training recognition results, the model is trained twice using the training recognition results to verify the accuracy of the model for recognition analysis. If the two recognition results are the same, it is determined that the bearing abnormal vibration supervision model to be trained has completed the training, and the bearing abnormal vibration supervision model is obtained.

[0061] S4: Acquire bearing vibration monitoring data in real time, and analyze the bearing vibration monitoring data through the bearing abnormal vibration supervision model, and issue an early warning signal when the bearing vibration monitoring data is greater than or equal to a monitoring threshold.

[0062] After completing the training of the bearing abnormal vibration supervision model, the bearing vibration monitoring data can be collected in real time, and the bearing vibration monitoring data can be analyzed through the bearing abnormal vibration supervision model. If the bearing vibration monitoring data is greater than or equal to the monitoring threshold, an early warning signal is issued to remind relevant personnel to inspect and replace the bearing; but if the bearing vibration monitoring data is less than the monitoring threshold, the bearing vibration monitoring data is updated to the training data set and combined to train and update the bearing abnormal vibration supervision model, so that the model's recognition performance is updated as the monitoring data is updated, thereby obtaining a continuously updated and corrected bearing abnormal vibration supervision model.

[0063] The present embodiment provides a method for monitoring abnormal vibration of a bearing, which obtains historical bearing monitoring data as a training data set; then extracts feature data of the abnormal vibration of the bearing in the training data set, and constructs a correlation between the feature data and the training data set; then performs memory training on a bearing abnormal vibration supervision model to be trained based on the training data set, thereby obtaining a bearing abnormal vibration supervision model, which has the performance of identifying data generated by abnormal vibration of the bearing. When the bearing vibration monitoring data is obtained in real time, the bearing vibration monitoring data can be analyzed by the bearing abnormal vibration supervision model. If the bearing vibration monitoring data is greater than or equal to the monitoring threshold, an early warning signal is issued to remind relevant technical personnel to inspect the bearing to ensure the normal operation of the bearing, thereby achieving accurate control of the abnormal vibration condition of the bearing.

[0064] In one embodiment, if the historical bearing monitoring data is bearing position monitoring data, the above-mentioned obtaining the historical bearing monitoring data as a training data set includes:

[0065] S11. Obtain force data of radial force and axial force applied to the bearing when it is installed on the equipment.

[0066] Equipment refers to any equipment that needs to install and use bearings, without specific restrictions. When the bearing is used on the equipment, the force loads in various directions in the three-dimensional space are different. Therefore, the vibration amplitude transmitted when it vibrates is not the same. In order to accurately evaluate the abnormal conditions of the bearing during vibration, it is necessary to obtain the force conditions of the bearing in various directions, so that the monitoring position will be determined according to the force data of the bearing in the future; among them, when obtaining the force data of the radial force and axial force of the bearing, the center point of the bearing can be used as the limit load point for force analysis, or other positions of the bearing can be used as the limit load point for force analysis, without specific restrictions.

[0067] S12. Determine the load limit point of the bearing according to the force data.

[0068] The force data of radial force and axial force are sorted and analyzed, and the maximum pressure-bearing position of the bearing is taken as the limit load stress point. There are multiple limit load stress points. A monitoring device for monitoring the vibration of the bearing is arranged at the limit load stress point. The monitoring device can be a sensor, etc., and there is no specific limitation. The monitoring data of the bearing vibration will be obtained through the monitoring device later.

[0069] S13, obtaining a position monitoring data set of the limit load force point under different acceleration conditions, wherein the position monitoring data set is a training data set.

[0070] Under different acceleration conditions, the bearing is subjected to different forces, and the corresponding vibration amplitude is also different. Based on this, it is necessary to obtain monitoring data of each load-limiting stress point of the bearing under different acceleration conditions. Among them, different accelerations include the maximum acceleration of the bearing when it rotates with the motor, and the acceleration of each limit.

[0071] As described above, the monitoring position of the bearing is determined according to the stress condition of the bearing, so as to monitor the position points of the bearing that are susceptible to fatigue damage, and then accurately control the abnormal conditions of the bearing. Specifically, the stress conditions of the bearing in all directions are obtained, and the maximum pressure-bearing position of the bearing is analyzed according to the stress conditions. Under the same conditions, the maximum pressure-bearing position is more prone to damage than other positions; and the bearing is subjected to different forces under different acceleration conditions. Therefore, by monitoring the vibration generated by the bearing operation at the maximum pressure position under different accelerations, it is helpful to control the health status of the bearing. Among them, the monitoring data obtained at the maximum pressure position under different accelerations includes abnormal bearing vibration data (vibration data when the bearing is damaged) and normal data (data of normal bearing operation).

[0072] In one embodiment, if the historical bearing monitoring data is electrostatic monitoring data of the bearing, the above-mentioned obtaining the historical bearing monitoring data as a training data set includes:

[0073] S14. Obtain charge data of the bearing under different acceleration conditions.

[0074] During the operation of mechanical equipment, static electricity occurs in the friction and wear area due to a combination of various charging mechanisms. Therefore, the static electricity generated when the bearing rotates can be used to analyze and evaluate the bearing status.

[0075] S15. Obtain frequency data of bearing rotation under different acceleration conditions.

[0076] S16, converting the charge data into a voltage signal according to Laplace transform.

[0077] In the electrostatic induction process, the induced charge is generally not easy to be detected directly. It is necessary to connect the sensor end to the signal conditioning unit, and then convert the induced charge in the probe into an electrostatic induced voltage signal that can be measured and studied. This is the principle of electrostatic monitoring. Therefore, it is necessary to convert the monitored charge data into a voltage signal.

[0078] S17. Convert the frequency data and the voltage signal through Fourier transform and match the features to obtain frequency domain features.

[0079] In order to correlate the frequency data of the bearing and the collected voltage signal, Fourier transform can be used to convert the frequency data and the voltage signal so that the frequency data and the voltage signal match each other. For example, when the speed frequency gradually increases, the electrostatic fluctuation corresponding to the voltage signal becomes more and more obvious, and the electrostatic induction becomes stronger; when the speed frequency gradually decreases, the electrostatic fluctuation corresponding to the voltage signal becomes weaker and weaker, and the electrostatic induction becomes weaker and weaker; this intuitively shows the relationship between the frequency and voltage signals, and then quickly controls the abnormal vibration of the bearing.

[0080] S18. Filter out the noise in the frequency domain features according to preset screening rules to obtain effective frequency domain features, where the effective frequency domain features are a training data set.

[0081] After the frequency data and voltage signal are converted and matched, there may be some data that cannot be matched. For example, there are electrostatic signals caused by power frequency interference, background noise and random pulses. This electrostatic signal is different from the static electricity generated by bearing vibration, and is not related to the frequency data of bearing vibration. Based on this, the data of the electrostatic signal needs to be filtered out to prevent interference in the subsequent model training process, so as to obtain effective frequency domain features that are all interrelated.

[0082] As mentioned above, in order to enrich the model training samples and improve the model's recognition performance for various data generated by bearing vibration, the state of the bearing can also be analyzed and evaluated through the electrostatic phenomenon generated when the bearing vibrates; that is, the charge data of the bearing under different acceleration conditions is obtained and the charge data is converted into a voltage signal according to the Laplace transform; the frequency data of the bearing rotation under different acceleration conditions is obtained, and then the frequency data and the voltage signal are converted and matched by Fourier transform to obtain the frequency generated when the bearing vibrates and the frequency domain characteristics corresponding to the electrostatic phenomenon; in order to ensure the effectiveness of the frequency domain characteristics, some unmatched data in the frequency domain characteristics are also filtered out to prevent interference in the subsequent model training process and ensure the effectiveness of the training data set.

[0083] In one embodiment, the extracting of characteristic data of abnormal bearing vibration in the training data set includes:

[0084] S21, reading abnormal vibration data of the bearing including the cage, rolling element, outer ring and inner ring in the training data set when the bearing vibrates abnormally.

[0085] When a bearing fails, bearing components such as the cage, rolling elements, outer ring, inner ring, etc. usually have local damage, such as pitting, spalling, cracks, etc.; therefore, when the bearing has abnormal vibration, the data monitored by the cage, rolling elements, outer ring, inner ring as the bearing vibrates abnormally is abnormal vibration data.

[0086] S22. Calculate the peak acceleration of the bearing rotation under different acceleration conditions.

[0087] Vibration, such as velocity peak, is a common characteristic parameter that reflects the magnitude of vibration. The greater the acceleration, the greater the vibration amplitude of the bearing. Conversely, the smaller the acceleration, the smaller the vibration amplitude. Therefore, the acceleration peak can be used as a characteristic parameter for evaluating bearing failure.

[0088] S23. Calculate the fault frequencies of the abnormal vibration data according to the acceleration peak values ​​to obtain fault characteristic frequencies including a cage fault characteristic frequency, a rolling element fault characteristic frequency, an outer ring fault characteristic frequency, and an inner ring fault characteristic frequency.

[0089] Since the characteristic frequency of bearing fault is usually modulated with the high-frequency natural vibration frequency, it is difficult to identify the characteristic frequency of bearing fault by directly performing spectrum analysis on vibration acceleration data. Therefore, envelope analysis technology is used to calculate the fault characteristic frequency of each data when the bearing vibration is abnormal under the peak acceleration state, and then the frequency domain characteristic parameters for judging the bearing fault are obtained.

[0090] S24, comparing the fault characteristic frequency with a preset fault threshold, if the fault characteristic frequency is less than the preset fault threshold, recording it as first abnormal characteristic data; and,

[0091] The fault characteristic frequency is the abnormal data of bearing vibration abnormality / bearing fault. The fault characteristic frequency is compared with the preset fault threshold to screen out the degree of bearing abnormality. When the fault characteristic frequency is less than the preset fault threshold, it indicates that the bearing abnormality / bearing fault is relatively mild. The abnormal data with a relatively mild fault degree is recorded as the first abnormal characteristic data. The first abnormal characteristic data is not a fixed value, but a data interval that is less than the preset fault threshold.

[0092] S25. If the fault characteristic frequency is greater than or equal to the preset fault threshold, it is recorded as second abnormal characteristic data.

[0093] When the fault characteristic frequency is greater than or equal to the preset fault threshold, it indicates that the bearing is abnormal / the bearing fault is serious. The abnormal data with a serious fault degree is recorded as the second abnormal characteristic data. The second abnormal characteristic data is not a fixed value, but a data interval greater than or equal to the preset fault threshold.

[0094] As mentioned above, bearing components generally include cages, rolling elements, outer rings, inner rings, etc. When the bearing has abnormal vibration, the data monitored by the cage, rolling elements, outer rings, and inner rings along with the abnormal bearing vibration are abnormal vibration data; vibration such as velocity peak is a common characteristic parameter reflecting the vibration magnitude, and the acceleration peak of the bearing rotation under different acceleration conditions is also calculated; but it is difficult to identify the bearing fault characteristic frequency by directly performing spectrum analysis on the vibration acceleration data. Therefore, the envelope analysis technology is used to calculate the fault characteristic frequency of each data when the bearing vibration is abnormal under the acceleration peak state, and then the frequency domain characteristic parameters for judging the bearing fault are obtained. In order to distinguish the fault frequency size when the bearing vibration is abnormal, the fault characteristic frequency is also compared with the preset fault threshold. If the fault characteristic frequency is less than the preset fault threshold, it is recorded as the first abnormal characteristic data; if the fault characteristic frequency is greater than or equal to the preset fault threshold, it is recorded as the second abnormal characteristic data; the abnormal data is divided to clarify the classification of the bearing damage.

[0095] In one embodiment, the above-mentioned construction of the association relationship between the feature data and the training data set includes:

[0096] S26. Filter out first training data from the training data set according to the first abnormal feature data, establish a data link between the first abnormal feature data and the first training data, and add a first identification mark to obtain the first training data.

[0097] S27. Filter out second training data from the training data set according to the second abnormal feature data, establish a data link between the second abnormal feature data and the second training data, and add a second identification mark to obtain second associated data.

[0098] As described above, different abnormal characteristic data indicate different degrees of bearing damage. Correlating the abnormal data with the corresponding bearing damage degrees can provide an effective basis for the accurate identification of the subsequent accurate model. Specifically, the training data set includes data on normal bearing rotation and data on abnormal bearing vibration, wherein the data on abnormal bearing vibration includes data on a small degree of bearing damage and data on a large degree of bearing damage when the bearing vibration is abnormal. Therefore, the first training data corresponding to the first abnormal characteristic data with less bearing damage is screened out from the training data, and after the first training data is screened out, a data link is established between the first training data and the first abnormal characteristic data. The data link is a two-way link. For example, if the first abnormal data is identified during the model training process, the first abnormal characteristic data corresponding to the first abnormal data is read based on the data link, and then the bearing damage corresponding to the first abnormal data generated when the bearing vibration is abnormal is determined. In order to make the first training data targeted and easy to identify when training the model, a first identification identifier for identification is assigned to the first training data, and the first identification identifier is any identifier that is easy to identify. Similarly, according to the second abnormal characteristic data with greater bearing damage, the corresponding second training data is screened out from the training data, and after the second training data is screened out, a data link is established between the second training data and the second abnormal characteristic data, and a second identification mark for identification is assigned. In the training data set, after the first training data and the second training data are screened, the training data that is not screened out is normal data of bearing vibration, and the normal data corresponds to data monitored based on the vibration generated by the normal operation of the bearing. The data can be marked as third training data and assigned a third identification mark.

[0099] In one embodiment, the memory training of the abnormal vibration supervision model of the bearing to be trained based on the training data set includes:

[0100] S31, inputting the training data set into the bearing abnormal vibration supervision model to be trained.

[0101] S32, in the bearing abnormal vibration supervision model to be trained, using the first training data to train the bearing abnormal vibration supervision model to be trained for the recognition performance of the first-level damage caused by the bearing abnormal vibration, and grouping the training results of the first training data into a first group; and,

[0102] S33. Use the second training data to train the bearing abnormal vibration supervision model to be trained to identify the secondary damage caused by bearing abnormal vibration, and group the training results of the second training data into a second group until the bearing abnormal vibration supervision model to be trained converges to obtain an initial bearing abnormal vibration supervision model.

[0103] S34. Verify the recognition result of the initial bearing abnormal vibration supervision model on the damage caused by the bearing abnormal vibration based on the first grouping and the second grouping.

[0104] S35. If the recognition result is the same as the training results of the first group and the second group, it is determined that the abnormal vibration supervision model of the bearing to be trained has completed training, and the abnormal vibration supervision model of the bearing is obtained.

[0105] As described above, in order to obtain a supervision model with analysis and recognition functions for bearing data, the training data set is input into the supervision model for abnormal vibration of the bearing to be trained. In the supervision model for abnormal vibration of the bearing to be trained, the recognition and classification performance of the model for abnormal data and normal data during bearing vibration is mainly trained. Specifically, the training data set is input into the supervision model for abnormal vibration of the bearing to be trained. In the supervision model for abnormal vibration of the bearing to be trained, the first training data is used to train the supervision model for abnormal vibration of the bearing to be trained to perform the performance of first-level damage identification caused by abnormal vibration of the bearing, and the recognition result is grouped into the first group; and the second training data is used to train the supervision model for abnormal vibration of the bearing to be trained to perform the performance of second-level damage identification caused by abnormal vibration of the bearing, and the recognition result is grouped into the second group; wherein the degree of damage of the first-level damage is less than the degree of damage of the second-level damage; however, during the training process, there may be a situation where the associated data of the first-level damage shows a first-level damage indicator, but its bearing damage degree is a second-level damage. Therefore, in order to avoid misjudgment of the model and thereby improve the accuracy of model recognition, the identified first group is also secondary recognized, that is, through algorithms such as random forests. The method randomly deletes the data of the first group, and establishes multiple groups of deleted data sets to be restored after deletion for the same data; then, the deleted data set is restored through the recovery model established by the diagnosed damage degree type to obtain the restored data set; the restored data set is compared with the deleted data set, and the comparison process includes: constructing the vibration abnormality feature difference between the deleted data set and the restored data set according to the bearing continuous amplitude difference index combined with the cosine similarity, wherein the value range of the cosine similarity is between [-1, 1], and calculating the similarity between the deleted data set and the restored data set. When the cosine similarity of the deleted data set and the restored data set is closer to 1, they are more similar, indicating that the vibration abnormality feature is smaller (there is no misjudgment of the recognition result of the first group); but if the cosine similarity of the deleted data set and the restored data set is close to -1, it means that the deleted data set and the restored data set are extremely dissimilar. There is a situation in which the data in the first group shows a first-level damage indicator, but the bearing damage degree is a second-level damage. The misjudged data needs to be divided into the second group to realize the re-classification of the data information in the first group that does not correspond to the damage indicator and the damage data.In addition, the third training data is also used to evaluate the performance of the bearing abnormal vibration supervision model to be trained in identifying normal bearing vibration data. The training results are grouped into the third group until the bearing abnormal vibration supervision model to be trained converges, and the initial bearing abnormal vibration supervision model can be obtained. In order to verify the training effect of the initial bearing abnormal vibration supervision model and improve the confidence of its recognition results, the first group, the second group and the third group are input into the initial bearing abnormal vibration supervision model for recognition training again. If the initial recognition training result and the re-recognition training result are the same, the model is judged to have completed the training, and the bearing abnormal vibration supervision model can be obtained; but if the initial recognition training result and the re-recognition training result are different, iterative training is required until the recognition results are the same, and then the bearing abnormal vibration supervision model to be trained is judged to have completed the training, so as to ensure the accuracy of the recognition result of the bearing abnormal vibration supervision model.

[0106] In one embodiment, the real-time acquisition of bearing vibration monitoring data and the analysis of the bearing vibration monitoring data by the bearing abnormal vibration supervision model are performed, and when the bearing vibration monitoring data is greater than or equal to a monitoring threshold, an early warning signal is issued, including:

[0107] S41. Collect bearing vibration monitoring data in real time and read bearing abnormal vibration supervision model.

[0108] S42. Analyze the bearing vibration monitoring data through the bearing abnormal vibration supervision model, and issue a warning signal if the bearing vibration monitoring data is greater than or equal to a monitoring threshold.

[0109] S43. If the bearing vibration monitoring data is less than the monitoring threshold, the bearing vibration monitoring data is updated to the training data set to train and update the bearing abnormal vibration supervision model to obtain a modified bearing abnormal vibration supervision model.

[0110] As described above, by collecting bearing vibration monitoring data in real time and using the bearing abnormal vibration supervision model for analysis, the health status of the bearing under vibration can be accurately controlled. Specifically, the bearing monitoring data is obtained in real time and the bearing abnormal vibration supervision model is read. The bearing monitoring data is identified and analyzed through the bearing abnormal vibration supervision model. The identification and analysis include identifying the bearing monitoring data, and then comparing the identification result with the bearing abnormal vibration data. When they are the same or similar, it is judged that the bearing vibration is abnormal, and an early warning can be issued to remind relevant personnel to check and replace the bearing status; but if there is no sameness or similarity with the bearing abnormal vibration data, the real-time bearing monitoring data can be updated to the training data, and the bearing abnormal vibration supervision model can be updated and trained with the bearing monitoring data, so that the bearing abnormal vibration supervision model is kept updated in real time to adapt to the health monitoring of the updated bearing vibration status.

[0111] Reference Figure 2 , an embodiment of the present invention further provides a monitoring device for abnormal bearing vibration, comprising:

[0112] An acquisition module 10 is used to acquire historical bearing monitoring data as a training data set;

[0113] An extraction module 20, for extracting characteristic data of abnormal bearing vibration in the training data set, and constructing an association relationship between the characteristic data and the training data set;

[0114] A training module 30 is used to perform memory training on the bearing abnormal vibration supervision model to be trained based on the training data set to obtain the bearing abnormal vibration supervision model;

[0115] The monitoring module 40 is used to obtain bearing vibration monitoring data in real time, and analyze the bearing vibration monitoring data through the bearing abnormal vibration supervision model, and issue an early warning signal when the bearing vibration monitoring data is greater than or equal to a monitoring threshold.

[0116] As described above, the device for monitoring abnormal vibration of a bearing can implement a method for monitoring abnormal vibration of a bearing.

[0117] In one embodiment, the acquisition module 10 further includes:

[0118] Obtain the force data of radial force and axial force on the bearing when it is installed on the equipment;

[0119] Determine the load limit point of the bearing according to the force data;

[0120] A position monitoring data set of the load-limiting force point under different acceleration conditions is obtained, and the position monitoring data set is a training data set.

[0121] In one embodiment, the acquisition module 10 further includes:

[0122] Obtain the charge data of the bearing under different acceleration conditions;

[0123] Obtain the frequency data of bearing rotation under different acceleration conditions;

[0124] Converting the charge data into a voltage signal according to Laplace transform;

[0125] The frequency data and the voltage signal are transformed and matched with each other through Fourier transformation to obtain frequency domain features;

[0126] The noise in the frequency domain features is filtered out according to preset screening rules to obtain effective frequency domain features, and the effective frequency domain features are a training data set.

[0127] In one embodiment, the extraction module 20 further includes:

[0128] Reading abnormal vibration data of the bearing including the cage, rolling element, outer ring and inner ring in the training data set when the bearing has abnormal vibration;

[0129] Calculate the peak acceleration of the bearing rotation under different acceleration conditions;

[0130] Calculating the fault frequencies of the abnormal vibration data according to the acceleration peak values ​​to obtain fault characteristic frequencies including a cage fault characteristic frequency, a rolling element fault characteristic frequency, an outer ring fault characteristic frequency, and an inner ring fault characteristic frequency;

[0131] Compare the fault characteristic frequency with a preset fault threshold, and if the fault characteristic frequency is less than the preset fault threshold, record it as first abnormal characteristic data; and,

[0132] If the fault characteristic frequency is greater than or equal to the preset fault threshold, it is recorded as second abnormal characteristic data.

[0133] In one embodiment, the extraction module 20 further includes:

[0134] Filtering out first training data from the training data set according to the first abnormal feature data, establishing a data link between the first abnormal feature data and the first training data, and adding a first identification mark to obtain first training data;

[0135] The second training data is screened out from the training data set according to the second abnormal feature data, and a data link is established between the second abnormal feature data and the second training data, and a second identification mark is added to obtain second associated data.

[0136] In one embodiment, the training module 30 further includes:

[0137] Inputting the training data set into the bearing abnormal vibration supervision model to be trained;

[0138] In the bearing abnormal vibration supervision model to be trained, the first training data is used to train the bearing abnormal vibration supervision model to be trained for the recognition performance of first-level damage caused by bearing abnormal vibration, and the training results of the first training data are grouped into a first group; and,

[0139] Using the second training data to train the bearing abnormal vibration supervision model to be trained to identify the secondary damage caused by the bearing abnormal vibration, and grouping the training results of the second training data into a second group until the bearing abnormal vibration supervision model to be trained converges to obtain an initial bearing abnormal vibration supervision model;

[0140] Verifying the recognition result of the initial bearing abnormal vibration supervision model on the damage caused by the bearing abnormal vibration based on the first grouping and the second grouping;

[0141] If the recognition result is the same as the training results of the first group and the second group, it is determined that the abnormal vibration supervision model of the bearing to be trained has completed training, and the abnormal vibration supervision model of the bearing is obtained.

[0142] In one embodiment, the monitoring module 40 further includes:

[0143] Collect bearing vibration monitoring data in real time and read bearing abnormal vibration supervision model;

[0144] The bearing vibration monitoring data is analyzed by the bearing abnormal vibration supervision model, and if the bearing vibration monitoring data is greater than or equal to a monitoring threshold, an early warning signal is issued;

[0145] If the bearing vibration monitoring data is less than the monitoring threshold, the bearing vibration monitoring data is updated to the training data set to train and update the bearing abnormal vibration supervision model to obtain a modified bearing abnormal vibration supervision model.

[0146] Reference Figure 3 The embodiment of the present invention further provides a computer device, the internal structure of which can be as follows Figure 3As shown. The computer device includes a processor, a memory, a network interface and a database connected via a system bus. Among them, the processor designed for the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating device, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store relevant data of the monitoring method for abnormal vibration of bearings, etc. The network interface of the computer device is used to communicate with an external terminal through a network connection. Furthermore, the above-mentioned computer device can also be provided with an input device and a display screen, etc. When the above computer program is executed by a processor, a method for monitoring abnormal bearing vibration is implemented, which includes the following steps: obtaining historical bearing monitoring data as a training data set; extracting characteristic data of abnormal bearing vibration in the training data set, and constructing an association relationship between the characteristic data and the training data set; performing memory training on a training abnormal bearing vibration supervision model based on the training data set to obtain a bearing abnormal vibration supervision model; obtaining bearing vibration monitoring data in real time, and analyzing the bearing vibration monitoring data through the bearing abnormal vibration supervision model, and issuing a warning signal when the bearing vibration monitoring data is greater than or equal to a monitoring threshold. Those skilled in the art can understand that Figure 3 The structure shown in is merely a block diagram of a portion of the structure related to the present application solution, and does not constitute a limitation on the computer device on which the present application solution is photographed.

[0147] An embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, a method for monitoring abnormal bearing vibration is implemented, including the following steps: obtaining historical bearing monitoring data as a training data set; extracting characteristic data of abnormal bearing vibration in the training data set, and constructing an association relationship between the characteristic data and the training data set; performing memory training on a training abnormal bearing vibration supervision model based on the training data set to obtain a bearing abnormal vibration supervision model; obtaining bearing vibration monitoring data in real time, and analyzing the bearing vibration monitoring data through the bearing abnormal vibration supervision model, and issuing an early warning signal when the bearing vibration monitoring data is greater than or equal to a monitoring threshold. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0148] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided in this application and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0149] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0150] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for monitoring abnormal vibration of a bearing, characterized in that: The method comprises: Obtain historical bearing monitoring data as a training data set; Extracting characteristic data of abnormal bearing vibration in the training data set, and constructing an association relationship between the characteristic data and the training data set; Performing memory training on the bearing abnormal vibration supervision model to be trained based on the training data set to obtain the bearing abnormal vibration supervision model; Acquire bearing vibration monitoring data in real time, and analyze the bearing vibration monitoring data through the bearing abnormal vibration supervision model, and issue an early warning signal when the bearing vibration monitoring data is greater than or equal to a monitoring threshold; If the historical bearing monitoring data is electrostatic monitoring data of the bearing, then obtaining the historical bearing monitoring data as a training data set includes: Obtain the charge data of the bearing under different acceleration conditions; Obtain the frequency data of bearing rotation under different acceleration conditions; Converting the charge data into a voltage signal according to Laplace transform; The frequency data and the voltage signal are transformed and matched with each other through Fourier transformation to obtain frequency domain features; Filtering the noise in the frequency domain features according to a preset screening rule to obtain effective frequency domain features, wherein the effective frequency domain features are a training data set; The extracting characteristic data of abnormal bearing vibration in the training data set includes: Reading abnormal vibration data of the bearing including the cage, rolling element, outer ring and inner ring in the training data set when the bearing has abnormal vibration; Calculate the peak acceleration of the bearing rotation under different acceleration conditions; Calculating the fault frequencies of the abnormal vibration data according to the acceleration peak values ​​to obtain fault characteristic frequencies including a cage fault characteristic frequency, a rolling element fault characteristic frequency, an outer ring fault characteristic frequency, and an inner ring fault characteristic frequency; Compare the fault characteristic frequency with a preset fault threshold, and if the fault characteristic frequency is less than the preset fault threshold, record it as first abnormal characteristic data; and, If the fault characteristic frequency is greater than or equal to the preset fault threshold, it is recorded as the second abnormal characteristic data; The constructing the association relationship between the feature data and the training data set includes: Filtering out first training data from the training data set according to the first abnormal feature data, establishing a data link between the first abnormal feature data and the first training data, and adding a first identification mark to obtain first training data; Filtering out second training data from the training data set according to the second abnormal feature data, establishing a data link between the second abnormal feature data and the second training data, and adding a second identification mark to obtain second associated data; The memory training of the abnormal vibration supervision model of the bearing to be trained based on the training data set includes: Inputting the training data set into the abnormal vibration supervision model of the bearing to be trained; In the bearing abnormal vibration supervision model to be trained, the first training data is used to train the bearing abnormal vibration supervision model to be trained for the recognition performance of the first-level damage caused by the bearing abnormal vibration, and the training results of the first training data are grouped into a first group; and, Using the second training data to train the bearing abnormal vibration supervision model to be trained to identify the secondary damage caused by the bearing abnormal vibration, and grouping the training results of the second training data into a second group until the bearing abnormal vibration supervision model to be trained converges to obtain an initial bearing abnormal vibration supervision model; Verifying the recognition result of the initial bearing abnormal vibration supervision model on the damage caused by the bearing abnormal vibration based on the first grouping and the second grouping; If the recognition result is the same as the training results of the first group and the second group, it is determined that the abnormal vibration supervision model of the bearing to be trained has completed training, and the abnormal vibration supervision model of the bearing is obtained.

2. The method for monitoring abnormal vibration of a bearing according to claim 1, characterized in that: If the historical bearing monitoring data is bearing position monitoring data, then obtaining the historical bearing monitoring data as a training data set includes: Obtain the force data of radial force and axial force on the bearing when it is installed on the equipment; Determine the load limit point of the bearing according to the force data; A position monitoring data set of the load-limiting force point under different acceleration conditions is obtained, and the position monitoring data set is a training data set.

3. The method for monitoring abnormal vibration of a bearing according to claim 1, characterized in that: The real-time acquisition of bearing vibration monitoring data and the analysis of the bearing vibration monitoring data by the bearing abnormal vibration supervision model, when the bearing vibration monitoring data is greater than or equal to a monitoring threshold, an early warning signal is issued, including: Collect bearing vibration monitoring data in real time and read bearing abnormal vibration supervision model; The bearing vibration monitoring data is analyzed by the bearing abnormal vibration supervision model, and if the bearing vibration monitoring data is greater than or equal to a monitoring threshold, an early warning signal is issued; If the bearing vibration monitoring data is less than the monitoring threshold, the bearing vibration monitoring data is updated to the training data set to train and update the bearing abnormal vibration supervision model to obtain a modified bearing abnormal vibration supervision model.

4. A monitoring device for abnormal bearing vibration, characterized in that: The device comprises: An acquisition module, used for acquiring historical bearing monitoring data as a training data set; An extraction module, used to extract characteristic data of abnormal bearing vibration in the training data set, and to construct an association relationship between the characteristic data and the training data set; A training module, used for performing memory training on the bearing abnormal vibration supervision model to be trained based on the training data set to obtain the bearing abnormal vibration supervision model; A monitoring module is used to obtain bearing vibration monitoring data in real time, and analyze the bearing vibration monitoring data through the bearing abnormal vibration supervision model, and issue an early warning signal when the bearing vibration monitoring data is greater than or equal to a monitoring threshold; If the historical bearing monitoring data is electrostatic monitoring data of the bearing, then obtaining the historical bearing monitoring data as a training data set includes: Obtain the charge data of the bearing under different acceleration conditions; Obtain the frequency data of bearing rotation under different acceleration conditions; Converting the charge data into a voltage signal according to Laplace transform; The frequency data and the voltage signal are transformed and matched with each other through Fourier transformation to obtain frequency domain features; Filtering the noise in the frequency domain features according to a preset screening rule to obtain effective frequency domain features, wherein the effective frequency domain features are a training data set; The extracting characteristic data of abnormal bearing vibration in the training data set includes: Reading abnormal vibration data of the bearing including the cage, rolling element, outer ring and inner ring in the training data set when the bearing has abnormal vibration; Calculate the peak acceleration of the bearing rotation under different acceleration conditions; Calculating the fault frequencies of the abnormal vibration data according to the acceleration peak values ​​to obtain fault characteristic frequencies including a cage fault characteristic frequency, a rolling element fault characteristic frequency, an outer ring fault characteristic frequency, and an inner ring fault characteristic frequency; Compare the fault characteristic frequency with a preset fault threshold, and if the fault characteristic frequency is less than the preset fault threshold, record it as first abnormal characteristic data; and, If the fault characteristic frequency is greater than or equal to the preset fault threshold, it is recorded as the second abnormal characteristic data; The constructing the association relationship between the feature data and the training data set includes: Filtering out first training data from the training data set according to the first abnormal feature data, establishing a data link between the first abnormal feature data and the first training data, and adding a first identification mark to obtain first training data; Filtering out second training data from the training data set according to the second abnormal feature data, establishing a data link between the second abnormal feature data and the second training data, and adding a second identification mark to obtain second associated data; The memory training of the abnormal vibration supervision model of the bearing to be trained based on the training data set includes: Inputting the training data set into the abnormal vibration supervision model of the bearing to be trained; In the bearing abnormal vibration supervision model to be trained, the first training data is used to train the bearing abnormal vibration supervision model to be trained for the recognition performance of the first-level damage caused by the bearing abnormal vibration, and the training results of the first training data are grouped into a first group; and, Using the second training data to train the bearing abnormal vibration supervision model to be trained to identify the secondary damage caused by the bearing abnormal vibration, and grouping the training results of the second training data into a second group until the bearing abnormal vibration supervision model to be trained converges to obtain an initial bearing abnormal vibration supervision model; Verifying the recognition result of the initial bearing abnormal vibration supervision model on the damage caused by the bearing abnormal vibration based on the first grouping and the second grouping; If the recognition result is the same as the training results of the first group and the second group, it is determined that the abnormal vibration supervision model of the bearing to be trained has completed training, and the abnormal vibration supervision model of the bearing is obtained.

5. A computer device comprising a memory and a processor, wherein a computer program is stored in the memory, wherein: When the processor executes the computer program, the steps of the method for monitoring abnormal vibration of a bearing according to any one of claims 1 to 3 are implemented.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for monitoring abnormal vibration of a bearing according to any one of claims 1 to 3 are implemented.

Images