A method for detecting and analyzing vibration and noise in miniature bearings
By constructing a testing condition consistent with the target assembly scenario, collecting and calculating the operating condition response data of miniature bearings, and establishing an anomaly source discrimination model, the problem of identifying and distinguishing the source of early weak abnormal noises in miniature bearings was solved, achieving more accurate detection and risk assessment.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- NANTONG SK SEIKO CO LTD
- Filing Date
- 2026-06-03
- Publication Date
- 2026-06-30
Smart Images
Figure CN122306422A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of noise detection technology, and in particular to a method for detecting and analyzing vibration noise in miniature bearings. Background Technology
[0002] Miniature bearings are widely used in miniature fan motors, camera module drive mechanisms, miniature medical pumps, gimbal motors, and other small, precision rotating devices. These bearings are characterized by their small size, high speed, compact structure, sensitivity to assembly preload, and low noise tolerance. Problems such as localized raceway damage, cage misalignment, deteriorated lubrication, or assembly misalignment can often generate hissing, high-pitched noises, whistling, or intermittent abnormal sounds during operation, directly impacting the acoustic quality, operational stability, and user experience of the final product. Therefore, accurately identifying and analyzing the sources of abnormal vibration and noise during the screening process, pre-installation testing, and failure analysis of miniature bearings has become a crucial issue in this technical field.
[0003] Most existing methods for testing miniature bearings employ vibration acquisition under constant speed or single load conditions, judging the bearing's condition through time-domain statistics, frequency-domain peak values, or simple threshold rules. Some schemes also simultaneously acquire acoustic signals, combining this with empirical frequency band analysis or general machine learning methods to arrive at anomaly conclusions. While these methods are effective for larger bearings or those with more pronounced faults, they are less suitable for miniature bearings. Due to their low vibration response amplitude, weak initial anomaly characteristics, and significant influence from assembly boundaries, the performance of the same bearing in standard testing fixtures often differs significantly from its performance in actual products. This is especially true in miniature fan motors and camera module drive mechanisms, where axial preload, torque pulsation, directional off-center loading, and the acoustic transmission boundaries of the housing collectively alter the bearing's vibration and noise characteristics. This makes static testing results on traditional test benches insufficient to accurately reflect the overall machine's abnormal noise risk.
[0004] Furthermore, most existing technologies directly analyze the collected vibration and acoustic data in a unified manner. While this can determine whether an anomaly exists, it is difficult to further distinguish whether the anomaly originates from the raceway, cage, lubrication condition, or assembly misalignment. This is because different sources of anomalies often exhibit similar spectral increases or amplitude increases under constant detection conditions. The lack of a detection process that actively reveals the differences in various anomaly mechanisms leads to detection conclusions that typically remain at the level of whether an anomaly exists, rather than reaching the analytical depth of where the anomaly originates, under what operating conditions it will be amplified, and whether it will translate into abnormal noise in the entire machine. For critical components like miniature bearings, which are designed for high-quietness and miniaturized products, the inability to accurately identify the source of anomalies makes it difficult to guide subsequent process modifications, assembly parameter optimization, and quality screening strategy formulation.
[0005] Therefore, a new method for detecting and analyzing vibration and noise in miniature bearings is needed. This method should not be limited to passive detection under a single constant speed condition, but should be able to construct a realistic detection boundary by combining the target assembly scenario. Vibration response data and acoustic response data should be collected under multiple distinguishing operating conditions. Furthermore, operating condition response characterization data that can represent the abnormality pattern should be calculated, and an anomaly source discrimination model should be established based on historical detection datasets. This will enable stable identification of early weak abnormal noises in miniature bearings, analysis of abnormal source, and risk assessment of abnormal noises during installation. Summary of the Invention
[0006] The technical problem solved by this invention is that existing technologies are unable to stably identify and accurately distinguish the source of early weak abnormal noises under the combined effects of real assembly constraints, dynamic working condition excitation, and near-field acoustic-vibration coupling propagation of miniature bearings.
[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution: A method for detecting and analyzing vibration and noise in miniature bearings includes the following steps: Step S1: Construct the detection conditions corresponding to the target assembly scenario, obtain historical excitation data, historical vibration response data and historical acoustic response data from historical detection records, and establish a historical detection dataset; Step S2: Apply real-time excitation corresponding to the test condition to the micro bearing under test, collect real-time excitation data, real-time vibration response data and real-time acoustic response data, and establish a real-time test dataset; Step S3: Calculate the working condition response characterization data based on the historical detection dataset and the real-time detection dataset. The working condition response characterization data includes disturbance-sensitive trajectory data, energy redistribution trajectory data, and repeating visible trajectory data. Step S4: Establish an anomaly source discrimination model based on the working condition response characterization data in the historical detection dataset, input the working condition response characterization data corresponding to the real-time detection dataset into the anomaly source discrimination model, and output anomaly source analysis data. The anomaly source analysis data includes raceway anomaly data, cage anomaly data, lubrication anomaly data, and assembly eccentricity anomaly data. Step S5: Calculate abnormal noise risk data based on the abnormal source analysis data and the real-time detection dataset, and output detection conclusion data based on the abnormal noise risk data.
[0008] Preferably, step S1 includes the following sub-steps: Step S101: Obtain historical detection records corresponding to the target assembly scene. The historical detection records include historical working condition excitation data, historical vibration response data, and historical acoustic response data. Step S102: Classify the historical operating condition excitation data to obtain historical gradually changing speed excitation data, historical pulsating speed excitation data, historical axial preload excitation data, historical directional off-center load excitation data, and historical stop-start transition excitation data; Step S103: Associate and store various historical operating condition excitation data with corresponding historical vibration response data and historical acoustic response data to establish a historical detection dataset.
[0009] Preferably, step S2 includes the following sub-steps: Step S201: Install the miniature bearing to be tested in the detection position corresponding to the target assembly scene; Step S202: Apply real-time operating condition excitation to the micro bearing under test according to the operating condition category corresponding to the historical test dataset. The real-time operating condition excitation includes real-time gradual speed change excitation, real-time pulsating speed excitation, real-time axial preload excitation, real-time directional off-center load excitation, and real-time stop-start transition excitation. Step S203: During the application of real-time operating condition excitation, real-time vibration response data and real-time acoustic response data are collected simultaneously. Step S204: Store the real-time operating condition excitation data, real-time vibration response data, and real-time acoustic response data accordingly to establish a real-time detection dataset.
[0010] Preferably, step S3 includes the following sub-steps: Step S301: Based on historical operating condition excitation data and real-time operating condition excitation data, extract the vibration response segment and acoustic response segment corresponding to each operating condition stage respectively. Step S302: Calculate the disturbance-sensitive trajectory data based on the amplitude changes of the vibration response segment and acoustic response segment in each working condition stage as the working condition excitation changes. Step S303: Calculate the energy redistribution trajectory data based on the frequency band energy distribution changes of the vibration response segment and acoustic response segment in each working condition stage; Step S304: Calculate the repeated visible trajectory data based on the repeated triggering positions and repeated triggering intensities of the vibration response segments and acoustic response segments under multiple working condition excitations in each working condition stage. Step S305: Output the disturbance-sensitive trajectory data, energy redistribution trajectory data, and repetitive visible trajectory data as operating condition response characterization data.
[0011] Preferably, the calculation logic for the disturbance-sensitive trajectory data in step S302 is as follows: Based on the difference between adjacent excitation quantities in each working condition stage, calculate the vibration change of the corresponding vibration response segment and the acoustic change of the corresponding acoustic response segment. By correlating the vibration change with the corresponding excitation change, a vibration-sensitive change sequence is obtained. By correlating acoustic changes with corresponding excitation changes, an acoustically sensitive change sequence is obtained. Based on the vibration-sensitive change sequence and the acoustic-sensitive change sequence, disturbance-sensitive trajectory data corresponding to each working condition stage are generated.
[0012] Preferably, the calculation logic for the energy redistribution trajectory data in step S303 is as follows: Calculate the vibration frequency band energy data of the vibration response segment and the acoustic frequency band energy data of the acoustic response segment in each working condition stage; Based on the changes in vibration frequency band energy data between adjacent operating conditions, a vibration energy migration sequence is obtained; Based on the changes in acoustic frequency band energy data between adjacent operating conditions, an acoustic energy migration sequence is obtained; Energy redistribution trajectory data are established based on vibrational energy migration sequences and acoustic energy migration sequences.
[0013] Preferably, the calculation logic for the repeated visible trajectory data in step S304 is as follows: Peak trigger positions were statistically analyzed for vibration response segments collected multiple times under the same working conditions to obtain repeated vibration trigger data. Peak trigger positions were statistically analyzed for acoustic response segments acquired multiple times under the same operating conditions to obtain acoustic repeated trigger data. Vibration repetition intensity data is calculated based on vibration repetition trigger data, and acoustic repetition intensity data is calculated based on acoustic repetition trigger data. By correlating vibration repetition intensity data and acoustic repetition intensity data, repeating visible trajectory data is formed.
[0014] Preferably, step S4 includes the following sub-steps: Step S401: Establish an anomaly source discrimination model based on the working condition response characterization data in the historical detection dataset, and establish the correspondence between the anomaly source type and the working condition response characterization data; Step S402: Input the working condition response characterization data corresponding to the real-time detection dataset into the anomaly source discrimination model, and output the anomaly source score data corresponding to each anomaly source type; Step S403: Compare the anomaly source score data, find the anomaly source type with the highest score, and output the anomaly source analysis data.
[0015] Preferably, the analysis logic of step S403 is as follows: If the repetitive visible trajectory data in the real-time operating condition response characterization data continues to increase under real-time slowly varying speed excitation and real-time pulsating speed excitation, then the anomaly source score data corresponding to the raceway anomaly data will be improved. If the disturbance-sensitive trajectory data in the real-time operating condition response characterization data continues to increase under real-time pulsating speed excitation and real-time directional off-center load excitation, then the anomaly source score data corresponding to the cage anomaly data will be improved. If the energy redistribution trajectory data in the real-time operating condition response characterization data continues to increase under real-time axial preload excitation and real-time stop-start transition excitation, then the abnormal source score data corresponding to the lubrication abnormal data will be improved. If the disturbance-sensitive trajectory data and energy redistribution trajectory data in the real-time operating condition response characterization data increase synchronously under real-time slowly varying speed excitation and real-time directional off-center load excitation, then the anomaly source score data corresponding to the assembly eccentricity anomaly data will be improved.
[0016] Preferably, step S5 includes the following sub-steps: Step S501: Calculate abnormal noise risk data based on abnormal source analysis data and real-time detection dataset. The abnormal noise risk data includes installed abnormal noise probability data and operating condition amplification risk data. Step S502: If the installation abnormal noise probability data is greater than or equal to the preset abnormal noise threshold, then output the abnormal interception conclusion data. Step S503: If the installation abnormal noise probability data is less than the preset abnormal noise threshold, and the operating condition amplification risk data is greater than or equal to the preset amplification threshold, then output the conclusion data of limiting the use of the operating condition. Step S504: If the probability data of abnormal noise from the installed equipment is less than the preset abnormal noise threshold, and the risk data of amplified operating conditions is less than the preset amplification threshold, then output the qualified conclusion data. Step S505: Output the abnormal interception conclusion data, the restricted operating condition use conclusion data, and the qualified conclusion data as the test conclusion data.
[0017] The beneficial effects of this invention are as follows: By constructing a detection position consistent with the target assembly scenario, and collecting vibration response data and acoustic response data under conditions such as gradually changing speed, pulsating speed, axial preload, directional off-center loading, and stop-start transition, this invention further calculates disturbance-sensitive trajectory data, energy redistribution trajectory data, and repetitive visible trajectory data. This enables the differentiation and analysis of raceway abnormalities, cage abnormalities, lubrication abnormalities, and assembly eccentricity abnormalities. Compared with existing technologies, this invention can more realistically reflect the abnormal noise performance of miniature bearings after installation, improve the early weak abnormality identification capability and the accuracy of abnormality source analysis, and can output installation abnormal noise risk conclusions. Attached Figure Description
[0018] Figure 1 The flowchart illustrates the steps of a method for detecting and analyzing vibration and noise in a miniature bearing, as provided in one embodiment of the present invention. Detailed Implementation
[0019] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.
[0020] Example, refer to Figure 1 This paper provides a method for detecting and analyzing vibration and noise in miniature bearings, comprising the following steps: Step S1: Construct the detection conditions corresponding to the target assembly scenario, obtain historical excitation data, historical vibration response data and historical acoustic response data from historical detection records, and establish a historical detection dataset.
[0021] Step S2: Apply real-time operating condition excitation corresponding to the test condition to the micro bearing under test, collect real-time operating condition excitation data, real-time vibration response data and real-time acoustic response data, and establish a real-time test dataset.
[0022] Step S3: Calculate the working condition response characterization data based on the historical detection dataset and the real-time detection dataset. The working condition response characterization data includes disturbance-sensitive trajectory data, energy redistribution trajectory data, and repeating visible trajectory data.
[0023] Step S4: Establish an anomaly source discrimination model based on the working condition response characterization data in the historical detection dataset. Input the working condition response characterization data corresponding to the real-time detection dataset into the anomaly source discrimination model and output anomaly source analysis data. The anomaly source analysis data includes raceway anomaly data, cage anomaly data, lubrication anomaly data, and assembly eccentricity anomaly data.
[0024] Step S5: Calculate abnormal noise risk data based on the abnormal source analysis data and the real-time detection dataset, and output detection conclusion data based on the abnormal noise risk data.
[0025] This method is applicable to factory testing, pre-installation screening, and abnormal noise failure analysis of miniature bearings such as miniature fan motor bearings, camera module drive bearings, miniature pump bearings, and gimbal motor bearings. Unlike conventional methods that directly collect vibration or noise data under constant speed and load conditions, this embodiment constructs a detection position consistent with the target assembly scenario and applies discriminative excitation during the testing process. This causes different sources of anomalies to exhibit different response amplification patterns at different operating stages. Then, a discriminant model for the anomaly source is established using historical testing datasets, thereby obtaining anomaly source analysis data and abnormal noise risk data.
[0026] Step S1 includes the following sub-steps: Step S101: Obtain historical detection records corresponding to the target assembly scenario. The historical detection records include historical working condition excitation data, historical vibration response data, and historical acoustic response data.
[0027] Step S102: Classify the historical operating condition excitation data to obtain historical gradually changing speed excitation data, historical pulsating speed excitation data, historical axial preload excitation data, historical directional off-center load excitation data, and historical stop-start transition excitation data.
[0028] Step S103: Associate and store various historical operating condition excitation data with corresponding historical vibration response data and historical acoustic response data to establish a historical detection dataset.
[0029] In this embodiment, a detection position consistent with the target assembly scenario is first constructed. If the object under test is a miniature fan motor bearing, the detection position includes a simulated motor end cover constraint, an axial preload structure, a shaft connection stiffness structure, and a housing acoustic transmission boundary structure; if the object under test is a camera module drive bearing, the detection position includes a module clamping structure, a drive unit mounting base, and a near-field acoustic reflection boundary structure. The purpose of constructing the detection position is to ensure that historical detection records and real-time detection records form a comparable data source under the same assembly boundary, avoiding deviations between the bare bearing detection results and the overall assembly results.
[0030] Subsequently, historical test records were acquired, which included at least historical operating condition excitation data, historical vibration response data, and historical acoustic response data. The historical operating condition excitation data included historical gradually changing speed excitation data, historical pulsating speed excitation data, historical axial preload excitation data, historical directional off-center load excitation data, and historical start-stop transition excitation data. Historical vibration response data was collected by a miniature accelerometer installed at the support end of the test position, with a sampling frequency set to... Historical acoustic response data were acquired by a miniature sound pressure sensor positioned near the micro-bearing under test, with a sampling frequency set to [value missing]. All historical inspection records are recorded with corresponding abnormality source labels. The abnormality source labels only use four categories: raceway abnormality, cage abnormality, lubrication abnormality, and assembly eccentricity abnormality.
[0031] In this embodiment, historical operating condition excitation data is categorized. Let the total number of historical detection records be... , No. The historical test record is recorded as Then we have: ; in, This represents historical operating condition incentive data. This represents historical vibration response data. This represents historical acoustic response data. This indicates the source of the anomaly. Then... The excitation segments are categorized into five types based on operating conditions: ; in, This is historical data on gradually varying rotational speeds. Historical pulsating speed excitation data, This is based on historical axial preload excitation data. For historical directional biased incentive data, This provides historical start-stop transition excitation data. Historical vibration response data segments and historical acoustic response data segments are stored in one-to-one correspondence with excitation segments under each operating condition, thus forming a historical detection dataset.
[0032] Step S2 includes the following sub-steps: Step S201: Install the miniature bearing to be tested in the detection position corresponding to the target assembly scene; Step S202: Apply real-time operating condition excitation to the micro bearing under test according to the operating condition category corresponding to the historical test dataset. The real-time operating condition excitation includes real-time gradual speed excitation, real-time pulsating speed excitation, real-time axial preload excitation, real-time directional off-center load excitation, and real-time stop-start transition excitation.
[0033] Step S203: During the application of real-time operating condition excitation, real-time vibration response data and real-time acoustic response data are collected simultaneously. Step S204: Store the real-time operating condition excitation data, real-time vibration response data, and real-time acoustic response data accordingly to establish a real-time detection dataset.
[0034] In this embodiment, the miniature bearing under test is installed in the same detection position as in step S1. During detection, real-time operating condition excitation is applied to the miniature bearing under test. The real-time operating condition excitation uses the same operating condition category and excitation order as the historical operating condition excitation data, which includes real-time gradually changing speed excitation, real-time pulsating speed excitation, real-time axial preload excitation, real-time directional off-center load excitation, and real-time stop-start transition excitation. The purpose of this setting is to keep the real-time detection dataset consistent with the historical detection dataset in the operating condition dimension, so that the subsequent model can map the real-time operating condition response characterization data to the historical anomaly source space.
[0035] In real-time operating condition excitation, gradually changing speed excitation adopts interval... Linear variation sequence within; pulsating speed excitation at the center speed Nearby according to amplitude Periodic fluctuations; axial preload excitation within the preload range Internal graded application; directional off-center excitation applies a small off-center load in a predetermined direction. The stop-start transition excitation record tracks the complete excitation change process from shutdown to restart. Let the real-time detection record be... ; in, This represents real-time operating condition excitation data. This represents real-time vibration response data. This represents real-time acoustic response data. Further details include: ; Synchronous vibration response segments were collected during each operating condition phase. With synchronous acoustic response segments ,in All obtained real-time operating condition excitation data, real-time vibration response data, and real-time acoustic response data are included in the subsequent characterization data calculation process and are not discarded.
[0036] Step S3 includes the following sub-steps: Step S301: Based on historical operating condition excitation data and real-time operating condition excitation data, extract the vibration response segment and acoustic response segment corresponding to each operating condition stage.
[0037] Step S302: Calculate the disturbance-sensitive trajectory data based on the amplitude changes of the vibration response segment and acoustic response segment in each working condition stage as the working condition excitation changes.
[0038] Step S303: Calculate the energy redistribution trajectory data based on the frequency band energy distribution changes of the vibration response segment and acoustic response segment in each working condition stage.
[0039] Step S304: Calculate the repeated visible trajectory data based on the repeated triggering positions and repeated triggering intensities of the vibration response segments and acoustic response segments under multiple operating condition excitations in each working condition stage.
[0040] Step S305: Output the disturbance-sensitive trajectory data, energy redistribution trajectory data, and repetitive visible trajectory data as operating condition response characterization data.
[0041] The calculation logic for the disturbance-sensitive trajectory data in step S302 is as follows: Based on the difference between adjacent excitation quantities in each working condition stage, calculate the vibration change of the corresponding vibration response segment and the acoustic change of the corresponding acoustic response segment.
[0042] By correlating the vibration change with the corresponding excitation change, a vibration-sensitive change sequence is obtained.
[0043] By correlating acoustic changes with corresponding excitation changes, an acoustically sensitive change sequence is obtained.
[0044] Based on the vibration-sensitive change sequence and the acoustic-sensitive change sequence, disturbance-sensitive trajectory data corresponding to each working condition stage are generated.
[0045] Disturbance-sensitive trajectory data are used to characterize the sensitivity of vibration and acoustic responses to minute changes in operating excitation. Let the... The excitation amount of each working condition sampling point is The corresponding vibration envelope amplitude is The acoustic envelope amplitude is The change in excitation between adjacent sampling points Vibration change Harmonic variation They are defined as follows: ; ; ; The vibration sensitivity coefficient and the acoustic sensitivity coefficient are further defined as follows: ; ; in, This represents the average amplitude of the vibration envelope during this operating condition. This represents the average amplitude of the acoustic envelope during this operating condition. To prevent the denominator from approaching zero, a stable term is used. A weighted average is calculated for all sampling points under the same operating condition to obtain the disturbance sensitivity index for that condition. ; in, For the incentive phase weights, the following condition must be met: ; The disturbance sensitivity indices corresponding to the five operating conditions are connected in sequence to form disturbance sensitivity trajectory data: ; The physical meaning of this data is that if a certain source of anomaly rapidly increases both the vibration envelope amplitude and the acoustic envelope amplitude when the excitation under the operating conditions changes slightly, then the disturbance sensitivity index corresponding to that source of anomaly will increase significantly. For example, cage anomalies often exhibit a high disturbance sensitivity index under pulsating speed excitation, and assembly eccentricity anomalies often exhibit a high disturbance sensitivity index under slowly varying speed excitation and directional off-center load excitation.
[0046] The calculation logic for the energy redistribution trajectory data in step S303 is as follows: Calculate the vibration frequency band energy data of the vibration response segment and the acoustic frequency band energy data of the acoustic response segment in each working condition stage.
[0047] Based on the changes in vibration frequency band energy data between adjacent operating conditions, a vibration energy migration sequence is obtained.
[0048] The acoustic energy migration sequence is obtained based on the changes in acoustic frequency band energy data between adjacent operating conditions.
[0049] Energy redistribution trajectory data are established based on vibrational energy migration sequences and acoustic energy migration sequences.
[0050] Energy redistribution trajectory data is used to characterize the migration patterns of vibrational and acoustic energy across different frequency bands. First, the vibration spectrum... Acoustic spectrum Divided into The frequency band, the first The vibrational energy and acoustic energy of each frequency band are defined as follows: ; ; Then, the energy of each frequency band is normalized to obtain the energy ratio of the vibration frequency band. Harmony and acoustic frequency band energy ratio : ; ; Let the frequency band energy ratios corresponding to two adjacent sampling points under different operating conditions be respectively , and , Then, the vibrational energy transfer and acoustic energy transfer are defined as follows: ; ; The energy redistribution index for this operating condition is redefined as: ; in, For energy transfer weights, satisfying: ; Connect the energy redistribution indices corresponding to the five operating conditions in sequence to form energy redistribution trajectory data: ; The physical meaning of this data is that if an abnormal source causes vibration energy to redistribute across frequency bands, and acoustic energy also migrates synchronously, then its energy redistribution index is high. Lubrication abnormalities usually cause significant energy redistribution under axial preload excitation and start-stop transition excitation. This is because changes in the lubrication film state cause the contact state to change from relatively stable to locally friction-dominated, thus causing broadband vibration energy and broadband acoustic energy to change synchronously.
[0051] The calculation logic for the repeated visible trajectory data in step S304 is as follows: Peak trigger positions were statistically analyzed for vibration response segments collected multiple times under the same working conditions to obtain repeated vibration trigger data.
[0052] Peak trigger positions were statistically analyzed for acoustic response segments acquired multiple times under similar operating conditions to obtain acoustic repetitive trigger data.
[0053] Vibration repetition intensity data is calculated based on vibration repetition trigger data, and acoustic repetition intensity data is calculated based on acoustic repetition trigger data.
[0054] By correlating vibration repetition intensity data and acoustic repetition intensity data, repeating visible trajectory data is formed.
[0055] Repeated explicit trajectory data are used to characterize whether anomalous responses recur at the same or similar locations under multiple similar operating conditions. For the first instance of similar operating condition excitation... After repeated detections, let the set of significant peak locations detected in the vibration response segment be denoted as . The set of significant peak locations detected in the acoustic response segment is Perform location clustering on all duplicate detection results. The center position of a certain location cluster is denoted as... The corresponding vibration repetition count and acoustic repetition count are defined as follows: ; ; in, This represents the number of repeated tests. For The neighborhood window is centered. This is an indicator function.
[0056] Further define the first The repetition intensity of the cluster at each location is: ; in, This represents the degree of dispersion of the peak positions within this location cluster. This is the discrete penalty coefficient. Then, summing over all location clusters yields the repetition exponent for this operating condition stage: ; The repetitive visibility indices corresponding to the five types of working conditions are connected in sequence to form repetitive visibility trajectory data: ; The physical significance of this data is that if there is a local anomaly in the raceway of the miniature bearing, the vibration peak and noise peak will repeatedly appear at several fixed positions during the excitation of slowly varying speed or pulsating speed, resulting in a significant increase in the repetition index; if there is only random environmental disturbance, the significant peak positions will be dispersed and will not form a high repetition index.
[0057] Step S305: For any detection record, merge the disturbance-sensitive trajectory data, energy redistribution trajectory data, and repeated visible trajectory data to form the operating condition response characterization data, denoted as: ; For historical detection records The data obtained are historical operating condition response characterization data. For real-time detection records, real-time operating condition response characterization data are obtained. At this point, historical operating condition excitation data, historical vibration response data, historical acoustic response data, real-time operating condition excitation data, real-time vibration response data, and real-time acoustic response data have all been transformed into operating condition response characterization data that can be directly used in subsequent models.
[0058] Step S4 includes the following sub-steps: Step S401: Establish an anomaly source discrimination model based on the working condition response characterization data in the historical detection dataset, and establish the correspondence between the anomaly source type and the working condition response characterization data.
[0059] Step S402: Input the working condition response characterization data corresponding to the real-time detection dataset into the anomaly source discrimination model, and output the anomaly source score data corresponding to each anomaly source type.
[0060] Step S403: Compare the anomaly source score data, find the anomaly source type with the highest score, and output the anomaly source analysis data.
[0061] The analysis logic for step S403 is as follows: If the repetitive visible trajectory data in the real-time operating condition response characterization data continues to increase under real-time slowly varying speed excitation and real-time pulsating speed excitation, then the anomaly source score data corresponding to the raceway anomaly data will be improved.
[0062] If the disturbance-sensitive trajectory data in the real-time operating condition response characterization data continues to increase under real-time pulsating speed excitation and real-time directional off-center load excitation, then the anomaly source score data corresponding to the cage anomaly data will be improved.
[0063] If the energy redistribution trajectory data in the real-time operating condition response characterization data continues to increase under real-time axial preload excitation and real-time stop-start transition excitation, then the anomaly source score data corresponding to the lubrication anomaly data will be improved.
[0064] If the disturbance-sensitive trajectory data and energy redistribution trajectory data in the real-time operating condition response characterization data increase synchronously under real-time slowly varying speed excitation and real-time directional off-center load excitation, then the anomaly source score data corresponding to the assembly eccentricity anomaly data will be improved.
[0065] In this embodiment, the anomaly source discrimination model takes historical operating condition response characterization data as input and anomaly source labels from historical detection records as output, and is established using a combination of weighted distance discrimination and probability scoring. Let the total number of historical operating condition response characterization data be... , No. The historical operating condition response characterization data are as follows Its real-time corresponding tag is First, all historical operating condition response characterization data are divided into four sets according to the anomaly source label, corresponding to raceway anomaly, cage anomaly, lubrication anomaly, and assembly eccentricity anomaly, respectively.
[0066] For any anomaly source category, calculate the class center vector and the intra-class discrete matrix for that category. For real-time operational response characterization data, define its weighted distance relative to the category. for: ; Then, construct category scores based on weighted distances. Let the score temperature parameter be... Then the original score of the category Defined as: ; The four types of raw scores were normalized to obtain anomaly source scoring data. : ; in, This represents the set of anomaly source categories. From this, we obtain four anomaly source score data items: raceway anomaly data, cage anomaly data, lubrication anomaly data, and assembly eccentricity anomaly data.
[0067] To ensure the judgment logic aligns with the operational condition manifestation mechanism of this invention, rule corrections are further applied after model scoring. If the repetitive manifest trajectory data in the real-time operational condition response characterization data continuously increases under real-time gradual speed excitation and real-time pulsating speed excitation, an enhancement factor is applied to the raceway anomaly data; if the disturbance-sensitive trajectory data in the real-time operational condition response characterization data continuously increases under real-time pulsating speed excitation and real-time directional off-center load excitation, an enhancement factor is applied to the cage anomaly data; if the energy redistribution trajectory data in the real-time operational condition response characterization data continuously increases under real-time axial preload excitation and real-time stop-start transition excitation, an enhancement factor is applied to the lubrication anomaly data; if the disturbance-sensitive trajectory data and energy redistribution trajectory data in the real-time operational condition response characterization data increase simultaneously under real-time gradual speed excitation and real-time directional off-center load excitation, an enhancement factor is applied to the assembly eccentricity anomaly data. After the enhanced scores are re-normalized, the category corresponding to the highest score is taken as the anomaly source analysis data output.
[0068] Step S5 includes the following sub-steps: Step S501: Calculate abnormal noise risk data based on abnormal source analysis data and real-time detection dataset. Abnormal noise risk data includes installation abnormal noise probability data and operating condition amplification risk data.
[0069] Step S502: If the installation abnormal noise probability data is greater than or equal to the preset abnormal noise threshold, then output the abnormal interception conclusion data.
[0070] Step S503: If the probability data of abnormal noise from the installed equipment is less than the preset abnormal noise threshold, and the risk data of amplified operating conditions is greater than or equal to the preset amplification threshold, then output the conclusion data of limiting the use of operating conditions.
[0071] Step S504: If the probability data of abnormal noise from the installed equipment is less than the preset abnormal noise threshold, and the risk data of amplified operating conditions is less than the preset amplification threshold, then output the qualified conclusion data.
[0072] Step S505: Output the abnormal interception conclusion data, the restricted operating condition use conclusion data, and the qualified conclusion data as the test conclusion data.
[0073] After obtaining the anomaly source analysis data, this embodiment further calculates abnormal noise risk data based on the anomaly source analysis data and the real-time detection dataset. The abnormal noise risk data includes installation abnormal noise probability data and operating condition amplification risk data. The installation abnormal noise probability data characterizes the probability that the tested miniature bearing will generate a user-perceptible abnormal noise in the target assembly scenario, while the operating condition amplification risk data characterizes the degree of risk of further amplification under specific operating conditions.
[0074] Let the category with the highest anomaly source score be Its rating is Simultaneously, the operating condition response characteristics most relevant to this category are extracted to construct an installed unit abnormal noise risk vector. For example, when When the raceway is abnormal, select , , As a major risk component; when For abnormal lubrication, select , , As the primary risk component, let the risk weight vector be... The bias term is The probability data for abnormal noise during installation is defined as follows: ; in, The value ranges from 0 to 1, with a higher value indicating a higher probability of abnormal noise during installation.
[0075] Operating condition amplification risk data is used to determine whether a certain anomaly will be further amplified under specific operating conditions. Let the increments of the operating condition response characteristics corresponding to the five types of operating conditions be: ; in, This indicates that the real-time detection data is under operating conditions. The operating condition response characterization components, This indicates that historical normal samples are under operating conditions. The mean value represents the components. The amplified risk data under the working condition is redefined as follows: ; in, This refers to the risk weighting of a particular operating condition. If a certain operating condition stage corresponds to... If the value is significantly higher than in other operating conditions, it indicates that the anomaly has a significant risk of being amplified under that operating condition.
[0076] Finally, based on the probability data of abnormal noise during installation. and operating condition amplification risk data Output the detection conclusion data. If... Greater than or equal to the preset abnormal noise threshold Then output the exception interception conclusion data; if Less than the preset abnormal noise threshold ,and Greater than or equal to the preset amplification threshold Then output the conclusion data of the limiting operating conditions; if Less than the preset abnormal noise threshold, and Less than the preset amplification threshold If the result is satisfactory, then the output will be a qualified conclusion.
[0077] This invention first constructs a testing position consistent with the target assembly scenario, ensuring that the miniature bearing is simultaneously subjected to axial preload, support stiffness, and near-field sound transmission boundary conditions similar to the actual product during testing. This avoids the problem of abnormal noise in the actual machine despite passing the test bench test, a common issue in traditional bare bearing testing. For end products with high acoustic quality requirements, such as miniature fan motors, camera module drive mechanisms, miniature pumps, and gimbal motors, this method significantly improves the consistency between test results and actual usage scenarios.
[0078] This invention further employs gradually varying speed excitation, pulsating speed excitation, axial preload excitation, directional off-center load excitation, and stop-start transition excitation to constitute the detection conditions. It no longer relies on passive sampling under a single constant speed condition, but actively induces response differences from different anomaly sources through differentiated condition changes. Since raceway anomalies, cage anomalies, lubrication anomalies, and assembly eccentricity anomalies exhibit different response amplification patterns at different stages of the detection process, this invention enables anomaly mechanisms that are difficult to distinguish under conventional detection conditions to be revealed under specific conditions, thus providing a reliable data foundation for subsequent anomaly source analysis.
[0079] This invention does not simply perform threshold judgments on the raw vibration and acoustic response data. Instead, it further calculates disturbance-sensitive trajectory data, energy redistribution trajectory data, and repetitive manifest trajectory data, transforming the raw acquired data into condition response characterization data capable of representing the anomaly mechanism. Specifically, the disturbance-sensitive trajectory data reflects the sensitivity of the anomaly to minute changes in the operating conditions; the energy redistribution trajectory data reflects the migration patterns of vibration and acoustic energy across different frequency bands; and the repetitive manifest trajectory data reflects the repeated triggering characteristics of the anomaly response under multiple excitations. This data processing method overcomes the limitations of existing technologies that only focus on the mean, peak value, kurtosis, or single spectral peak value, elevating the detection result from simply whether the signal is abnormal to how the anomaly manifests under specific operating conditions.
[0080] This invention also establishes an anomaly source discrimination model based on historical testing datasets, and utilizes the correspondence between operating condition response characterization data and anomaly source labels to identify anomalies in the output raceway, cage, lubrication, and assembly eccentricity data of the miniature bearing under test. Therefore, this invention can not only determine whether anomalies exist in the miniature bearing under test, but also further determine which specific location or mechanism the anomaly is more likely to originate from. These detection results have direct guiding significance for defect tracing in production, parameter correction in processes, and preload control in assembly, significantly improving the usability of the detection and analysis results.
[0081] Furthermore, based on the anomaly source analysis data, this invention further calculates the probability data of abnormal noise from the installed equipment and the risk data of amplified operating conditions, and outputs anomaly interception conclusion data, operating condition restriction conclusion data, or qualified conclusion data. Therefore, the detection results are no longer limited to simple qualified or unqualified judgments, but can provide quality decision conclusions that are closer to the actual usage needs of the end product. In other words, this invention can not only detect anomalies, but also determine whether the anomaly will develop into a user-perceptible abnormal noise under specific operating conditions, thereby improving the targeting of quality screening and reducing over-rejection and missed detection.
[0082] In summary, this invention achieves stable detection of early-stage subtle abnormal noises in miniature bearings, accurate analysis of abnormal sources, and effective prediction of installation risks through a complete technical chain, including the construction of real assembly boundaries, the application of differentiated operating condition excitations, the calculation of operating condition response characterization data, the analysis of abnormal source discrimination models, and the risk assessment of installation abnormal noises. It can significantly improve the problems of insufficient detection accuracy, weak ability to distinguish abnormal sources, and insufficiently refined screening decisions in existing technologies under real-world application scenarios of miniature bearings.
[0083] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0084] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the protection scope of the present invention.
Claims
1. A method for detecting and analyzing vibration and noise in miniature bearings, characterized in that, Includes the following steps: Step S1: Construct the detection conditions corresponding to the target assembly scenario, obtain historical excitation data, historical vibration response data and historical acoustic response data from historical detection records, and establish a historical detection dataset; Step S2: Apply real-time excitation corresponding to the test condition to the micro bearing under test, collect real-time excitation data, real-time vibration response data and real-time acoustic response data, and establish a real-time test dataset; Step S3: Calculate the working condition response characterization data based on the historical detection dataset and the real-time detection dataset. The working condition response characterization data includes disturbance-sensitive trajectory data, energy redistribution trajectory data, and repeating visible trajectory data. Step S4: Establish an anomaly source discrimination model based on the working condition response characterization data in the historical detection dataset, input the working condition response characterization data corresponding to the real-time detection dataset into the anomaly source discrimination model, and output anomaly source analysis data. The anomaly source analysis data includes raceway anomaly data, cage anomaly data, lubrication anomaly data, and assembly eccentricity anomaly data. Step S5: Calculate abnormal noise risk data based on the abnormal source analysis data and the real-time detection dataset, and output detection conclusion data based on the abnormal noise risk data.
2. The method for detecting and analyzing vibration and noise in miniature bearings as described in claim 1, characterized in that, Step S1 includes the following sub-steps: Step S101: Obtain historical detection records corresponding to the target assembly scene. The historical detection records include historical working condition excitation data, historical vibration response data, and historical acoustic response data. Step S102: Classify the historical operating condition excitation data to obtain historical gradually changing speed excitation data, historical pulsating speed excitation data, historical axial preload excitation data, historical directional off-center load excitation data, and historical stop-start transition excitation data; Step S103: Associate and store various historical operating condition excitation data with corresponding historical vibration response data and historical acoustic response data to establish a historical detection dataset.
3. The method for detecting and analyzing vibration and noise in miniature bearings as described in claim 2, characterized in that, Step S2 includes the following sub-steps: Step S201: Install the miniature bearing to be tested in the detection position corresponding to the target assembly scene; Step S202: Apply real-time operating condition excitation to the micro bearing under test according to the operating condition category corresponding to the historical test dataset. The real-time operating condition excitation includes real-time gradual speed change excitation, real-time pulsating speed excitation, real-time axial preload excitation, real-time directional off-center load excitation, and real-time stop-start transition excitation. Step S203: During the application of real-time operating condition excitation, real-time vibration response data and real-time acoustic response data are collected simultaneously. Step S204: Store the real-time operating condition excitation data, real-time vibration response data, and real-time acoustic response data accordingly to establish a real-time detection dataset.
4. The method for detecting and analyzing vibration and noise in miniature bearings as described in claim 3, characterized in that, Step S3 includes the following sub-steps: Step S301: Based on historical operating condition excitation data and real-time operating condition excitation data, extract the vibration response segment and acoustic response segment corresponding to each operating condition stage respectively. Step S302: Calculate the disturbance-sensitive trajectory data based on the amplitude changes of the vibration response segment and acoustic response segment in each working condition stage as the working condition excitation changes. Step S303: Calculate the energy redistribution trajectory data based on the frequency band energy distribution changes of the vibration response segment and acoustic response segment in each working condition stage; Step S304: Calculate the repeated visible trajectory data based on the repeated triggering positions and repeated triggering intensities of the vibration response segments and acoustic response segments under multiple working condition excitations in each working condition stage. Step S305: Output the disturbance-sensitive trajectory data, energy redistribution trajectory data, and repetitive visible trajectory data as operating condition response characterization data.
5. The method for detecting and analyzing vibration and noise in miniature bearings as described in claim 4, characterized in that, The calculation logic for the disturbance-sensitive trajectory data in step S302 is as follows: Based on the difference between adjacent excitation quantities in each working condition stage, calculate the vibration change of the corresponding vibration response segment and the acoustic change of the corresponding acoustic response segment. By correlating the vibration change with the corresponding excitation change, a vibration-sensitive change sequence is obtained. By correlating acoustic changes with corresponding excitation changes, an acoustically sensitive change sequence is obtained. Based on the vibration-sensitive change sequence and the acoustic-sensitive change sequence, disturbance-sensitive trajectory data corresponding to each working condition stage are generated.
6. The method for detecting and analyzing vibration and noise in miniature bearings as described in claim 5, characterized in that, The calculation logic for the energy redistribution trajectory data in step S303 is as follows: Calculate the vibration frequency band energy data of the vibration response segment and the acoustic frequency band energy data of the acoustic response segment in each working condition stage; Based on the changes in vibration frequency band energy data between adjacent operating conditions, a vibration energy migration sequence is obtained; Based on the changes in acoustic frequency band energy data between adjacent operating conditions, an acoustic energy migration sequence is obtained; Energy redistribution trajectory data are established based on vibrational energy migration sequences and acoustic energy migration sequences.
7. The method for detecting and analyzing vibration and noise in miniature bearings as described in claim 6, characterized in that, The calculation logic for the repeated visible trajectory data in step S304 is as follows: Peak trigger positions were statistically analyzed for vibration response segments collected multiple times under the same working conditions to obtain repeated vibration trigger data. Peak trigger positions were statistically analyzed for acoustic response segments acquired multiple times under the same operating conditions to obtain acoustic repeated trigger data. Vibration repetition intensity data is calculated based on vibration repetition trigger data, and acoustic repetition intensity data is calculated based on acoustic repetition trigger data. By correlating vibration repetition intensity data and acoustic repetition intensity data, repeating visible trajectory data is formed.
8. The method for detecting and analyzing vibration and noise in miniature bearings as described in claim 7, characterized in that, Step S4 includes the following sub-steps: Step S401: Establish an anomaly source discrimination model based on the working condition response characterization data in the historical detection dataset, and establish the correspondence between the anomaly source type and the working condition response characterization data; Step S402: Input the working condition response characterization data corresponding to the real-time detection dataset into the anomaly source discrimination model, and output the anomaly source score data corresponding to each anomaly source type; Step S403: Compare the anomaly source score data, find the anomaly source type with the highest score, and output the anomaly source analysis data.
9. The method for detecting and analyzing vibration and noise in miniature bearings as described in claim 8, characterized in that, The analysis logic of step S403 is as follows: If the repetitive visible trajectory data in the real-time operating condition response characterization data continues to increase under real-time slowly varying speed excitation and real-time pulsating speed excitation, then the anomaly source score data corresponding to the raceway anomaly data will be improved. If the disturbance-sensitive trajectory data in the real-time operating condition response characterization data continues to increase under real-time pulsating speed excitation and real-time directional off-center load excitation, then the anomaly source score data corresponding to the cage anomaly data will be improved. If the energy redistribution trajectory data in the real-time operating condition response characterization data continues to increase under real-time axial preload excitation and real-time stop-start transition excitation, then the abnormality source score data corresponding to the lubrication abnormality data will be improved. If the disturbance-sensitive trajectory data and energy redistribution trajectory data in the real-time operating condition response characterization data increase synchronously under real-time gradually changing speed excitation and real-time directional off-center load excitation, then the anomaly source score data corresponding to the assembly eccentricity anomaly data will be improved.
10. The method for detecting and analyzing vibration noise in a miniature bearing as described in claim 9, characterized in that, Step S5 includes the following sub-steps: Step S501: Calculate abnormal noise risk data based on abnormal source analysis data and real-time detection dataset. The abnormal noise risk data includes installed abnormal noise probability data and operating condition amplification risk data. Step S502: If the installation abnormal noise probability data is greater than or equal to the preset abnormal noise threshold, then output the abnormal interception conclusion data. Step S503: If the installation abnormal noise probability data is less than the preset abnormal noise threshold, and the operating condition amplification risk data is greater than or equal to the preset amplification threshold, then output the conclusion data of limiting the use of the operating condition. Step S504: If the probability data of abnormal noise from the installed equipment is less than the preset abnormal noise threshold, and the risk data of amplified operating conditions is less than the preset amplification threshold, then output the qualified conclusion data. Step S505: Output the abnormal interception conclusion data, the restricted operating condition use conclusion data, and the qualified conclusion data as the test conclusion data.