A method for evaluating the dynamic balance performance of miniature fan blades

Abstract

This invention relates to the field of balance assessment, and more specifically, to a method for assessing the dynamic balance performance of micro fan blades. The method includes: reconstructing a high-precision instantaneous phase sequence using Hall signals and performing a two-stage fitting: first, identifying interference frequencies through preliminary fitting; then, constructing a hybrid domain full-basis matrix containing rotor phase basis and time-domain interference basis, and globally fitting the original vibration signal to achieve accurate separation of the target signal and interference. Finally, calculating the unbalance amplitude from the fitting coefficients and calculating the goodness of fit to quantify the assessment measurement confidence. This invention effectively mitigates speed fluctuations, separates interference, and performs self-verification, significantly improving the accuracy and reliability of the assessment.

Description

Technical Field

[0001] This invention relates to the field of balance assessment. More specifically, this invention relates to a method for assessing the dynamic balance performance of miniature fan blades. Background Technology

[0002] The dynamic balance performance of miniature fans is a core quality indicator that determines the product's noise reduction and lifespan. Therefore, online dynamic balance testing of fans on the assembly line is crucial. However, the automated testing station environment is harsh. On the one hand, the fan's own start-stop and control can cause speed fluctuations; on the other hand, servo motors, frequency converters, and switching power supplies on the production line can introduce strong, time-locked electromagnetic and mechanical interference.

[0003] Existing vibration analysis methods struggle to address this challenge. First, traditional frequency domain analysis based on Fast Fourier Transform (FFT) relies on the assumption that the signal must be strictly periodic, i.e., the rotational speed must be constant. When the rotational speed fluctuates, the energy of the target order diffuses across the frequency spectrum, forming fuzzy energy clumps that prevent accurate amplitude extraction. Second, the more advanced Time Synchronous Averaging (TSA) technique, designed to eliminate the effects of rotational speed fluctuations, relies on a crucial assumption: all non-rotor-synchronous signals are random noise. When the production line environment introduces highly deterministic time-domain interference, TSA fails. If the interference has no simple integer ratio to the rotational speed, TSA will misclassify it as noise and attempt to suppress it, but due to its high energy, suppression is often incomplete, leaving residual interference energy that contaminates the evaluation results. If the interference happens to have a specific harmonic relationship with the rotational speed, TSA will retain it as a valid signal, causing it to overlap with the actual imbalance signal and resulting in serious misjudgments.

[0004] Therefore, there is an urgent need in this field for a new technical solution that can simultaneously solve the two coexisting and contradictory technical problems of speed fluctuation and strong time-domain periodic interference, and achieve accurate separation and evaluation of rotor imbalance signals under complex working conditions. Summary of the Invention

[0005] To address the challenge of eliminating speed fluctuations and strong time-domain periodic interference and achieving accurate separation of rotor imbalance signals, this invention proposes a method for evaluating the dynamic balance performance of micro fan blades. This method includes the following steps:

[0006] The main vibration signal and Hall synchronization signal of the micro fan under test are acquired simultaneously to obtain the main vibration sequence and Hall sequence containing a predetermined total number of sampling points;

[0007] Based on the data points of multiple sampling times of the Hall sequence, the instantaneous phase sequence corresponding to each sampling time of the predetermined total number of sampling points is reconstructed;

[0008] The rotor basis matrix is ​​constructed based on the instantaneous phase sequence. The first stage fitting is performed using the main vibration sequence to extract the residual signal sequence. The spectrum analysis of the residual signal sequence is then performed to identify the time-domain interference frequency.

[0009] A time-domain basis matrix is ​​constructed based on the sampling time of the time-domain interference frequency and the predetermined total number of sampling points. The time-domain basis matrix is ​​combined with the rotor basis matrix to form a hybrid domain full basis matrix. The hybrid domain full basis matrix is ​​then used to perform a second-stage fitting on the main vibration sequence to solve for the full basis coefficient vector and the final residual sum of squares.

[0010] The target order coefficients are extracted from the full basis coefficient vector to calculate the unbalance amplitude, and the goodness of fit is calculated based on the final residual sum of squares and the total deviation sum of squares of the master vibration sequence to evaluate the measurement confidence.

[0011] This invention employs a two-stage fitting process—first identifying and then separating—to precisely decouple synchronous rotor vibrations from different physical sources from time-domain fixed-frequency interference. Simultaneously, it introduces goodness-of-fit as an evaluation metric, providing a quantitative basis for the validity of each measurement. This significantly improves the accuracy and reliability of dynamic balancing performance evaluation in complex industrial production line environments.

[0012] Preferably, the main vibration sequence is a composite signal, including: a rotor imbalance signal locked to the instantaneous phase sequence, and a production line periodic interference locked to the sampling time.

[0013] Preferably, the method for reconstructing the instantaneous phase sequence includes:

[0014] Extract the key sampling point index of the pulse edge from the Hall sequence;

[0015] Assign corresponding theoretical Hall phases to the key sampling point indices to form control point groups;

[0016] The instantaneous phase sequence is generated by fitting an interpolator based on the control point group using spline interpolation.

[0017] This invention uses spline interpolation based on the Hall pulse edge to accurately reconstruct the instantaneous rotor angular displacement corresponding to each vibration sampling point from discrete Hall signals. This provides a high-resolution, continuous phase reference for subsequently constructing a velocity-varying rotor basis matrix.

[0018] Preferably, the rotor basis matrix includes multiple sets of orthogonal basis vectors based on the instantaneous phase sequence, the orthogonal basis vectors including and ,in For the target order related to the rotor, This is the index of the sampling time. It represents the instantaneous phase at the index of the sampling time in the instantaneous phase sequence.

[0019] This invention constructs a set of speed-adaptive orthogonal basis vectors for all harmonic vibrations related to rotor rotation. Since the phase of the basis vectors is directly derived from a high-precision instantaneous phase sequence, this matrix can perfectly describe the instantaneous changes in harmonic frequencies caused by speed fluctuations, thereby ensuring that the model can fit the rotor synchronization signal without distortion.

[0020] Preferably, the first stage of fitting is to solve... The least squares solution, where For the rotor basis matrix, The rotor vibration coefficients obtained from the first stage fitting solution;

[0021] The residual signal sequence ,in This refers to the main vibration sequence.

[0022] This invention, by fitting and subtracting the dominant rotor vibration component from the original signal, highlights the relatively weak time-domain periodic interference, which was originally masked by the strong signal, in the residual signal. This creates favorable conditions for subsequent accurate and reliable spectral identification of these interference frequencies.

[0023] Preferably, the time-domain basis matrix includes multiple sets of orthogonal basis vectors based on the sampling time, sampling frequency, and time-domain interference frequency, wherein the orthogonal basis vectors include and , Indicates the sampling frequency. Here, n represents the time-domain interference frequency, and n is the index of the sampling time.

[0024] The hybrid domain full basis matrix is ​​constructed by matrix splicing of the rotor basis matrix and the time-domain basis matrix.

[0025] This invention constructs a complete linear model space for the composite signal under test, capable of describing all known major signal components. This matrix, through concatenation, integrates two sets of orthogonal subspaces used to describe rotor synchronization signals and time-domain fixed-frequency signals, respectively, thus forming a unified mathematical framework capable of synchronous and unbiased parameter estimation for these two types of signals with different physical origins.

[0026] Preferably, the second-stage fitting is achieved by solving the mixed-domain full basis matrix. The least squares solution is used to obtain the full basis coefficient vector and the final residual sum of squares; where The main vibration sequence, For the final residual sequence, The vector of all basis coefficients. It is a mixed-domain full basis matrix.

[0027] This invention solves the mixed-domain full basis matrix using least squares. This process can optimally distribute the total signal energy to each component based on the correlation between each basis vector and the original signal. Even when the rotor harmonic frequency is very close to the interference frequency, this method can effectively suppress spectral leakage and energy misjudgment, achieving accurate decoupling of the coefficients of each signal component.

[0028] Preferably, the unbalanced amplitude satisfies the following relationship:

[0029]

[0030] in, and These are the cosine and sine coefficients corresponding to the first order, extracted from the full basis coefficient vector. This represents the unbalanced amplitude.

[0031] Preferably, the goodness of fit satisfies the following relationship:

[0032]

[0033] in, It is the final sum of squared residuals. It is the total sum of squares of deviations of the main vibration sequence. This indicates the goodness of fit.

[0034] Preferred options also include:

[0035] The goodness of fit is compared with a preset goodness of fit threshold;

[0036] When the goodness of fit is lower than the goodness of fit threshold, the calculated unbalance amplitude is determined to be invalid, and a measurement alarm is output.

[0037] The present invention has the following beneficial effects:

[0038] This invention constructs an innovative hybrid domain full-basis fitting model, which can accurately decouple the rotor synchronization signal under speed fluctuations from the fixed frequency interference in the production line environment. This fundamentally overcomes the problem of spectral aliasing and distortion in traditional frequency domain analysis methods under non-ideal working conditions, ensuring the accuracy of the evaluation results.

[0039] Furthermore, goodness of fit is introduced as a confidence index for each measurement, enabling the system to perform self-verification. This mechanism can automatically identify and eliminate invalid data caused by unmodeled abnormal disturbances, ensuring the long-term reliability and consistency of the final output dynamic balance performance data.

[0040] Furthermore, it proactively identifies and isolates environmental interference through algorithms, rather than relying on physical vibration isolation or strictly constant-speed fan operation. Therefore, it can be directly applied to real production line environments with background vibration and power supply fluctuations, significantly reducing the stringent environmental control requirements of high-precision dynamic balancing tests and possessing strong engineering practical value. Attached Figure Description

[0041] Figure 1 This is a flowchart illustrating the steps of a method for evaluating the dynamic balance performance of a micro fan blade provided in an embodiment of the present invention.

[0042] Figure 2 These are the images of the main vibration signal and rotor signal provided in the embodiments of the present invention;

[0043] Figure 3 The first-stage fitting residual spectrum is provided for an embodiment of the present invention;

[0044] Figure 4 A comparison diagram of the second-stage fitting results and the main vibration signal provided in an embodiment of the present invention. Detailed Implementation

[0045] Please see Figure 1 The diagram illustrates a flowchart of a method for evaluating the dynamic balance performance of a micro fan blade provided in Embodiment 1. The method includes the following steps:

[0046] S1: Synchronously acquire the main vibration signal and Hall synchronization signal of the micro fan under test to obtain the main vibration sequence and Hall sequence containing a predetermined total number of sampling points.

[0047] It should be noted that in a production line environment, the main vibration sequence captured by the sensor is a physically mixed signal. It includes both the target vibration signal caused by fan imbalance and production line interference caused by servo motors, etc. In order to separate the true target vibration signal from this mixed signal, the main vibration sequence alone is not enough; a Hall effect sequence must be acquired simultaneously. This provides a phase reference for calibrating the fan rotation position and provides a data foundation for subsequent angular domain analysis.

[0048] Specifically, at the automated testing station, the main accelerometer is mounted on a test fixture supporting a miniature fan. The main vibration signal is acquired at a preset sampling frequency for a preset duration, and the Hall signal from the fan motor is simultaneously acquired at the same sampling frequency and duration. For example, the preset sampling frequency is... The preset duration is 2 seconds.

[0049] The main vibration signals collected at all sampling times are arranged in time sequence to obtain the main vibration sequence.

[0050] Similarly, the Hall signals collected at all sampling times are arranged in time sequence to obtain the Hall sequence.

[0051] Figure 2 The image shows the main vibration signal and the rotor signal. The blue curve in the image shows a schematic diagram of the main vibration sequence, and the red curve shows a schematic diagram of the rotor signal. It can be seen from the image that the rotor signal is masked by other signals.

[0052] S2: Based on the data points of multiple sampling times of the Hall sequence, the instantaneous phase sequence corresponding to each sampling time of the predetermined total number of sampling points is reconstructed.

[0053] It should be noted that the Hall sequence acquired in the above steps is a discrete and non-ideal phase reference. Its limitation lies in the fact that it only generates pulse edges when the fan rotates to a specific angle, resulting in sparse phase information in time, and the timing of the pulse edges is affected by jitter and is therefore inaccurate. Directly using this sparse and inaccurate phase information for subsequent order analysis will introduce significant computational errors. Therefore, this step must use an interpolation algorithm to reconstruct and refine these discrete phase reference points to generate a high-resolution, smooth instantaneous phase sequence. This sequence can provide an accurate physical angle estimate at each sampling moment, thus providing a data foundation for subsequent angular domain analysis to eliminate the influence of rotational speed fluctuations.

[0054] Preferably, as an example, based on the data points of multiple sampling times of the Hall sequence, the instantaneous phase sequence corresponding to each sampling time of the predetermined total number of sampling points is reconstructed, including:

[0055] First, using a threshold comparison method or a differential method, the sampling point indices of all pulse edges are extracted from the Hall sequence and denoted as key sampling indices.

[0056] Next, based on prior knowledge of the fan motor structure, known theoretical Hall phases are assigned to the key sampling point indices to obtain a control point group. For example, per revolution... Each pulse is assigned a known theoretical Hall phase, if Then the theoretical Hall phase sequence is .

[0057] Subsequently, based on the control point set, a continuous mathematical model was fitted using cubic spline interpolation as an interpolator. This interpolator physically reproduced the smooth nonlinear relationship between the fan rotor and its rotation angle at any sampling time.

[0058] Finally, all sampling times are input into the interpolator for calculation to obtain a high-resolution instantaneous phase sequence.

[0059] Understandably, these discrete phase reference points are reconstructed and refined using interpolation algorithms to generate a high-resolution, smooth instantaneous phase sequence. This sequence provides an accurate physical angle estimate at each sampling moment, thus providing a data foundation for subsequent angular domain analysis to eliminate the influence of rotational speed fluctuations.

[0060] S3: Construct the rotor basis matrix based on the instantaneous phase sequence, perform the first stage fitting using the main vibration sequence to extract the residual signal sequence, and perform spectral analysis on the residual signal sequence to identify the time-domain interference frequency.

[0061] It should be noted that the main vibration sequence is mixed with unbalanced vibrations from the fan and time-domain vibration interference from the production line. Therefore, in order to achieve accurate fan imbalance analysis, it is necessary to identify these time-domain vibration interferences from the production line. If the original main vibration sequence is directly subjected to spectral analysis, the broadband spectrum caused by the speed fluctuation will obscure the narrowband spectrum of the time-domain vibration interference from the production line, making it difficult to identify.

[0062] It should be further explained that, since the instantaneous phase sequence mainly reflects the fan rotation information and is less affected by production line timing information, it can be used to fit a theoretical rotor vibration model. This model can then be used to fit the unbalanced vibration components of all fans in the main vibration sequence. After subtracting these fitted rotor synchronization components from the main vibration sequence, the resulting residual signal sequence is physically composed mainly of time-domain interference and random noise. Spectral analysis of this residual signal sequence then reveals a clear spectral background, highlighting the frequency peaks of time-domain interference, making it easy to automatically identify.

[0063] S30: Construct the rotor basis matrix based on the instantaneous phase sequence.

[0064] Preferably, as an example, constructing the rotor basis matrix based on the instantaneous phase sequence includes:

[0065] Iterate through all sampling time indices n, and calculate for each n. and Where k is the preset rotor order, such as k=1,2,3). The first in the instantaneous phase sequence The instantaneous phase indexed at each sampling time point reflects the physical rotation angle of the fan. and Reflected respectively in the The sampling time, the first The instantaneous amplitudes of the cosine and sine components of the order rotor vibration.

[0066] These calculation results are used as elements of the nth row to fill a matrix, resulting in the rotor basis matrix. The rotor basis matrix is ​​a... The matrix, where Q represents the total number of sampling times, This represents the total number of rotor orders. Its [number]th [order]... The elements of the row describe the first row. The theoretical rotor vibration waveform components at each sampling time.

[0067] Understandably, fitting the rotor basis matrix is ​​the fundamental descriptive model for the unbalanced vibration of all fans.

[0068] S31: Perform a first-stage fitting using the master vibration sequence to extract the residual signal sequence.

[0069] It should be noted that the above steps fit the basic description model of the fan's unbalanced vibration. Therefore, by describing the fan's unbalanced vibration signal through the basic description model, and removing the fan's unbalanced vibration from the main vibration sequence, we can obtain the vibration components that are out of sync with the fan's rotation, namely production line interference and random noise.

[0070] Preferably, as an example, the first-stage fitting is performed using the master vibration sequence to extract the residual signal sequence, and spectral analysis is performed on the residual signal sequence to identify time-domain interference frequencies, including:

[0071] The residual signal sequence is fitted using the least squares method, and the residual signal sequence satisfies the following relationship:

[0072]

[0073] in, It is a residual signal sequence. The main vibration sequence, For the rotor basis matrix, The rotor vibration coefficients are obtained from the fitting solution in the first stage. The unbalanced vibration signal of the fan described using the basic description model.

[0074] Understandably, the vibration components related to fan rotation are subtracted from the original main vibration sequence. The resulting residual signal sequence consists of all vibration components that are out of sync with fan rotation, primarily including production line interference and random noise that need to be identified.

[0075] S32: Perform spectral analysis on the residual signal sequence to identify time-domain interference frequencies.

[0076] Preferably, as an example, performing spectral analysis on the residual signal sequence to identify time-domain interference frequencies includes:

[0077] Perform a Fast Fourier Transform (FFT) on the residual signal sequence, and then perform a peak search algorithm to obtain the time-domain interference frequency.

[0078] Understandably, after removing the rotor vibration component from the main vibration sequence, the spectrum here only contains production line interference signals and random noise information. The signal components are relatively simple, so the time-domain interference frequency of the production line can be easily and accurately identified.

[0079] S4: Construct a time-domain basis matrix based on the sampling time of the time-domain interference frequency and the predetermined total number of sampling points. Combine the time-domain basis matrix with the rotor basis matrix to form a hybrid domain full basis matrix. Use the hybrid domain full basis matrix to perform a second-stage fitting on the main vibration sequence to solve for the full basis coefficient vector and the final residual sum of squares.

[0080] It should be noted that although a model describing the fan vibration signal has been constructed, the lack of a directly corresponding fan vibration signal to build the relevant equations prevents accurate fitting of the parameters describing the fan vibration signal. Currently, the main vibration signal is known to be a mixed signal containing both fan vibration and production line interference signals. If a descriptive model can also be constructed for the production line interference signal within the main vibration signal, combining these two descriptive models will yield a comprehensive descriptive model that can describe the main vibration signal. Furthermore, based on the descriptive model of the active signal, the main vibration signal can be described to construct the relevant equations, thereby accurately fitting the parameters of the fan vibration signal.

[0081] S40: Construct a time-domain basis matrix based on the sampling time of the time-domain interference frequency and the predetermined total number of sampling points.

[0082] Preferably, as an example, constructing a time-domain basis matrix based on the sampling times of the time-domain interference frequency and the predetermined total number of sampling points includes:

[0083] Traverse all sampling time indexes n takes values ​​from 1 to n , The number of sampling times, and for each calculate and And fill the first part of the matrix with the calculation results. Okay, we get the time-domain basis matrix.

[0084] For time-domain interference frequency, The preset sampling frequency, Reflects the first The absolute time corresponding to each sampling point This reflects the angular frequency under an interference waveform. , They respectively reflect the first Each sampling time, with a frequency of The instantaneous amplitudes of the cosine and sine components of the theoretical disturbance vibration.

[0085] It is understandable that the time-domain basis matrix is ​​a fundamental descriptive model for interference signals on the production line.

[0086] Figure 3 The image shows the residual spectrum of the first-stage fitting. As can be seen from the image, the time-domain interference frequency can be identified relatively accurately through the first-stage fitting.

[0087] S41: Combine the time-domain basis matrix with the rotor basis matrix to form a mixed-domain full-basis matrix.

[0088] Preferably, as an example, combining the time-domain basis matrix and the rotor basis matrix into a mixed-domain full-basis matrix includes:

[0089] rotor basis matrix With the time-domain basis matrix Perform a matrix horizontal concatenation operation to obtain the full basis matrix of the mixed domain. .

[0090] It is understandable that the hybrid domain full basis matrix is ​​a hybrid model that can describe fan vibration signals and production line interference signals.

[0091] Figure 4 The image shows a comparison between the second-stage fitting results and the main vibration signal. The image reveals that the difference between the second-stage fitting results and the main vibration signal is small, indicating that the constructed fitting model can accurately fit the signal. Therefore, accurate signal separation can be achieved based on this model.

[0092] S42: The mixed-domain full basis matrix is ​​used to perform a second-stage fitting on the main vibration sequence to solve for the full basis coefficient vector and the final residual sum of squares.

[0093] It should be noted that the above steps yielded a hybrid model that can describe the fan vibration signal and the production line interference signal. The hybrid model is then used to fit the main vibration signal, and the fitting relationship is used to fit the relevant parameters required to describe the fan vibration signal.

[0094] Preferably, as an example, a second-stage fitting is performed on the master vibration sequence using the mixed-domain full basis matrix to solve for the full basis coefficient vector and the final residual sum of squares, including:

[0095] The fitting relationship is obtained by fitting the main vibration sequence using the least squares method:

[0096]

[0097] in, The main vibration sequence, For a mixed-domain full basis matrix, Let be the vector of all basis coefficients that need to be fitted and solved. This is the final residual sequence for fitting.

[0098] It is understandable that the hybrid domain full basis matrix contains both the descriptive model of the fan vibration signal and the descriptive model of the production line interference signal. Therefore, the main vibration sequence can be well fitted through the hybrid domain full basis matrix, and based on this relatively accurate fitting relationship, the relevant parameters required to describe the fan vibration signal can be fitted relatively accurately.

[0099] Furthermore, the rotor basis matrix and the time-domain basis matrix in the mixed-domain full basis matrix are mathematically orthogonal. The least squares solver can accurately segment the energy in the original main vibration sequence to the corresponding domain. In other words, the coefficients in the full basis coefficient vector corresponding to the rotor basis matrix reflect the energy information of the fan vibration signal, and the coefficients in the full basis coefficient vector corresponding to the time-domain basis matrix reflect the energy information related to production line interference.

[0100] S5: Extract the target order coefficients from the full basis coefficient vector to calculate the unbalance amplitude, and calculate the goodness of fit based on the final residual sum of squares and the total deviation sum of squares of the main vibration sequence to evaluate the measurement confidence.

[0101] It should be noted that the above steps can extract the energy information of the fan vibration, and the imbalance of the fan can be analyzed based on this energy information.

[0102] S50: Extract the target order coefficients from the full basis coefficient vector to calculate the unbalance magnitude.

[0103] Preferably, as an example, extracting the target order coefficients from the full basis coefficient vector to calculate the unbalance magnitude includes:

[0104] Extracting the first-order cosine coefficients from the full basis coefficient vector and sine coefficient And based on this, the unbalance amplitude is calculated:

[0105]

[0106] in, This represents the amplitude of unbalanced vibration.

[0107] It is understandable that the first-order cosine and sine coefficients are related parameters corresponding to the rotor basis matrix. These parameters mainly describe the energy situation of fan vibration, and therefore the unbalance amplitude calculated by them reflects the unbalance of the fan.

[0108] S51: Calculate the goodness of fit based on the final residual sum of squares and the total sum of squares of deviations of the master vibration sequence to evaluate the measurement confidence.

[0109] It should be noted that the reliability of the calculated unbalanced vibration amplitude needs to be analyzed to determine whether the fan imbalance needs to be reassessed.

[0110] Preferably, as an example, the goodness of fit is calculated based on the final sum of squared residuals and the total sum of squared deviations of the master vibration sequence to assess measurement confidence, including:

[0111] Calculate the goodness of fit:

[0112]

[0113] in, The goodness of fit is used to reflect the reliability of the calculated unbalanced amplitude. The variance of the final residual sequence, The variance of the main vibration sequence.

[0114] Understandably, if the energy of the final residual sequence is large, it indicates a lower reliability of the signal fit, which in turn indicates a worse signal separation effect and a less accurate fan imbalance measurement; otherwise, it indicates a better signal separation effect and a more accurate fan imbalance measurement.

[0115] The goodness-of-fit is compared with a preset goodness-of-fit threshold. For example, the goodness-of-fit threshold is set to... .

[0116] If the goodness of fit is greater than the goodness of fit threshold, the measurement is considered reliable. If the goodness of fit is lower than the goodness of fit threshold, it indicates the presence of unknown interference, resulting in a sharp increase in residual energy. The system then alarms that the measurement is invalid, thus preventing the use of erroneous unbalanced amplitude readings for evaluation.

[0117] This concludes the embodiment.

[0118] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for evaluating the dynamic balance performance of a micro fan blade, characterized in that, The method comprises: synchronously collecting a main vibration signal and a Hall synchronization signal of a measured micro-fan to obtain a main vibration sequence and a Hall sequence containing a predetermined total number of sampling points; reconstructing an instantaneous phase sequence corresponding to each sampling time of the predetermined total number of sampling points based on data points of multiple sampling times of the Hall sequence; constructing a rotor basis matrix based on the instantaneous phase sequence, performing a first stage fitting using the main vibration sequence to extract a residual signal sequence, and performing a frequency spectrum analysis on the residual signal sequence to identify a time domain interference frequency; constructing a time domain basis matrix based on the time domain interference frequency and the sampling time of the predetermined total number of sampling points, combining the time domain basis matrix and the rotor basis matrix into a hybrid domain full basis matrix, and performing a second stage fitting on the main vibration sequence using the hybrid domain full basis matrix to solve a full basis coefficient vector and a final residual sum of squares; extracting target order coefficients from the full basis coefficient vector to calculate an unbalance amplitude, and calculating a goodness of fit based on the final residual sum of squares and a total deviation sum of squares of the main vibration sequence to evaluate a measurement confidence.

2. The method for evaluating the dynamic balance performance of a micro fan blade according to claim 1, wherein, The main vibration sequence is a composite signal, which includes a rotor unbalance signal locked with the instantaneous phase sequence and a line periodic interference locked with the sampling time.

3. The method of claim 1, wherein the method further comprises: The reconstruction method of the instantaneous phase sequence comprises: extracting key sampling point indexes of pulse edges from the Hall sequence; assigning corresponding theoretical Hall phases to the key sampling point indexes to form a control point group; fitting an interpolator based on the control point group using a spline interpolation method to generate the instantaneous phase sequence.

4. The method of claim 1, wherein the micro fan blade dynamic balance performance evaluation method is characterized by, The rotor basis matrix comprises a plurality of sets of orthogonal basis vectors based on the instantaneous phase sequence, the orthogonal basis vectors comprising and wherein is a rotor dependent target order, is an index of a sampling instant, is an instantaneous phase in the instantaneous phase sequence at the sampling instant index.

5. The method of claim 1, wherein the micro fan blade dynamic balance performance evaluation method is characterized by, The first stage fit is to solve the least square solution of where is the rotor base matrix, is the rotor vibration coefficient solved by the first stage fit; said sequence of residual signals wherein is said sequence of primary vibrations.

6. The method of claim 1, wherein the micro fan blade dynamic balance performance evaluation method is characterized by, The time-domain basis matrix comprises a plurality of sets of orthogonal basis vectors based on the sampling time, sampling frequency and the time-domain interference frequency, the orthogonal basis vectors comprising and , denotes the sampling frequency, is the time-domain interference frequency, and n is an index of the sampling time. The hybrid domain full basis matrix is constructed by matrix splicing the rotor basis matrix and the time domain basis matrix.

7. The method of claim 1 or 6, wherein The second stage fitting obtains the full basis coefficient vector and the final residual sum of squares by solving for the least squares solution of ; wherein is the primary vibration sequence, is the final residual sequence, is the full basis coefficient vector, is the hybrid domain full basis matrix.

8. The method of claim 1, wherein the micro fan blade dynamic balance performance evaluation method is characterized by, The unbalance amplitude satisfies the relationship: wherein and are cosine and sine coefficients of order 1 extracted from said full basis coefficient vector, is the unbalance amplitude.

9. The method of claim 7, wherein the method further comprises: The goodness of fit satisfies the relationship: wherein is the final sum of squared residuals, is the total sum of squared deviations of the main vibration sequence, denotes the goodness of fit.

10. The method of claim 1 or 9, wherein The method further comprises: comparing the goodness of fit with a preset goodness of fit threshold; when the goodness of fit is lower than the goodness of fit threshold, determining that the calculated unbalance amplitude is invalid, and outputting a measurement alarm.

Images