Dynamic balancing adjustment apparatus and method for a magnetic levitation fan rotor

Abstract

The application discloses a magnetic suspension fan rotor dynamic balance adjusting device and method, and relates to the technical field of self-adaptive adjustment.The method comprises the following steps: collecting suspension bearing displacement signals, magnetic suspension coil current signals, machine body vibration acceleration signals, motor end electromagnetic response signals and rotor angular displacement parameters, and constructing a rotor operation signal set; respectively pre-processing and extracting features of the signals to obtain four types of feature data; inputting the comprehensive feature data into a rotor balance diagnosis model to identify the rotor unbalance state; solving the amplitude and phase adjustment amount of the virtual counterweight based on the amplitude and phase inversion algorithm and injecting the suspension coil to perform adjustment; judging whether dynamic correction is needed according to the real-time comprehensive feature data, generating a dynamic gain coefficient to correct the adjustment parameters when needed, and then injecting the suspension coil again; the application realizes real-time identification and self-adaptive adjustment of the mass unbalance vibration of the magnetic suspension fan rotor, effectively suppresses the rotor operation vibration, and improves the balance stability and reliability of the system.

Description

Technical Field

[0001] This invention relates to the field of adaptive adjustment technology, and more specifically, to a device and method for dynamic balancing adjustment of magnetic levitation fan rotors. Background Technology

[0002] Magnetic levitation fans, due to their low friction loss, high-speed operation capability, and maintenance-free characteristics, have been widely used in high-efficiency ventilation, compression, and energy recovery. The magnetic levitation rotor achieves non-contact levitation support through electromagnetic force, and its operational stability is highly dependent on the rotor's mass distribution and the precision of the magnetic levitation control. However, during long-term operation, the rotor is prone to slight mass imbalances due to manufacturing and assembly processes, material wear, or external disturbances. This imbalance causes periodic excitation forces during rotor rotation, resulting in fluctuations in levitation displacement, electromagnetic current disturbances, and increased machine vibration. This not only affects the fan's operational stability and noise control but may also lead to frequent corrections of the levitation control system, shortening the equipment's lifespan.

[0003] Existing magnetic levitation fan balancing technologies mainly rely on static counterweights or manual offline dynamic balancing methods. These methods require shutdown for counterweight adjustment, a cumbersome process that cannot address dynamic imbalances that occur during operation. Some studies have attempted to adjust imbalances through electromagnetic compensation, but these methods generally suffer from technical shortcomings. Specifically, existing solutions rely on a single sensing signal, failing to simultaneously reflect the rotor's motion state and electromagnetic response characteristics, resulting in low imbalance identification accuracy. Compensation parameters often employ fixed gain control, making it difficult to adapt to dynamic changes in imbalance characteristics under load fluctuations or speed variations. Furthermore, the lack of a feedback correction mechanism for the adjustment effect easily leads to overcompensation or undercompensation, resulting in insufficient system convergence and stability.

[0004] Therefore, how to accurately identify, adaptively adjust, and continuously optimize the vibration caused by the mass imbalance of the magnetic levitation fan rotor without shutting down the machine has become a pressing technical problem in this field. This invention addresses this problem by proposing a dynamic balancing adjustment device and method for magnetic levitation fan rotors based on multi-source vibration sensing and amplitude-reverse algorithm, enabling real-time detection and electromagnetic self-compensation adjustment of rotor imbalance vibration. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art and achieve the above objectives, the present invention provides the following technical solution: a method for dynamic balance adjustment of a magnetic levitation fan rotor, comprising:

[0006] The collection includes rotor operation signal sets such as suspension bearing displacement signal, magnetic levitation coil current signal, body vibration acceleration signal, motor end electromagnetic response signal, and rotor angular displacement parameters;

[0007] Preprocessing and feature extraction of the suspension bearing displacement signal yields first feature data including the major axis of the rotor motion trajectory, the minor axis of the rotor motion trajectory, the ellipticity of the rotor motion trajectory axis, the displacement synchronization amplitude, and the displacement synchronization phase.

[0008] Preprocessing and feature extraction are performed on the vibration acceleration signal of the body to obtain second feature data including structural energy concentration, fundamental frequency amplitude of acceleration and structural phase shift.

[0009] The electromagnetic response signal at the motor end is preprocessed and its features are extracted to obtain the third feature data, including the fundamental frequency amplitude of the electromagnetic response, the air gap side band ratio, and the magnetoelectric phase difference indication.

[0010] Preprocessing and feature extraction are performed on the magnetic levitation coil current signal to obtain fourth feature data including the proportion of synchronization energy and the current harmonic distortion coefficient;

[0011] The comprehensive feature data, including the first feature data, the second feature data, the third feature data, and the fourth feature data, is input into the rotor balance diagnosis model to identify the unbalanced vibration caused by rotor mass imbalance during operation and generate a rotor unbalanced state identifier.

[0012] The adjustment parameters are solved based on the amplitude-inverse transformation algorithm and injected into the magnetic levitation coil for adjustment. Based on the real-time comprehensive characteristic data, it is determined whether the adjustment parameters still need to be dynamically adjusted. If the determination result is yes, a dynamic gain coefficient is generated to correct the adjustment parameters before injecting them into the magnetic levitation coil. The adjustment parameters include the virtual counterweight amplitude and the phase adjustment amount.

[0013] Furthermore, methods for solving the adjustment parameters based on the amplitude-inverse transformation algorithm include:

[0014] The displacement synchronization amplitude is converted into the unbalanced force amplitude based on the equivalent dynamic proportional coefficient corresponding to the entire magnetic levitation fan.

[0015] The phase of the unbalanced force is calculated based on the displacement synchronization phase and the structural phase offset.

[0016] Construct a complex vector of unbalanced force based on the amplitude and phase of the unbalanced force;

[0017] The virtual counterweight amplitude is constructed based on the principle of antiphase superposition of unbalanced force complex vectors;

[0018] The phase adjustment amount is calculated based on the unbalanced force phase and the magnetoelectric phase difference indication;

[0019] The virtual counterweight amplitude and phase adjustment amount are constructed as adjustment parameters.

[0020] Furthermore, methods for determining whether adjustment parameters still need dynamic adjustment based on real-time comprehensive characteristic data include:

[0021] The real-time comprehensive feature data is input into the rotor balance diagnosis model to obtain the second rotor balance diagnosis score;

[0022] If the second rotor balance diagnostic score is greater than or equal to the preset rotor balance score threshold, the output judgment result is negative and the current diagnostic process ends.

[0023] If the second rotor balance diagnostic score is less than the preset rotor balance score threshold, the output judgment result is yes, and the current diagnostic process ends.

[0024] Furthermore, methods for generating dynamic gain coefficients to correct adjustment parameters include:

[0025] The second rotor balance diagnostic score is matched with the pre-constructed score-gain coefficient matching table to obtain the corresponding dynamic gain coefficient.

[0026] Add the dynamic gain coefficient to the constant 1 to obtain the adjustment coefficient used for correction calculation;

[0027] Multiply the adjustment coefficient by the virtual weight amplitude and the phase adjustment amount respectively to obtain the corrected virtual weight amplitude and the corrected phase adjustment amount;

[0028] The corrected virtual counterweight amplitude and the corrected phase adjustment amount are used to construct the corrected adjustment parameters.

[0029] Furthermore, the method for generating the rotor imbalance state identifier includes:

[0030] The comprehensive feature data, including the first feature data, the second feature data, the third feature data, and the fourth feature data, is input into the rotor balance diagnosis model to obtain the first rotor balance diagnosis score.

[0031] If the first rotor balance diagnosis score is greater than or equal to the preset rotor balance score threshold, no rotor imbalance status identifier is generated, and the current diagnosis process ends.

[0032] If the first rotor balance diagnosis score is less than the preset rotor balance score threshold, a rotor imbalance status indicator is generated, and the current diagnosis process ends.

[0033] Furthermore, the method for obtaining the first feature data includes:

[0034] The displacement signal of the suspension bearing is orthogonally decomposed and fitted with the axis trajectory to obtain the elliptic equation of the rotor motion trajectory; the elliptic equation of the rotor motion trajectory is solved to obtain the major axis, minor axis and axis ellipticity of the rotor motion trajectory.

[0035] Amplitude and phase synchronization demodulation is performed based on rotor angular displacement parameters to obtain displacement synchronization amplitude and displacement synchronization phase.

[0036] The first feature data is constructed from the major axis of the rotor motion trajectory, the minor axis of the rotor motion trajectory, the ellipticity of the rotor motion trajectory axis, the displacement synchronization amplitude, and the displacement synchronization phase.

[0037] Furthermore, the methods for obtaining the displacement synchronization amplitude and displacement synchronization phase include:

[0038] The displacement signals of the suspension bearing are mapped to time series and angle series to construct an angle domain displacement signal model;

[0039] From the angular domain displacement signal model, select the first sine reference signal and the first cosine reference signal with the same frequency as the rotor angular displacement parameter. Multiply the suspension bearing displacement signal with the first sine reference signal to obtain the first in-phase component signal. Multiply the suspension bearing displacement signal with the first cosine reference signal to obtain the first quadrature component signal.

[0040] The first in-phase component signal and the first quadrature component signal are filtered and the corresponding average energy components are retained to obtain the average value of the first in-phase component and the average value of the first quadrature component.

[0041] The displacement synchronization amplitude and displacement synchronization phase are calculated based on the average value of the first in-phase component and the average value of the first quadrature component.

[0042] Furthermore, the method for obtaining the second feature data includes:

[0043] The vibration acceleration signal of the machine body is subjected to trend removal and bandpass filtering to obtain the net acceleration waveform signal; based on the net acceleration waveform signal, a fast Fourier transform is performed, the energy distribution is calculated within a preset analysis frequency range and normalized to obtain the structural energy concentration.

[0044] Synchronous amplitude and phase demodulation is performed on the net acceleration waveform signal based on the rotor angular displacement parameters to obtain the fundamental frequency amplitude and phase of the acceleration.

[0045] The structural energy concentration, the fundamental frequency amplitude of acceleration, and the structural phase shift are used to construct the second feature data.

[0046] Furthermore, the method for obtaining the third feature data includes:

[0047] Bandpass filtering and harmonic suppression are performed on the electromagnetic response signal at the motor end to obtain the net electromagnetic response signal;

[0048] Based on the rotor angular displacement parameters, the electromagnetic net response signal is synchronously demodulated in amplitude and phase to extract the component consistent with the rotor rotation frequency, thereby obtaining the fundamental frequency amplitude and phase of the electromagnetic response.

[0049] Sideband demodulation is performed on the electromagnetic net response signal to extract the characteristic frequencies of the lower and upper sides of the fundamental frequency; the air gap sideband ratio is calculated based on the amplitudes of the lower and upper sides of the fundamental frequency extracted from the characteristic frequencies of the lower and upper sides of the fundamental frequency.

[0050] The fundamental frequency phase of the electromagnetic response is compared with the displacement synchronization phase of the suspension bearing displacement signal to obtain the magnetoelectric phase difference indication.

[0051] The fundamental frequency amplitude of the electromagnetic response, the air gap side band ratio, and the magnetoelectric phase difference indication are constructed as the third characteristic data.

[0052] Furthermore, the method for obtaining the fourth feature data includes:

[0053] Trend removal and bandpass filtering are performed on the magnetic levitation coil current signal to obtain the net current waveform signal;

[0054] Perform a fast Fourier transform on the net current waveform signal to obtain the amplitude function of the amplitude distribution of the net current waveform signal at each frequency component;

[0055] Based on the rotor rotation fundamental frequency and the preset bandwidth compensation ratio, a corresponding fundamental frequency analysis range is constructed, and energy integration is performed on the amplitude function within the fundamental frequency analysis range to obtain the fundamental frequency neighborhood energy.

[0056] The amplitude function's spectral range is constructed using its lower and upper spectral limits, and the total frequency domain energy is obtained by integrating the energy of the amplitude function within its spectral range.

[0057] The synchronization energy ratio is obtained by dividing the energy in the fundamental frequency neighborhood by the total energy in the frequency domain.

[0058] The net current waveform signal is harmonic decomposed according to the preset maximum harmonic order Q, Q harmonic amplitudes are extracted, and the current harmonic distortion coefficient is calculated.

[0059] The proportion of synchronous energy and the current harmonic distortion coefficient are used to construct the fourth characteristic data.

[0060] A dynamic balancing adjustment device for a magnetic levitation fan rotor, used to implement the aforementioned dynamic balancing adjustment method for a magnetic levitation fan rotor, comprising:

[0061] The multi-source vibration sensing module is used to collect a set of rotor operation signals, including the displacement signal of the suspension bearing, the current signal of the magnetic levitation coil, the vibration acceleration signal of the machine body, the electromagnetic response signal at the motor end, and the rotor angular displacement parameters.

[0062] The first data extraction module is used to preprocess and extract features from the displacement signal of the suspension bearing to obtain first feature data including the major axis of the rotor motion trajectory, the minor axis of the rotor motion trajectory, the ellipticity of the rotor motion trajectory axis, the displacement synchronization amplitude, and the displacement synchronization phase.

[0063] The second data processing module is used to preprocess and extract features from the body vibration acceleration signal to obtain second feature data including structural energy concentration, acceleration fundamental frequency amplitude and structural phase shift.

[0064] The third data processing module is used to preprocess and extract features from the electromagnetic response signal at the motor end to obtain third feature data including the fundamental frequency amplitude of the electromagnetic response, the air gap side band ratio and the magnetoelectric phase difference indication.

[0065] The fourth data processing module is used to preprocess and extract features from the magnetic levitation coil current signal to obtain fourth feature data including the proportion of synchronous energy and the current harmonic distortion coefficient.

[0066] The rotor balance diagnosis module is used to input comprehensive feature data, including first feature data, second feature data, third feature data and fourth feature data, into the rotor balance diagnosis model, identify the unbalanced vibration caused by rotor mass imbalance during operation, and generate rotor unbalanced state identifier.

[0067] The intelligent balance adjustment module solves the adjustment parameters based on the amplitude-reverse transformation algorithm and injects them into the magnetic levitation coil for adjustment. Based on real-time comprehensive characteristic data, it determines whether the adjustment parameters still need to be dynamically adjusted. If the determination result is yes, a dynamic gain coefficient is generated to correct the adjustment parameters before injecting them into the magnetic levitation coil. The adjustment parameters include the virtual counterweight amplitude and phase adjustment amount.

[0068] Compared with existing technologies, the magnetic levitation fan rotor dynamic balancing adjustment device and method proposed in this invention have the following technical effects and advantages:

[0069] This invention addresses the vibration problem caused by rotor mass imbalance during operation of existing magnetic levitation fans by proposing a comprehensive solution integrating multi-source sensing, intelligent identification, and adaptive adjustment. By establishing a multi-source vibration sensing module, real-time acquisition of displacement signals from the suspension bearing, current signals from the magnetic levitation coil, vibration acceleration signals from the machine body, and electromagnetic response signals from the motor end is achieved, forming comprehensive feature data covering the mechanical, electromagnetic, and control domains, providing a multi-dimensional information foundation for identifying the imbalance state. Furthermore, an amplitude-phase inversion algorithm is used to analyze the synchronous amplitude and phase characteristics of the rotor, deriving the amplitude and phase of the equivalent imbalance force, and generating virtual counterweight amplitude and phase adjustment amounts. By injecting reverse electromagnetic compensation force through the suspension coil, real-time suppression of rotor imbalance excitation is achieved, thus completing dynamic balance adjustment without shutting down the machine.

[0070] Furthermore, this invention calculates a balance diagnosis score through a rotor balance diagnosis model and realizes dynamic gain correction based on a score-gain coefficient matching table. When insufficient compensation is detected, the amplitude and phase of the adjustment parameters are automatically adjusted to form a closed-loop adaptive optimization mechanism.

[0071] In summary, the technical solution of the present invention can achieve rapid identification and self-learning correction of rotor imbalance vibration, significantly reduce vibration amplitude and energy consumption fluctuations, and improve the dynamic balance performance and long-term operational stability of magnetic levitation fans under complex operating conditions. Attached Figure Description

[0072] Figure 1 This is a schematic diagram of the dynamic balancing adjustment device for the magnetic levitation fan rotor according to Embodiment 1 of the present invention;

[0073] Figure 2 This is a flowchart of the dynamic balance adjustment method for the rotor of a magnetic levitation fan according to Embodiment 2 of the present invention;

[0074] Figure 3 The flowchart shows the method for solving the adjustment parameters based on the amplitude-inverse algorithm.

[0075] Figure 4 Here is a flowchart of the method for obtaining the first feature data;

[0076] Figure 5 The flowchart shows the method for obtaining the displacement synchronization amplitude and displacement synchronization phase. Detailed Implementation

[0077] The technical solutions of the embodiments of the present invention will be described in detail, clearly, and completely below with reference to the accompanying drawings. It should be particularly noted that the specific embodiments described below are only for better illustrating and explaining the technical solutions of the present invention, and are intended to enable those skilled in the art to better understand and implement the present invention, and should not be construed as limiting the scope of protection of the present invention. Without departing from the spirit and substance of the present invention, those skilled in the art can modify, adjust, or make equivalent substitutions based on the content disclosed in the present invention, and these should all be considered within the scope of protection of the present invention.

[0078] Example 1

[0079] Please see Figure 1 As shown, this embodiment discloses a dynamic balancing adjustment device for a magnetic levitation fan rotor, including a multi-source vibration sensing module, a first data extraction module, a second data extraction module, a third data extraction module, a fourth data extraction module, a rotor balance diagnosis module, and an intelligent balance adjustment module. Each module is connected via wired and / or wireless means to achieve data transmission.

[0080] The multi-source vibration sensing module is used to collect a set of rotor operation signals, including the displacement signal of the suspension bearing, the current signal of the magnetic levitation coil, the vibration acceleration signal of the machine body, the electromagnetic response signal of the motor end, and the rotor angular displacement parameters.

[0081] It should be noted that the suspension bearing displacement signal is measured in real time by a displacement sensor arranged within the bearing air gap. This sensor outputs a voltage signal proportional to the change in air gap. The suspension bearing displacement signal characterizes the rotor's displacement changes in the radial and axial directions, reflecting the spatial offset caused by unbalanced excitation and serving as a direct basis for judging the rotor's dynamic stability. The magnetic levitation coil current signal is measured in real time by a current detection unit built into the control system unit, reflecting the magnitude of the electromagnetic force output by the suspension coil to maintain rotor balance. When the rotor is unbalanced, the magnetic levitation coil current signal exhibits periodic fluctuations related to the rotational speed, which can be used to analyze the intensity and phase characteristics of unbalanced excitation.

[0082] The vibration acceleration signal of the fan body is measured by acceleration sensors installed on the fan body or supporting structure. This signal characterizes the acceleration response caused by rotor imbalance excitation transmitted through the structure. By monitoring the vibration acceleration signal, the overall vibration level of the rotor and the range of imbalance effects can be reflected. The electromagnetic response signal at the motor end is obtained by monitoring changes in voltage, current, or magnetic flux at the motor stator end. It characterizes the electromagnetic field disturbance caused by rotor eccentricity or air gap unevenness. The electromagnetic response signal at the motor end reflects the response characteristics of the electromagnetic system to the dynamic displacement of the rotor and is an important input for electromagnetic domain imbalance analysis.

[0083] It should be noted that the rotor angular displacement parameter is a physical quantity describing the instantaneous angular position of the rotor, representing the rotor's rotational angular position at any given moment. It can be considered as the angular coordinates of the rotor relative to the zero reference position during its rotation around the axis of rotation. When the rotor completes one full rotation, the rotor angle changes from 0° to 360°. This angular quantity allows for the establishment of a correspondence between the vibration signals of each channel and the actual mechanical angular position of the rotor. This enables amplitude-phase comparison during phase analysis using the same zero angular point as a reference, which is used for subsequent amplitude-phase synchronous demodulation, shaft center trajectory fitting, and unbalanced direction inversion.

[0084] The first data extraction module is used to preprocess and extract features from the displacement signal of the suspension bearing to obtain first feature data including the major axis of the rotor motion trajectory, the minor axis of the rotor motion trajectory, the ellipticity of the rotor motion trajectory axis, the displacement synchronization amplitude, and the displacement synchronization phase.

[0085] like Figure 4 As shown, the method for obtaining the first feature data includes:

[0086] The displacement signal of the suspension bearing is orthogonally decomposed and fitted with the axis trajectory to obtain the elliptic equation of the rotor motion trajectory; the elliptic equation of the rotor motion trajectory is solved to obtain the major axis, minor axis and axis ellipticity of the rotor motion trajectory.

[0087] Amplitude and phase synchronization demodulation is performed based on rotor angular displacement parameters to obtain displacement synchronization amplitude and displacement synchronization phase.

[0088] The first feature data is constructed from the major axis of the rotor motion trajectory, the minor axis of the rotor motion trajectory, the ellipticity of the rotor motion trajectory axis, the displacement synchronization amplitude, and the displacement synchronization phase.

[0089] The method for calculating the elliptical equation of the rotor motion trajectory includes:

[0090] A first planar coordinate system is established with the horizontal and vertical axes; the displacement signal of the suspension bearing is mapped to the first planar coordinate system to obtain the horizontal axis displacement signal and the vertical axis displacement signal;

[0091] Bandpass filtering and noise reduction are performed on the horizontal and vertical displacement signals to eliminate temperature drift and low-frequency drift components, while retaining the effective vibration components related to the rotor rotation frequency.

[0092] The horizontal axis displacement signal and the vertical axis displacement signal are orthogonally decomposed to obtain the projected displacement of the rotor in the horizontal axis direction and the vertical axis direction, which are denoted as the horizontal axis synchronous displacement and the vertical axis synchronous displacement respectively, and a set of synchronous displacement trajectory points corresponding to the first plane coordinate system is constructed.

[0093] The ellipse equation of the rotor motion trajectory is obtained by ellipse fitting of the synchronous displacement trajectory point set using the least squares method.

[0094] It should be noted that the major axis, minor axis, and ellipticity of the rotor trajectory can be obtained from the geometric parameters of the elliptic equation of the rotor trajectory.

[0095] The method for calculating the ellipticity of the rotor motion trajectory axis includes:

[0096] ;

[0097] in, The ellipticity of the rotor's trajectory axis. For the long axis of the rotor's motion trajectory, It is the short axis of the rotor's motion trajectory.

[0098] It should be noted that the magnitude of the ellipticity of the rotor's trajectory axis can intuitively reflect the degree of eccentricity and the direction of imbalance during rotor operation. When the rotor is affected by mass imbalance, its trajectory axis gradually evolves from an ideal circle to an eccentric ellipse, with the major axis of the ellipse corresponding to the principal direction of the unbalanced force. The greater the difference between the major and minor axes of the rotor's trajectory, the more pronounced the unbalanced excitation. Therefore, the major, minor, and ellipticity of the rotor's trajectory obtained based on orthogonal decomposition and trajectory fitting can not only be used to quantitatively assess the degree of dynamic imbalance of the rotor, but also provide a geometric basis for subsequent imbalance inversion and correction plane selection, thereby supporting the adaptive dynamic balance adjustment scheme of this invention.

[0099] like Figure 5 As shown, the methods for obtaining the displacement synchronization amplitude and displacement synchronization phase include:

[0100] The displacement signal of the suspension bearing is mapped to the time series and the angle series to construct the angle domain displacement signal model. Through this mapping process, the time non-uniformity of the displacement signal affected by the rotation speed fluctuation can be transformed into an angle synchronization signal that corresponds one-to-one with the mechanical rotation angle, thereby realizing the phase unification of signal sampling.

[0101] The first sine reference signal and the first cosine reference signal with the same frequency as the rotor angular displacement parameter are selected from the angular domain displacement signal model. The first in-phase component signal is obtained by multiplying the suspension bearing displacement signal with the first sine reference signal, and the first quadrature component signal is obtained by multiplying the suspension bearing displacement signal with the first cosine reference signal. The essence of this operation is to decompose the suspension bearing displacement signal into two parts in phase and orthogonal to the first sine reference signal and the first cosine reference signal in the rotating reference coordinate system, thereby achieving amplitude and phase separation of the signal.

[0102] The first in-phase component signal and the first quadrature component signal are filtered and the corresponding average energy components are retained to obtain the average value of the first in-phase component and the average value of the first quadrature component.

[0103] The displacement synchronization amplitude and displacement synchronization phase are calculated based on the average value of the first in-phase component and the average value of the first quadrature component.

[0104] The method for calculating the displacement synchronization amplitude includes:

[0105] ;

[0106] in, The displacement synchronization amplitude represents the effective vibration intensity of the rotor's radial displacement under a unified rotational phase reference. The average value of the first in-phase component. It is the average value of the first orthogonal component.

[0107] The method for calculating the displacement synchronization phase includes:

[0108] ;

[0109] in, The displacement synchronization phase represents the phase shift of the unbalanced excitation relative to the rotor rotation angle. It is the arctangent function.

[0110] The second data processing module is used to preprocess and extract features from the body vibration acceleration signal to obtain second feature data including structural energy concentration, acceleration fundamental frequency amplitude, and structural phase shift.

[0111] The method for obtaining the second feature data includes:

[0112] The vibration acceleration signal of the machine body is subjected to trend removal and bandpass filtering to eliminate structural noise and environmental interference, retaining the effective vibration components related to the fundamental frequency of rotor rotation, thus obtaining the net acceleration waveform signal. Based on the net acceleration waveform signal, a fast Fourier transform is performed to calculate the energy distribution within a preset analysis frequency range and perform normalization processing to obtain the structural energy concentration, which is used to characterize whether the structural response energy is concentrated in the principal order frequency range.

[0113] It should be noted that trend removal and bandpass filtering are signal processing techniques known to those skilled in the art. Their function is to eliminate DC drift, low-frequency temperature variations, and high-frequency noise components in the measured signal, thereby retaining the effective vibration components related to the rotor's rotational frequency. Specific implementation methods can employ various known algorithms, such as polynomial trend fitting elimination, moving average drift removal, or bandpass filtering methods based on fast Fourier transform. Since these processing methods are already well-disclosed in the fields of signal analysis and vibration detection, and their implementation principles are known to those skilled in the art, this invention will not elaborate further on them.

[0114] It should be further explained that the preset analysis frequency range is used to determine the frequency domain boundary for energy calculation and feature extraction of the net acceleration waveform signal. The setting principle is to make the analysis frequency band cover the rotor rotation frequency and the adjacent frequency bands where the main energy is concentrated, so as to fully reflect the structural response characteristics caused by mass imbalance excitation. For example, in this invention, the preset analysis frequency range can be set to [160Hz, 240Hz].

[0115] Synchronous amplitude and phase demodulation is performed on the net acceleration waveform signal based on the rotor angular displacement parameters to obtain the fundamental frequency amplitude and phase of the acceleration. By comparing the fundamental frequency phase of the acceleration with the synchronous phase of the displacement signal of the suspension bearing, the structural phase offset is calculated to reflect the additional phase delay characteristics of unbalanced excitation during the propagation process of the structure.

[0116] The structural energy concentration, fundamental frequency amplitude of acceleration, and structural phase shift are used to construct a second characteristic data set to describe the response characteristics of the structure under unbalanced excitation. This second characteristic data reflects the propagation concentration and phase lag characteristics of unbalanced energy along the structural path, providing a dynamic constraint basis for estimating unbalanced quantities and determining the correction plane.

[0117] Methods for obtaining the fundamental frequency amplitude and phase of acceleration include:

[0118] The net acceleration waveform signal is mapped to the time series and the angle series to construct the angle domain acceleration signal model. Through this mapping process, the acceleration signal can be converted from the time domain to the angle domain expression that corresponds one-to-one with the rotor angular position, thereby eliminating the time scale instability problem caused by speed fluctuation.

[0119] From the angular domain acceleration signal model, select a second sine reference signal and a second cosine reference signal with the same frequency as the rotor angular displacement parameter. Multiply the net acceleration waveform signal with the second sine reference signal to obtain the second in-phase component signal. Multiply the net acceleration waveform signal with the second cosine reference signal to obtain the second quadrature component signal.

[0120] The second in-phase component signal and the second quadrature component signal are filtered and the corresponding average energy components are retained to obtain the average value of the second in-phase component and the average value of the second quadrature component.

[0121] The fundamental frequency amplitude and phase of acceleration are calculated based on the average value of the second in-phase component and the average value of the second quadrature component.

[0122] The method for calculating the fundamental frequency amplitude of the acceleration includes:

[0123] ;

[0124] in, The fundamental frequency amplitude of the acceleration is used to represent the vibration intensity of the rotor structure under fundamental frequency excitation. The average value of the second in-phase component. It is the average value of the second orthogonal component.

[0125] The method for calculating the fundamental frequency phase of the acceleration includes:

[0126] ;

[0127] in, The fundamental frequency phase of the acceleration is used to describe the phase shift of the acceleration response relative to the rotor angular position.

[0128] The method for calculating the phase shift of the structure includes:

[0129] ;

[0130] in, This represents the structural phase shift.

[0131] It should be noted that when calculating the structural phase offset, the fundamental frequency phase of the acceleration and the displacement synchronization phase of the suspension bearing displacement signal are time-aligned and normalized to ensure that they have the same starting point and period in the angular domain. The structural phase offset reflects the degree of phase delay in the transmission of unbalanced excitation from the rotor to the machine structure. A larger value indicates a more significant hysteresis or additional damping effect of vibration energy in the structural propagation path. Analysis of the structural phase offset allows for the determination of the machine structure's response characteristics and the propagation law of unbalanced excitation, providing phase characteristic basis for unbalance inversion and structural path optimization.

[0132] The third data processing module is used to preprocess and extract features from the electromagnetic response signal at the motor end to obtain third feature data including the fundamental frequency amplitude of the electromagnetic response, the air gap side band ratio, and the magnetoelectric phase difference indication.

[0133] The method for obtaining the third feature data includes:

[0134] Bandpass filtering and harmonic suppression are performed on the electromagnetic response signal at the motor end to remove power supply ripple and high-frequency noise interference, and obtain a stable net electromagnetic response signal.

[0135] Based on the rotor angular displacement parameters, the electromagnetic net response signal is synchronously demodulated in amplitude and phase to extract the component consistent with the rotor rotation frequency, thereby obtaining the fundamental frequency amplitude and phase of the electromagnetic response.

[0136] Sideband demodulation is performed on the electromagnetic net response signal to extract the characteristic frequencies of the lower and upper sides of the fundamental frequency. Based on the amplitudes of the lower and upper sides of the fundamental frequency extracted from the characteristic frequencies of the lower and upper sides of the fundamental frequency, the air gap sideband ratio is calculated. The air gap sideband ratio is used to reflect the intensity of electromagnetic field disturbance caused by rotor eccentricity or air gap unevenness.

[0137] The fundamental frequency phase of the electromagnetic response is compared with the displacement synchronization phase of the suspension bearing displacement signal to obtain the magnetoelectric phase difference indicator. The magnetoelectric phase difference indicator is used to describe the phase shift characteristics of the electromagnetic excitation relative to the mechanical response, characterize the phase difference relationship between the electromagnetic force and the unbalanced force, and provide a dynamic reference for the phase adjustment of the electromagnetic virtual counterweight.

[0138] The fundamental frequency amplitude of the electromagnetic response, the air gap side band ratio, and the magnetoelectric phase difference indication are constructed as the third characteristic data. The third characteristic data is used to comprehensively characterize the electromagnetic disturbance characteristics caused by rotor imbalance and its energy propagation law in the magnetic levitation system.

[0139] The methods for calculating the fundamental frequency amplitude and phase of the electromagnetic response include:

[0140] The electromagnetic response signal at the motor end is mapped to the time series and the angle series to construct the electromagnetic response signal model in the angle domain. Through this mapping process, the time variation characteristics of the original electromagnetic signal can be transformed into phase response characteristics that are synchronized with the rotor rotation angle, providing a unified reference for subsequent amplitude and phase demodulation.

[0141] From the electromagnetic response signal model in the angular domain, select the third sine reference signal and the third cosine reference signal with the same frequency as the rotor angular displacement parameter. Multiply the net acceleration waveform signal with the third sine reference signal to obtain the third in-phase component signal. Multiply the net acceleration waveform signal with the third cosine reference signal to obtain the third quadrature component signal.

[0142] The third in-phase component signal and the third quadrature component signal are filtered and the corresponding average energy components are retained to obtain the average value of the third in-phase component and the average value of the third quadrature component.

[0143] The fundamental frequency amplitude and phase of the electromagnetic response are calculated based on the average value of the third in-phase component and the average value of the third quadrature component.

[0144] The method for calculating the fundamental frequency amplitude of the electromagnetic response includes:

[0145] ;

[0146] in, This is the fundamental frequency amplitude of the electromagnetic response, used to reflect the intensity of electromagnetic field disturbance at the fundamental frequency. The average value of the third in-phase component. It is the average value of the third orthogonal component.

[0147] The method for calculating the fundamental frequency phase of the electromagnetic response includes:

[0148] ;

[0149] in, The fundamental frequency phase of the electromagnetic response is used to reflect the phase distribution of the electromagnetic excitation relative to the rotor's angular position.

[0150] The method for calculating the air gap side band ratio includes:

[0151] ;

[0152] in, The air gap side band ratio, The amplitude of the lower band of the fundamental frequency. The amplitude of the upper band of the fundamental frequency. This represents the average amplitude of the lower band of the fundamental frequency and the upper band of the fundamental frequency. The main peak amplitude corresponds to the rotor's fundamental frequency. The rotor's fundamental frequency refers to the rotational frequency of the rotor under stable operating conditions. The main peak amplitude refers to the maximum amplitude point on the spectrum curve corresponding to the rotor's fundamental frequency after performing spectrum analysis on the electromagnetic net response signal. It reflects the main energy intensity of the signal at that frequency component.

[0153] The method for calculating the magnetoelectric phase difference indication includes:

[0154] ;

[0155] in, It is a magnetoelectric phase difference indicator.

[0156] The fourth data processing module is used to preprocess and extract features from the magnetic levitation coil current signal to obtain fourth feature data including the proportion of synchronous energy and the current harmonic distortion coefficient.

[0157] The method for obtaining the fourth feature data includes:

[0158] Trend removal and bandpass filtering are performed on the magnetic levitation coil current signal to eliminate low-frequency drift and high-frequency noise interference, and obtain a stable net current waveform signal.

[0159] Perform a fast Fourier transform on the net current waveform signal to obtain an amplitude function that describes the amplitude distribution of the net current waveform signal at each frequency component;

[0160] Based on the rotor's fundamental frequency and a preset bandwidth compensation ratio, a corresponding fundamental frequency analysis range is constructed. Energy integration is then performed on the amplitude function within this range to obtain the energy in the fundamental frequency neighborhood. The fundamental frequency analysis range is... ,in, The fundamental frequency of rotor rotation. For example, in this invention, the bandwidth compensation ratio is set to a range of [5%, 20%]. Preferably, the bandwidth compensation ratio can be set to 10%.

[0161] The amplitude function's spectral range is constructed using its lower and upper spectral limits, and the total frequency domain energy is obtained by integrating the energy of the amplitude function within its spectral range.

[0162] The synchronization energy ratio is obtained by dividing the energy in the fundamental frequency neighborhood by the total energy in the frequency domain. The synchronization energy ratio is used to measure the degree of concentration of coil current energy in the fundamental frequency synchronization component. The larger the value of the synchronization energy ratio, the more the system output current is mainly distributed in the energy region consistent with the rotor rotation frequency, and the more stable and effective the compensation effect is.

[0163] The net current waveform signal is harmonic decomposed according to the preset maximum harmonic order Q, Q harmonic amplitudes are extracted, and the current harmonic distortion coefficient is calculated. The current harmonic distortion coefficient is used to reflect the degree of nonlinear response of the coil current under unbalanced excitation.

[0164] The synchronous energy ratio and the current harmonic distortion coefficient are constructed as the fourth characteristic data, which are used to comprehensively characterize the energy concentration and control stability of the magnetic levitation actuator during the unbalance compensation process.

[0165] The method for calculating the fundamental frequency neighborhood energy includes:

[0166] ;

[0167] in, Represents the energy in the fundamental frequency neighborhood. Let df be the amplitude function, and df represent the frequency differential component.

[0168] The method for calculating the total energy in the frequency domain includes:

[0169] ;

[0170] in, The total energy in the frequency domain. This is the upper limit of the spectrum of the amplitude function. This is the lower limit of the spectrum of the amplitude function.

[0171] The method for calculating the current harmonic distortion coefficient includes:

[0172] ;

[0173] in, Here, is the current harmonic distortion coefficient, q is the index variable in the summation formula, and Q represents the maximum harmonic order. This represents the amplitude of the q-th harmonic. This represents the amplitude of the first harmonic, i.e., the fundamental frequency amplitude. The current harmonic distortion coefficient is used to characterize the nonlinear response characteristics of the coil current under rotor unbalanced excitation. A larger current harmonic distortion coefficient indicates a higher proportion of higher-order harmonic components in the current, suggesting that the control system generates stronger compensation waveform distortion to counteract the unbalanced force, reflecting a tendency for magnetic saturation or overcompensation in the magnetic levitation actuator. By monitoring changes in the current harmonic distortion coefficient, the system's ability to suppress unbalanced disturbances and its compensation stability can be assessed in real time, providing feedback for adaptive dynamic balance adjustment.

[0174] The rotor balance diagnosis module is used to input comprehensive feature data, including first feature data, second feature data, third feature data and fourth feature data, into the rotor balance diagnosis model, identify the unbalanced vibration caused by rotor mass imbalance during operation, and generate rotor unbalanced state identifier.

[0175] The method for generating the rotor imbalance state identifier includes:

[0176] The comprehensive feature data, including the first feature data, the second feature data, the third feature data, and the fourth feature data, is input into the rotor balance diagnosis model to obtain the first rotor balance diagnosis score.

[0177] If the first rotor balance diagnosis score is greater than or equal to the preset rotor balance score threshold, the rotor is determined to be in a balanced operating state, no rotor imbalance status identifier is generated, and the current diagnosis process ends.

[0178] If the first rotor balance diagnosis score is less than the preset rotor balance score threshold, then the rotor has unbalanced operation characteristics, a rotor unbalanced state identifier is generated, and the current diagnosis process ends.

[0179] It should be noted that the first rotor balance diagnostic score is used to characterize the rotor's operational stability. A higher score indicates smoother rotor operation and the absence of significant unbalanced vibrations; conversely, a lower score indicates more unstable rotor operation and the presence of significant unbalanced excitation. In an exemplary embodiment of the present invention, the first rotor balance diagnostic score ranges from [0, 100], and the rotor balance score threshold can be set to 80. A score higher than or equal to 80 indicates a balanced rotor state, while a score lower than 80 indicates an unbalanced rotor state.

[0180] The training method for the rotor balance diagnostic model includes:

[0181] A rotor balance diagnosis dataset is pre-constructed, comprising W sets of rotor balance diagnosis data and the first rotor balance diagnosis score corresponding to the W sets of rotor balance diagnosis data, where W is a positive integer; the rotor balance diagnosis data includes first feature data, second feature data, third feature data, and fourth feature data; the rotor balance diagnosis dataset is divided into a training set and a validation set, the training set is used for learning the parameters of the rotor balance diagnosis model, and the validation set is used for real-time monitoring of the generalization performance and overfitting degree of the rotor balance diagnosis model;

[0182] A deep neural network with a multilayer perceptron as its core is used as the rotor balance diagnosis model. The rotor balance diagnosis data is standardized and vectorized before being input into the deep neural network, which consists of an input layer, hidden layers, and an output layer. Each hidden layer extracts features using a nonlinear activation function, and the output layer uses a softmax activation function to obtain the probability distribution corresponding to each first rotor balance diagnosis score. Finally, the first rotor balance diagnosis score corresponding to the highest probability is taken as the prediction result of the rotor balance diagnosis model. During training, the cross-entropy loss function is used as the optimization objective, and a gradient descent-type optimization algorithm is used to update the network weights. An early stopping strategy is set: when the prediction accuracy on the validation set reaches or exceeds the prediction accuracy threshold, the rotor balance diagnosis model is considered to have converged and training is terminated. For example, the prediction accuracy threshold can be set to 95% in this application.

[0183] The intelligent balance adjustment module solves the adjustment parameters based on the amplitude-reverse transformation algorithm and injects them into the magnetic levitation coil for adjustment. Based on real-time comprehensive characteristic data, it determines whether the adjustment parameters still need to be dynamically adjusted. If the determination result is yes, a dynamic gain coefficient is generated to correct the adjustment parameters before injecting them into the magnetic levitation coil. The adjustment parameters include the virtual counterweight amplitude and phase adjustment amount.

[0184] It should be noted that when current is passed through the magnetic levitation coil, an electromagnetic field is generated. This electromagnetic field exerts an electromagnetic force on the rotor, which counteracts the rotor's own weight and unbalanced forces, thereby achieving levitation, stability, and active attitude adjustment. Specifically, the magnetic levitation coil acts directly on the rotor through electromagnetic force, enabling active control of the rotor's position, attitude, and dynamic imbalance. The adaptive adjustment process essentially involves intelligently adjusting the amplitude and phase of the driving current of the magnetic levitation coil, causing the electromagnetic force to form a compensating component in space opposite to the unbalanced force. The system performs secondary adjustment by calculating the amplitude and phase adjustment of the virtual counterweight in real time and introducing a dynamic gain coefficient, achieving adaptive dynamic balance and stable operation of the magnetic levitation fan rotor.

[0185] like Figure 3 As shown, the method for solving the adjustment parameters based on the amplitude-inverse transformation algorithm includes:

[0186] The displacement synchronization amplitude is converted into the unbalanced force amplitude based on the equivalent dynamic proportional coefficient corresponding to the entire magnetic levitation fan.

[0187] The phase of the unbalanced force is calculated based on the displacement synchronization phase and the structural phase offset.

[0188] An unbalanced force complex vector is constructed based on the unbalanced force amplitude and phase to simultaneously characterize the amplitude and direction information of unbalanced excitation.

[0189] Based on the principle of anti-phase superposition of unbalanced force complex vectors, a virtual counterweight amplitude is constructed so that the generated electromagnetic compensation force is equivalent to the unbalanced force in amplitude and acts in opposite phase.

[0190] The phase adjustment amount is calculated based on the unbalanced force phase and the magnetoelectric phase difference indication, and is used to correct the phase command of electromagnetic compensation.

[0191] The virtual counterweight amplitude and phase adjustment amount are constructed as adjustment parameters, and the adjustment parameters are injected into the magnetic levitation coil to realize real-time electromagnetic compensation and adaptive balance adjustment of rotor unbalance force.

[0192] The method for obtaining the amplitude of the unbalanced force includes:

[0193] ;

[0194] in, The magnitude of the unbalanced force. This represents the equivalent dynamic proportionality coefficient for the entire magnetic levitation fan. The unbalanced force amplitude is used to quantitatively describe the equivalent excitation intensity generated by the rotor during rotation due to mass eccentricity.

[0195] It should be noted that the term "magnetic levitation fan as a whole" refers to the entire magnetic levitation fan operating system, consisting of a rotor, magnetic levitation bearings, levitation coils, a drive motor, and a control device. Its dynamic characteristics are used to define the equivalent proportional relationship between the rotor displacement response and the applied unbalanced force. The equivalent dynamic proportionality coefficient characterizes the proportional relationship between the rotor displacement response and the unbalanced force of the entire magnetic levitation fan under fundamental frequency excitation conditions, and can be obtained through structural dynamics modeling, trial operation identification, or online adaptive identification. In this invention, the displacement synchronization amplitude is converted into the equivalent unbalanced force amplitude based on the equivalent dynamic proportionality coefficient, providing a force domain benchmark for subsequent amplitude inversion to solve the virtual counterweight adjustment parameters.

[0196] The method for obtaining the phase of the unbalanced force includes:

[0197] ;

[0198] in, For unbalanced force phase, For displacement synchronization phase, This refers to the structural phase offset. The unbalanced force phase is used to characterize the relative direction and time position of the unbalanced force within one rotation cycle, i.e., the phase delay of the excitation force relative to the rotor angular position, and is used to determine the direction of the compensating force. When the electromagnetic compensation phase is 180° out of phase with the unbalanced force phase, maximum vibration cancellation can be achieved.

[0199] The method for constructing the complex vector of the unbalanced force:

[0200] ;

[0201] in, Let be the complex vector of the unbalanced force, where e is a constant and j is the imaginary unit. The complex vector of the unbalanced force unifies the amplitude and phase of the unbalanced force into a single complex number representation, thus mathematically describing both the excitation intensity and direction information simultaneously. In this invention, this complex vectorization allows force superposition, reverse compensation, and multi-coil allocation to be completed within a unified complex plane, significantly improving the computational efficiency and stability of the compensation solution.

[0202] The method for constructing the virtual counterweight amplitude includes:

[0203] ;

[0204] in, For virtual counterweight amplitude, The vibration suppression weight coefficient is defined as [0, 1]. For example, in this invention, the vibration suppression weight coefficient can be set to 0.65.

[0205] The method for calculating the phase adjustment amount includes:

[0206] ;

[0207] in, For phase adjustment amount, It is a constant. It is a magnetoelectric phase difference indicator.

[0208] Methods for determining whether adjustment parameters still need dynamic adjustment based on real-time comprehensive characteristic data include:

[0209] The real-time comprehensive feature data is input into the rotor balance diagnosis model to obtain the second rotor balance diagnosis score;

[0210] If the second rotor balance diagnosis score is greater than or equal to the preset rotor balance score threshold, the adjusted rotor is determined to be in a balanced operating state, no dynamic adjustment operation is required, the judgment result is output as negative, and the current diagnosis process ends.

[0211] If the second rotor balance diagnosis score is less than the preset rotor balance score threshold, it is determined that the adjusted rotor still has unbalanced operation characteristics, dynamic adjustment operation needs to be performed, the judgment result is output as yes, and the current diagnosis process ends.

[0212] Methods for generating dynamic gain coefficients to correct adjustment parameters include:

[0213] The second rotor balance diagnostic score is matched with the pre-constructed score-gain coefficient matching table to obtain the corresponding dynamic gain coefficient; an example of the score-gain coefficient matching table is shown in Table 1.

[0214] Add the dynamic gain coefficient to the constant 1 to obtain the adjustment coefficient used for correction calculation;

[0215] Multiply the adjustment coefficient by the virtual weight amplitude and the phase adjustment amount respectively to obtain the corrected virtual weight amplitude and the corrected phase adjustment amount;

[0216] The corrected virtual counterweight amplitude and the corrected phase adjustment amount are used to construct the corrected adjustment parameters, and the adjustment parameters are injected into the magnetic levitation coil to form a new electromagnetic compensation force, thereby achieving secondary suppression and adaptive adjustment of rotor unbalance vibration.

[0217] Table 1. Rating-Gain Coefficient Matching Table

[0218]

[0219] Example 2

[0220] Please see Figure 2As shown, this embodiment provides a method for dynamic balance adjustment of a magnetic levitation fan rotor, including:

[0221] The collection includes rotor operation signal sets such as suspension bearing displacement signal, magnetic levitation coil current signal, body vibration acceleration signal, motor end electromagnetic response signal, and rotor angular displacement parameters;

[0222] Preprocessing and feature extraction of the suspension bearing displacement signal yields first feature data including the major axis of the rotor motion trajectory, the minor axis of the rotor motion trajectory, the ellipticity of the rotor motion trajectory axis, the displacement synchronization amplitude, and the displacement synchronization phase.

[0223] Preprocessing and feature extraction are performed on the vibration acceleration signal of the body to obtain second feature data including structural energy concentration, fundamental frequency amplitude of acceleration and structural phase shift.

[0224] The electromagnetic response signal at the motor end is preprocessed and its features are extracted to obtain the third feature data, including the fundamental frequency amplitude of the electromagnetic response, the air gap side band ratio, and the magnetoelectric phase difference indication.

[0225] Preprocessing and feature extraction are performed on the magnetic levitation coil current signal to obtain fourth feature data including the proportion of synchronization energy and the current harmonic distortion coefficient;

[0226] The comprehensive feature data, including the first feature data, the second feature data, the third feature data, and the fourth feature data, is input into the rotor balance diagnosis model to identify the unbalanced vibration caused by rotor mass imbalance during operation and generate a rotor unbalanced state identifier.

[0227] The adjustment parameters are solved based on the amplitude-inverse transformation algorithm and injected into the magnetic levitation coil for adjustment. Based on the real-time comprehensive characteristic data, it is determined whether the adjustment parameters still need to be dynamically adjusted. If the determination result is yes, a dynamic gain coefficient is generated to correct the adjustment parameters before injecting them into the magnetic levitation coil. The adjustment parameters include the virtual counterweight amplitude and the phase adjustment amount.

[0228] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of protection of the present invention.

[0229] In conclusion, 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 spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method of dynamic balancing adjustment of a magnetic levitation fan rotor, characterized by, The method comprises the following steps: collecting a rotor operation signal set comprising a suspension bearing displacement signal, a magnetic suspension coil current signal, a machine body vibration acceleration signal, a motor end electromagnetic response signal and a rotor angular displacement parameter; preprocessing and feature extraction are performed on the suspension bearing displacement signal to obtain first feature data comprising a rotor motion trajectory long axis, a rotor motion trajectory short axis, a rotor motion trajectory axis eccentricity, a displacement synchronous amplitude and a displacement synchronous phase; preprocessing and feature extraction are performed on the machine body vibration acceleration signal to obtain second feature data comprising a structure energy concentration degree, an acceleration fundamental frequency amplitude and a structure phase shift; preprocessing and feature extraction are performed on the motor end electromagnetic response signal to obtain third feature data comprising an electromagnetic response fundamental frequency amplitude, an air gap sideband ratio and a magneto-electric phase difference indicator; preprocessing and feature extraction are performed on the magnetic suspension coil current signal to obtain fourth feature data comprising a synchronous energy proportion and a current harmonic distortion coefficient; the comprehensive feature data comprising the first feature data, the second feature data, the third feature data and the fourth feature data are input into a rotor balance diagnosis model to identify unbalanced vibration caused by rotor mass imbalance during operation and generate a rotor unbalance state identifier; adjustment parameters are solved based on an amplitude-phase inverse algorithm and are injected into the magnetic suspension coil for adjustment; whether the adjustment parameters still need to be dynamically adjusted is determined according to real-time comprehensive feature data, and if the determination result is yes, a dynamic gain coefficient is generated to modify the adjustment parameters and then the modified adjustment parameters are injected into the magnetic suspension coil; the adjustment parameters comprise a virtual counterweight amplitude and a phase adjustment amount; the method for solving the adjustment parameters based on the amplitude-phase inverse algorithm comprises the following steps: the displacement synchronous amplitude is converted into an unbalanced force amplitude based on an equivalent dynamic proportional coefficient corresponding to the whole magnetic suspension fan; the unbalanced force phase is calculated based on the displacement synchronous phase and the structure phase shift; the unbalanced force complex vector is constructed based on the unbalanced force amplitude and the unbalanced force phase; the virtual counterweight amplitude is constructed according to the unbalanced force complex vector according to the principle of inverse superposition; the phase adjustment amount is calculated based on the unbalanced force phase and the magneto-electric phase difference indicator; and the virtual counterweight amplitude and the phase adjustment amount are constructed into the adjustment parameters.

2. The magnetic levitation fan rotor dynamic balance adjustment method according to claim 1, characterized by, the method for determining whether the adjustment parameters still need to be dynamically adjusted according to real-time comprehensive feature data comprises the following steps: the real-time comprehensive feature data are input into the rotor balance diagnosis model to obtain a second rotor balance diagnosis score; if the second rotor balance diagnosis score is greater than or equal to a preset rotor balance score threshold, a determination result of no is output and the current diagnosis process is ended; if the second rotor balance diagnosis score is less than the preset rotor balance score threshold, a determination result of yes is output and the current diagnosis process is ended.

3. The magnetic levitation fan rotor dynamic balance adjustment method according to claim 2, characterized by, the method for generating a dynamic gain coefficient to modify the adjustment parameters comprises the following steps: the second rotor balance diagnosis score is matched with a pre-constructed score-gain coefficient matching table to obtain a corresponding dynamic gain coefficient; the dynamic gain coefficient is added to a constant 1 to obtain an adjustment coefficient used for modification calculation; the adjustment coefficient is multiplied by the virtual counterweight amplitude and the phase adjustment amount respectively to obtain a modified virtual counterweight amplitude and a modified phase adjustment amount. The corrected virtual counterweight amplitude and the corrected phase adjustment amount are constructed as a corrected adjustment parameter.

4. The magnetic levitation fan rotor dynamic balance adjustment method of claim 1, wherein, The method for generating the rotor unbalance state identifier comprises: comprehensive feature data including first feature data, second feature data, third feature data and fourth feature data is input into a rotor balance diagnosis model to obtain a first rotor balance diagnosis score; if the first rotor balance diagnosis score is greater than or equal to a preset rotor balance score threshold, no rotor unbalance state identifier is generated, and the current diagnosis process is ended; if the first rotor balance diagnosis score is less than the preset rotor balance score threshold, a rotor unbalance state identifier is generated, and the current diagnosis process is ended.

5. The magnetic levitation fan rotor dynamic balance adjustment method of claim 1, wherein, The method for obtaining the first feature data comprises: orthogonal decomposition and shaft orbit fitting are performed on the suspension bearing displacement signal to obtain a rotor motion orbit ellipse equation; the rotor motion orbit ellipse equation is solved to obtain a rotor motion orbit major axis, a rotor motion orbit minor axis and a rotor motion orbit axis center ellipticity; synchronous amplitude and phase demodulation is performed based on the rotor angular displacement parameter to obtain a displacement synchronous amplitude and a displacement synchronous phase; the rotor motion orbit major axis, the rotor motion orbit minor axis, the rotor motion orbit axis center ellipticity, the displacement synchronous amplitude and the displacement synchronous phase are constructed into the first feature data.

6. The magnetic levitation fan rotor dynamic balancing adjustment method of claim 5, wherein, The method for obtaining the displacement synchronous amplitude and the displacement synchronous phase comprises: the suspension bearing displacement signal is correspondingly mapped according to a time sequence and an angle sequence to construct an angle domain displacement signal model; first sine reference signals and first cosine reference signals of the same frequency as the rotor angular displacement parameter are selected from the angle domain displacement signal model, the suspension bearing displacement signal is multiplied by the first sine reference signals to obtain first in-phase component signals, and the suspension bearing displacement signal is multiplied by the first cosine reference signals to obtain first quadrature component signals; the first in-phase component signals and the first quadrature component signals are filtered, and corresponding average energy components are retained to obtain a first in-phase component average value and a first quadrature component average value; the displacement synchronous amplitude and the displacement synchronous phase are calculated based on the first in-phase component average value and the first quadrature component average value.

7. The magnetic levitation fan rotor dynamic balance adjustment method of claim 1, wherein, The method for obtaining the second feature data comprises: trend removal and band-pass filtering are performed on the machine body vibration acceleration signal to obtain an acceleration net waveform signal; fast Fourier transform is performed on the acceleration net waveform signal, energy distribution is calculated in a preset analysis frequency range, and normalization processing is performed to obtain a structure energy concentration degree; synchronous amplitude and phase demodulation is performed on the acceleration net waveform signal based on the rotor angular displacement parameter to obtain an acceleration fundamental frequency amplitude and an acceleration fundamental frequency phase; the structure energy concentration degree, the acceleration fundamental frequency amplitude and the structure phase offset are constructed into the second feature data.

8. The magnetic levitation fan rotor dynamic balance adjustment method of claim 1, wherein, The method for obtaining the third feature data comprises: band-pass filtering and harmonic suppression are performed on the motor end electromagnetic response signal to obtain an electromagnetic net response signal; synchronous amplitude and phase demodulation is performed on the electromagnetic net response signal based on the rotor angular displacement parameter to extract a component consistent with the rotor rotation frequency, and an electromagnetic response fundamental frequency amplitude and an electromagnetic response fundamental frequency phase are obtained. The sideband demodulation is performed on the electromagnetic net response signal to extract a lower sideband characteristic frequency and an upper sideband characteristic frequency of the base frequency; the air gap sideband ratio is calculated according to a lower sideband amplitude extracted from the lower sideband characteristic frequency and an upper sideband amplitude extracted from the upper sideband characteristic frequency; The electromagnetic response base frequency phase is compared with the displacement synchronous phase of the suspension bearing displacement signal to obtain a magneto-electric phase difference indicator; The electromagnetic response base frequency amplitude, the air gap sideband ratio and the magneto-electric phase difference indicator are constructed as third feature data.

9. The magnetic levitation fan rotor dynamic balancing adjustment method of claim 1, wherein, The fourth feature data acquisition method comprises: Performing trend removal and band-pass filtering on the magnetic suspension coil current signal to obtain a current net waveform signal; Performing fast Fourier transform on the current net waveform signal to obtain an amplitude function of the amplitude distribution of the current net waveform signal at each frequency component; Based on the rotor rotation base frequency and a preset bandwidth compensation ratio, a corresponding base frequency analysis range is constructed, and energy integration is performed on the amplitude function in the base frequency analysis range to obtain a base frequency neighborhood energy; The frequency spectrum range of the amplitude function is constructed with the lower limit and the upper limit of the frequency spectrum of the amplitude function, and energy integration is performed on the amplitude function in the amplitude function frequency spectrum range to obtain a total energy in the frequency domain; The synchronous energy proportion is obtained by dividing the base frequency neighborhood energy by the total energy in the frequency domain; The current harmonic distortion coefficient is calculated by performing harmonic decomposition on the current net waveform signal according to a preset maximum harmonic number Q and extracting Q harmonic amplitudes; The synchronous energy proportion and the current harmonic distortion coefficient are constructed as the fourth feature data.

10. A magnetic levitation fan rotor dynamic balance adjustment apparatus for implementing the magnetic levitation fan rotor dynamic balance adjustment method according to any one of claims 1 to 9, characterized by, Comprise: A multi-source vibration perception module is configured to collect a rotor operation signal set including a suspension bearing displacement signal, a magnetic suspension coil current signal, a machine body vibration acceleration signal, a motor end electromagnetic response signal and a rotor angular displacement parameter; A first data extraction module is configured to preprocess and extract features of the suspension bearing displacement signal to obtain first feature data including a rotor motion trajectory major axis, a rotor motion trajectory minor axis, a rotor motion trajectory axis eccentricity, a displacement synchronous amplitude and a displacement synchronous phase; A second data processing module is configured to preprocess and extract features of the machine body vibration acceleration signal to obtain second feature data including a structure energy concentration degree, an acceleration base frequency amplitude and a structure phase shift; A third data processing module is configured to preprocess and extract features of the motor end electromagnetic response signal to obtain third feature data including an electromagnetic response base frequency amplitude, an air gap sideband ratio and a magneto-electric phase difference indicator; A fourth data processing module is configured to preprocess and extract features of the magnetic suspension coil current signal to obtain fourth feature data including a synchronous energy proportion and a current harmonic distortion coefficient; A rotor balance diagnosis module is configured to input comprehensive feature data including the first feature data, the second feature data, the third feature data and the fourth feature data into a rotor balance diagnosis model, identify unbalanced vibrations caused by rotor mass imbalance during operation, and generate a rotor unbalance state identifier. The intelligent balance adjustment module is based on amplitude-phase inverse algorithm to solve adjustment parameters and injects the adjustment parameters into the magnetic suspension coil for adjustment; whether the adjustment parameters still need dynamic adjustment is determined according to real-time comprehensive characteristic data, if the determination result is yes, a dynamic gain coefficient is generated to modify the adjustment parameters and then inject the adjustment parameters into the magnetic suspension coil; the adjustment parameters include virtual counterweight amplitude and phase adjustment amount.

Images