Method for monitoring the operating state of a drive motor

Abstract

The present application relates to the technical field of mechanical vibration measurement, and discloses a vibration monitoring method for driving motor operation state, comprising: a sensor acquires shell surface vibration acceleration waveform; an energy operator transformation is applied to separate high-order carrier components from the waveform to determine a transient time-varying frequency line, and a rotor tooth slot number is combined to calculate a transient rotating frequency; an angular acceleration trend is determined according to a historical period rotating frequency difference value to update an adaptive filter center frequency prediction rule, and a cutoff boundary is controlled to follow a frequency shift trajectory to move; a torsional vibration deviation amount is obtained by implementing a second-order discrete difference calculation on the rotating frequency, and a sampling interval is fine-tuned to reconstruct an angular domain discrete stationary signal, a self-demodulation path homologous to a physical measuring point is constructed, transmission lag caused by shafting torsion is avoided, the reconstructed signal has a boundary convergence characteristic, and an order spectrum line energy divergence defect is eliminated at the source.

Description

Technical Field

[0001] This invention relates to a vibration monitoring method for the operating status of drive motors, belonging to the field of mechanical vibration measurement technology. Background Technology

[0002] Currently, under non-stationary operating conditions, the mainstream approach to identifying shaft imbalance, rotor misalignment, and localized bearing damage involves acquiring acceleration waveforms using vibration sensors deployed on the housing surface and converting them from the time domain to the angular domain to generate a fault order spectrum. The physical basis of angular domain analysis lies in matching the actual rotational phase of the shaft using a discrete point sequence, thereby converting the characteristic components that generate continuous frequency shifts into fixed order components and extracting the amplitude of characteristic spectral lines within a specific rotational period. However, under the impact of large inertia loads and variable speed conditions, the sudden torque changes in the motor generate transient physical slip and flexible torsional vibration on the shaft stiffness transmission chain. Fluctuations cause an unavoidable spatial stiffness span and phase propagation delay between the transient rotational speed phase recorded by the outer speed encoder and the dynamic phase at the vibration sensor measurement point. At the same time, the electromagnetic velocity parameters output by the control layer cannot track the transient torsional deformation of the structural components due to the lack of perception of physical slip yield. Conventional linear paths such as increasing the number of encoder lines and increasing the filter order cannot overcome the physical mismatch in stiffness space. Instead, the inherent group delay of the filter exacerbates the cumulative phase distortion of the time-domain signal resampling, resulting in energy divergence and signal aliasing of the order spectrum under high dynamic conditions.

[0003] Besides the physical limitations of external encoders in terms of spatial stiffness mismatch and phase propagation delay, relying on software algorithms to implement keyless phase order tracking also has bottlenecks. For example, Chinese invention patent application CN113125179A discloses a keyless phase order tracking method for rotating machinery speed fluctuations. It performs short-time Fourier transform on the variable speed vibration signal, uses variational nonlinear frequency modulation mode decomposition to extract the dynamic features of the signal and reconstruct the instantaneous phase, thus getting rid of the hardware keyer installation constraints. However, in industrial sites where drive motors are subjected to high-contrast variable load impacts, the pure data-driven mathematical iterative optimization algorithm lacks deterministic physical and dynamic boundary rigidity constraints. When faced with high peak sudden noise and drastic slip fluctuations between the stator and rotor, the objective function is prone to getting trapped in local optima or the iteration does not converge. Moreover, the technical path of relying on time-frequency ridge numerical demodulation to approximate the instantaneous frequency has group delay and high computational delay under non-stationary electromagnetic interference, and cannot dynamically capture the weak torsional vibration at the source of the shell measuring point, resulting in phase divergence and cumulative drift in the reconstructed equiangular resampling grid.

[0004] Therefore, the technical problem to be solved by this invention is how to directly extract the transient phase evolution process of the rotor that is completely physically homologous to the vibration sensor measurement point and has no transmission delay under a simpler sensor framework, and to establish an angle domain resampling grid sequence with monotonic deterministic correlation under variable noise interference conditions to eliminate process phase distortion. Summary of the Invention

[0005] To address the problems in the background art, the technical solution of the present invention is as follows: a vibration monitoring method for the operating status of a drive motor, comprising the following steps: Step S101: The continuous mechanical vibration acceleration waveform data of the drive motor housing surface is acquired by a high-frequency vibration acceleration sensor; Step S102: Apply energy operator transformation with a specific amplitude gating threshold to separate the high-order vibration carrier component containing the stator slot passage frequency component from the continuous mechanical vibration acceleration waveform data of the drive motor housing surface, determine its corresponding transient time-varying frequency line, and calculate the transient shaft rotation frequency based on the algebraic division relationship between the transient time-varying frequency line and the inherent rotor slot number constant, thereby establishing a self-demodulation path with adaptive demodulation capability for shaft slip fluctuations; Step S103: Determine the angular acceleration trend of the current period based on the difference in the rotational frequency of the shaft measured in the two adjacent historical periods in the previous period, and calculate the center frequency prediction update increment based on the angular acceleration trend to update the center frequency prediction update rule of the adaptive digital bandpass filter. Control the cutoff boundary of the adaptive digital bandpass filter to move in real time following the transient frequency shift trajectory of the higher-order vibration carrier component to block the zero-point pseudo-triggering caused by sudden changes in rotational speed and torque. Step S104: Perform second-order discrete-time difference calculation on the transient shaft rotation frequency to obtain the transient torsional vibration deviation of the shaft, which characterizes the dynamic flutter intensity of the shaft. Based on the transient torsional vibration deviation of the shaft, add a step fine-tuning correction amount on the nominal sampling interval to implement step fine-tuning, thereby dynamically adjusting the discrete sampling time of each sampling point and reconstructing the angle domain discrete stationary signal that is rigidly aligned with the spatial rotation angle of the shaft.

[0006] Preferably, after step S101, the method further includes the following: applying a time-series hysteresis damping operator to configure positive dead zone boundaries and negative dead zone boundaries on both sides of the zero-crossing point axis of the amplitude of the continuous mechanical vibration acceleration waveform data on the surface of the drive motor housing, and truncating the continuous mechanical vibration acceleration waveform data on the surface of the drive motor housing at zero points through the positive dead zone boundaries and negative dead zone boundaries, wherein the absolute values ​​of the positive dead zone boundaries and the negative dead zone boundaries are equal, and the magnitude of the absolute values ​​increases with the increase of the root mean square value of the background ambient noise.

[0007] Preferably, in step S102, separating the high-order vibration carrier component containing the stator slot passage frequency component from the continuous mechanical vibration acceleration waveform data of the drive motor housing surface to determine its corresponding transient time-varying frequency line includes the following sub-steps: Step S1021, capturing non-uniform time discrete point signals containing the intrinsic transient geometric phase evolution history of the shaft system by a high-frequency dynamic vibration sensing link; Step S1022, extracting the time-frequency ridge of the non-uniform time discrete point signals by an energy operator to separate the high-order vibration carrier component containing the stator slot passage frequency component, and converting the high-order vibration carrier component containing the stator slot passage frequency component into a transient time-varying frequency line.

[0008] Preferably, in step S103, updating the center frequency prediction update rule of the adaptive digital bandpass filter includes the following sub-steps: Step S1031, calculating the transient rotational frequency difference between adjacent historical periods to obtain the real-time angular acceleration trend of the mechanical shaft; Step S1032, linearly extrapolating the transient frequency shift of the current period based on the real-time angular acceleration trend, and calculating the center frequency prediction update increment to adjust the cutoff boundary of the adaptive digital bandpass filter, so that the passband of the adaptive digital bandpass filter moves in real time following the transient frequency shift trajectory of the higher-order vibration carrier component containing the stator slot passage frequency component.

[0009] Preferably, in step S104, the step-by-step fine-tuning based on the nominal sampling interval by superimposing a step-by-step correction amount on the basis of the transient torsional vibration deviation of the shaft includes the following sub-steps: Step S1041, applying the second-order difference operator to differentiate the transient shaft rotation frequency twice in consecutive time, and calculating the transient torsional vibration acceleration value characterizing the transient torsional vibration deviation of the shaft; Step S1042, comparing the transient torsional vibration acceleration value with the determined safety threshold range, if the transient torsional vibration acceleration value exceeds the safety threshold range, then adjusting the step-by-step fine-tuning correction amount according to a fixed step step until the generated angular domain discrete stationary signal satisfies the boundary convergence condition.

[0010] Preferably, after reconstructing the angle domain discrete stationary signal that is rigidly aligned with the rotation angle of the shaft space in step S104, the following steps are also included: Step S105, importing the angle domain discrete stationary signal into the characteristic order spectrum analysis engine, and performing Fourier transform on the angle domain discrete stationary signal to calculate the characteristic order spectrum of the vibration of the corresponding drive motor mechanical structure.

[0011] Preferably, after step S105, the method further includes the following: step S106, extracting the amplitude of the mechanical dynamic fault order features caused by shaft imbalance or shaft misalignment from the characteristic order spectrum lines, and evaluating the structural health of the drive motor based on the amplitude of the fault order features.

[0012] Preferably, in step S106, assessing the structural health of the drive motor based on the amplitude of the fault order feature includes the following sub-steps: step S1061, comparing the amplitude of the extracted fault order feature with a fixed degradation alarm threshold; step S1062, when the amplitude of the fault order feature continuously exceeds the degradation alarm threshold, outputting a fault warning signal indicating that there is a risk of unexpected shutdown in the mechanical structure of the drive motor.

[0013] Preferably, in step S104, during the period when the drive motor is subjected to impact load causing transient mechanical slip fluctuations in the shaft system, the digital signal processor controls the phase deviation of the discrete stationary signal in the angle domain to be maintained within a safe deviation range, so that the discrete stationary signal in the angle domain remains in a boundary convergence state.

[0014] Compared with the prior art, the beneficial effects of the present invention are: 1. In vibration monitoring of motor operation, by discretely extracting high-order vibration carrier components containing rotor tooth passage harmonics from single-channel structural vibration acceleration signals, and combining this with the zero-crossing point capture operator to locate the intersection time series of positive and negative polarity switching, the continuous transient geometric phase angle grid at the physical measurement point is directly reconstructed. This ensures that the sampling grid and the vibration characteristic source of the mechanical structure are in the same physical measurement point and a unified time reference, solving the stiffness span between the mechanical mounting point of the traditional external encoder and the vibration measurement point of the housing, avoiding the transmission lag caused by the flexible torsion of the shaft system in physical space, and enabling the non-uniform angle domain sampling grid sequence to maintain high-fidelity spatiotemporal mapping under continuous speed fluctuation conditions.

[0015] 2. By using energy operators to transform and process the carrier components to obtain transient time-varying frequency lines that are monotonically modulated with rotational speed, and further combining the inherent rotor cogging number constant to calculate the transient shaft rotation frequency, a self-demodulation path that is completely controlled by the laws of mechanical dynamics is constructed. This not only eliminates the dependence of traditional designs on the electromagnetic speed parameters of the inverter control layer, but also avoids the angle sampling deviation caused by the conventional electromagnetic speed estimation ignoring rotor slip and torsional vibration fluctuations. When the system is subjected to impact loads that cause severe fluctuations in the transient mechanical slip of the shaft system, the generated discrete stationary signal in the angle domain still has rigid boundary convergence characteristics, eliminating the energy divergence defect of the order spectral line at the mechanism level.

[0016] 3. By utilizing the angular acceleration trend measured in the previous historical period to construct the discretized prediction update rule of the filter center frequency, the cutoff boundary of the adaptive digital bandpass filter can dynamically wrap the transient frequency shift trajectory of the higher-order vibration carrier component. This effectively overcomes the problem of filter phase step response lag caused by high acceleration under emergency start-stop or impact load of the drive motor, and blocks the risk of zero-point false triggering and spectral line loss caused by sudden speed change. In conjunction with the second-order differential operator to calculate the logical rack bounce amount obtained by the shaft rotation frequency transformation, the nominal sampling interval is finely adjusted in a step-by-step manner to ensure that the order spectral lines maintain high resolution and clarity under high dynamic conditions. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the processing logic of the vibration monitoring method for the operating state of a drive motor according to the present invention. Figure 2 This is a module topology diagram of the vibration monitoring method for the driving motor operating status of the present invention.

[0018] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0020] A vibration monitoring method for drive motor operating status includes the following steps: Step S101: The continuous mechanical vibration acceleration waveform data of the drive motor housing surface is acquired by a high-frequency vibration acceleration sensor; Step S102: Apply energy operator transformation with a specific amplitude gating threshold to separate the high-order vibration carrier component containing the stator slot passage frequency component from the continuous mechanical vibration acceleration waveform data of the drive motor housing surface, determine its corresponding transient time-varying frequency line, and calculate the transient shaft rotation frequency based on the algebraic division relationship between the transient time-varying frequency line and the inherent rotor slot number constant, thereby establishing a self-demodulation path with adaptive demodulation capability for shaft slip fluctuations; Step S103: Determine the angular acceleration trend of the current period based on the difference in the rotational frequency of the shaft measured in the two adjacent historical periods in the previous period, and calculate the center frequency prediction update increment based on the angular acceleration trend to update the center frequency prediction update rule of the adaptive digital bandpass filter. Control the cutoff boundary of the adaptive digital bandpass filter to move in real time following the transient frequency shift trajectory of the higher-order vibration carrier component to block the zero-point pseudo-triggering caused by sudden changes in rotational speed and torque. Step S104: Perform second-order discrete-time difference calculation on the transient shaft rotation frequency to obtain the transient torsional vibration deviation of the shaft, which characterizes the dynamic flutter intensity of the shaft. Based on the transient torsional vibration deviation of the shaft, add a step fine-tuning correction amount on the nominal sampling interval to implement step fine-tuning, thereby dynamically adjusting the discrete sampling time of each sampling point and reconstructing the angle domain discrete stationary signal that is rigidly aligned with the spatial rotation angle of the shaft.

[0021] Preferably, after step S101, the method further includes the following: applying a time-series hysteresis damping operator to configure positive dead zone boundaries and negative dead zone boundaries on both sides of the zero-crossing point axis of the amplitude of the continuous mechanical vibration acceleration waveform data on the surface of the drive motor housing, and truncating the continuous mechanical vibration acceleration waveform data on the surface of the drive motor housing at zero points through the positive dead zone boundaries and negative dead zone boundaries, wherein the absolute values ​​of the positive dead zone boundaries and the negative dead zone boundaries are equal, and the magnitude of the absolute values ​​increases with the increase of the root mean square value of the background ambient noise.

[0022] Preferably, in step S102, separating the high-order vibration carrier component containing the stator slot passage frequency component from the continuous mechanical vibration acceleration waveform data of the drive motor housing surface to determine its corresponding transient time-varying frequency line includes the following sub-steps: Step S1021, capturing non-uniform time discrete point signals containing the intrinsic transient geometric phase evolution history of the shaft system by a high-frequency dynamic vibration sensing link; Step S1022, extracting the time-frequency ridge of the non-uniform time discrete point signals by an energy operator to separate the high-order vibration carrier component containing the stator slot passage frequency component, and converting the high-order vibration carrier component containing the stator slot passage frequency component into a transient time-varying frequency line.

[0023] Preferably, in step S103, updating the center frequency prediction update rule of the adaptive digital bandpass filter includes the following sub-steps: Step S1031, calculating the transient rotational frequency difference between adjacent historical periods to obtain the real-time angular acceleration trend of the mechanical shaft; Step S1032, linearly extrapolating the transient frequency shift of the current period based on the real-time angular acceleration trend, and calculating the center frequency prediction update increment to adjust the cutoff boundary of the adaptive digital bandpass filter, so that the passband of the adaptive digital bandpass filter moves in real time following the transient frequency shift trajectory of the higher-order vibration carrier component containing the stator slot passage frequency component.

[0024] Preferably, in step S104, the step-by-step fine-tuning based on the nominal sampling interval by superimposing a step-by-step correction amount on the basis of the transient torsional vibration deviation of the shaft includes the following sub-steps: Step S1041, applying the second-order difference operator to differentiate the transient shaft rotation frequency twice in consecutive time, and calculating the transient torsional vibration acceleration value characterizing the transient torsional vibration deviation of the shaft; Step S1042, comparing the transient torsional vibration acceleration value with the determined safety threshold range, if the transient torsional vibration acceleration value exceeds the safety threshold range, then adjusting the step-by-step fine-tuning correction amount according to a fixed step step until the generated angular domain discrete stationary signal satisfies the boundary convergence condition.

[0025] Preferably, after reconstructing the angle domain discrete stationary signal that is rigidly aligned with the rotation angle of the shaft space in step S104, the following steps are also included: Step S105, importing the angle domain discrete stationary signal into the characteristic order spectrum analysis engine, and performing Fourier transform on the angle domain discrete stationary signal to calculate the characteristic order spectrum of the vibration of the corresponding drive motor mechanical structure.

[0026] Preferably, after step S105, the method further includes the following: step S106, extracting the amplitude of the mechanical dynamic fault order features caused by shaft imbalance or shaft misalignment from the characteristic order spectrum lines, and evaluating the structural health of the drive motor based on the amplitude of the fault order features.

[0027] Preferably, in step S106, assessing the structural health of the drive motor based on the amplitude of the fault order feature includes the following sub-steps: step S1061, comparing the amplitude of the extracted fault order feature with a fixed degradation alarm threshold; step S1062, when the amplitude of the fault order feature continuously exceeds the degradation alarm threshold, outputting a fault warning signal indicating that there is a risk of unexpected shutdown in the mechanical structure of the drive motor.

[0028] Preferably, in step S104, during the period when the drive motor is subjected to impact load causing transient mechanical slip fluctuations in the shaft system, the digital signal processor controls the phase deviation of the discrete stationary signal in the angle domain to be maintained within a safe deviation range, so that the discrete stationary signal in the angle domain remains in a boundary convergence state.

[0029] Example 1: When a large industrial rolling mill drive motor faces start-stop switching, high dynamic speed regulation, and variable load conditions caused by sudden biting of cold-rolled billets in the hot rolling section, a single-axis dynamic vibration sensor fixed at a measuring point on the surface of the drive motor housing collects continuous mechanical vibration acceleration signals in the time domain. Based on the transient physical slip and flexible torsional vibration fluctuations generated on the shaft stiffness transmission chain due to the transient torque change of the drive motor, a spatial stiffness exists between the transient speed phase recorded by the external speed encoder or the electromagnetic velocity parameter output by the frequency converter control layer and the dynamic phase at the measuring point of the single-axis dynamic vibration sensor. Span and phase propagation delay; Under the working condition of nonlinear continuous abrupt changes in the rotation frequency of the shaft, the vibration characteristic line generates a continuous frequency shift in the time domain, causing the conventional isochronous sampling Fourier transform method to output spectral line energy divergence, signal ambiguity, and dynamic aliasing distortion due to feature overlap. In order to obtain the intrinsic transient geometric phase evolution characteristics of the shaft system without spatiotemporal propagation delay under a single-sensor hardware topology, the digital signal processor receives the time-domain continuous mechanical vibration acceleration signal collected by the single-axis dynamic vibration sensor, and uses an energy operator transformation processing method with a specific amplitude gating threshold to process the signal, and inputs the signal... An adaptive digital bandpass filter discretely removes the high-order vibration carrier component containing the stator slot passage frequency component. From a physical mechanism perspective, the stator slot passage frequency component is essentially a high-frequency vibration generated by the electromagnetic force excitation produced by the rotor magnetic field scanning the stator slot openings, which is directly transmitted to the motor housing through the stator core. Its signal transmission path is limited to the rigid contact surface between the stator core and the housing, and there is no mechanical hysteresis caused by the torsional deformation of the shaft system in traditional encoders. Therefore, by extracting the instantaneous phase of this carrier from the housing vibration signal, a quasi-real-time phase reference synchronized with the internal electromagnetic excitation can be obtained. The spatiotemporal mismatch between the velocimetry device and the vibration measurement point is eliminated at the physical source. A timing hysteresis damping operator is configured at the front-end input of the digital signal processor. Positive and negative dead zone boundaries are set on both sides of the axis at the zero-crossing point of the amplitude of the higher-order vibration carrier component. The absolute values ​​of the positive and negative dead zone boundaries are equal, and their magnitudes increase with the root mean square value of the background noise as the noise increases. During this process, the safety gain coefficient is set between 3.0 and 6.0 based on the system signal-to-noise ratio characteristics. In this embodiment, a value of 4.5 is preferred to conform to the normal distribution. The principle is to ensure that the dead zone range can cover the amplitude fluctuation range of most random thermal noise, allowing only the cogging impact component that actually carries energy to cross the boundary, effectively filtering out parasitic vibrations in the environment. When the amplitude of the higher-order vibration carrier component completely crosses the positive and negative dead zone boundaries, the zero-crossing point capture operator latches the corresponding zero-crossing point time series, thereby calculating the local time interval between two adjacent zero-crossing points. The reciprocal of the local time interval is multiplied by one-half to calculate the local intrinsic modulation transient frequency. The digital signal processor subtracts the preset discrete transient slip correction coefficient from the calculated local intrinsic modulation transient frequency, and then divides the difference by the inherent rotor cogging number constant of the drive motor to calculate the transient shaft rotation frequency; the discrete transient slip correction coefficient... The value is obtained in advance through the calibration procedure described in Embodiment 5 below. It reflects the hysteresis component of the mechanical shaft relative to the electromagnetic synchronous speed under a specific load torque. During actual calculation, the processor reads the torque percentage of the current current loop feedback in real time, retrieves the corresponding slip voltage value from the calibration mapping table of the non-volatile memory, and converts it into a frequency compensation term to correct the calculation deviation caused by electromagnetic slip. The digital signal processor calculates the transient rotational frequency difference between two consecutive historical cycles, obtains the real-time angular acceleration trend of the mechanical shaft, linearly extrapolates the transient frequency shift of the current cycle based on the real-time angular acceleration trend, and calculates the center frequency prediction update increment accordingly. Specifically, the processor uses a first-order Taylor expansion for linear extrapolation, assuming the center frequency of the previous cycle is... The calculated trend of angular acceleration is as follows Then the predicted frequency shift increment for the current period ,in, The center frequency of the updated filter is set to the rotation period. Through this rule, the passband of the filter can sense the trend of speed change in advance, thereby completing the pre-adjustment of the cutoff boundary before the frequency change occurs. The center frequency prediction update rule of the adaptive digital bandpass filter is adjusted, and the passband cutoff boundary of the filter is controlled to move synchronously with the transient frequency shift trajectory of the higher-order vibration carrier component, eliminating the risk of zero-point false triggering and spectral line loss caused by speed and torque changes.

[0030] The digital signal processor (DSP) uses a second-order differential operator to differentiate the transient shaft rotation frequency twice in consecutive time, calculating the transient torsional acceleration value that characterizes the transient torsional vibration deviation of the shaft. From a control logic perspective, the transient torsional acceleration value essentially reflects the deviation of the second derivative of the rotation angle with respect to time. The processor performs two numerical integrations of this acceleration value in the discrete domain to restore it to the deviation of the spatial angular displacement, thereby determining the direction and amplitude of the sampling timestamp that need to be fine-tuned. When the transient torsional acceleration value exceeds the preset torsional vibration safety threshold, the DSP adjusts the step fine-tuning correction amount according to the preset step step. The step fine-tuning correction amount is superimposed on the nominal sampling interval, and the discrete sampling time of each sampling point is dynamically adjusted between two adjacent zero-crossing points to reconstruct an angle domain sampling grid aligned with the spatial rotation angle of the shaft. Based on the angle domain sampling grid, the original time-domain continuous mechanical vibration acceleration signal is nonlinearly interpolated and sampled to convert the time-domain waveform into an angle-domain discrete stationary signal.

[0031] The digital signal processor imports the discrete stationary signal in the angle domain into the characteristic order spectrum analysis engine. It performs Fourier transform on the discrete stationary signal in the angle domain to calculate the characteristic order spectrum of the mechanical structure vibration of the corresponding drive motor. It separately extracts the amplitude of the mechanical dynamic fault order characteristics caused by shaft imbalance or shaft misalignment, compares the amplitude with the preset deterioration alarm threshold, and outputs a fault alarm prompt. This ensures that the reconstructed angle domain sampling grid and the defect source of mechanical vibration are kept at the same physical measurement point and time reference. It controls the convergence of the generated discrete stationary signal in the angle domain under varying working conditions and noise interference conditions, eliminating the energy divergence defect of the order spectrum.

[0032] Example 2: In a mechanical vibration test platform containing a test drive motor with a power of 110kW and 28 rotor tooth slots, and a biaxial magnetic powder brake simulating the impact load of a rolling mill, a piezoelectric accelerometer mounted on the surface of the motor bearing housing acquires continuous mechanical vibration acceleration signals. The piezoelectric accelerometer has a measurement frequency response range of 1Hz to 20kHz and an amplitude linearity within 1.5%. The data acquisition unit has a continuous sampling frequency of 100kHz. Background noise with root mean square values ​​of 0.15m / s², 0.45m / s², and 1.25m / s² is input to the signal conditioning circuit, respectively. The test groups include the sample group of the present invention using the complete procedure, and the sample group with the timing hysteresis damping removed. The control group with partial operator loss and the control group with dead zone threshold deviation from the parameter window were included. The positive and negative dead zone boundaries of the time-series hysteresis damping operator were adjusted within the range of 0.05V to 0.50V. The dead zone boundary value was proportional to the spectral bandwidth of the monitored higher-order vibration carrier component to filter out zero-point fluctuations caused by high-frequency electromagnetic interference. Under a background noise of 1.25 m / s², the set values ​​for the positive and negative dead zone boundaries were 0.05V, 0.20V, and 0.50V, respectively. Under a variable load condition where the drive motor output speed increased from 300 r / min to 1200 r / min, the piezoelectric accelerometer output a time-domain continuous mechanical vibration acceleration signal containing dynamic aliasing distortion. At 1.25 m / s², the dead zone boundary value was... Under background noise of m / s², the energy amplitude of the stator slot passing frequency component in the input original vibration signal is less than the energy amplitude of the background noise. In the prototype of this invention, the digital signal processor uses an energy operator transformation including an amplitude gating threshold to extract the higher-order vibration carrier component containing the stator slot passing frequency component from the continuous mechanical vibration acceleration signal in the time domain. The voltage amplitude of the higher-order vibration carrier component fluctuates between ±0.85V. The digital signal processor sets positive and negative dead zone boundaries with equal absolute values ​​on both sides of the axis at the zero-crossing point of the amplitude of the higher-order vibration carrier component. The absolute value of the dead zone boundary increases monotonically with the increase of the root mean square value of the input background noise. When the amplitude of the higher-order vibration carrier component crosses the positive and negative dead zone boundaries... At the boundary, the zero-crossing point capture operator latches the zero-crossing point time series and calculates the local time interval between two adjacent zero-crossing points. Under the operating condition of 600 r / min, the local time interval is measured to be 35.2 μs to 36.4 μs. The digital signal processor multiplies the reciprocal of the local time interval by one-half to obtain the local intrinsic modulation transient frequency of 13.7 kHz to 14.2 kHz. In the partially missing control group, the noise level in the range of ±0.35V superimposed in the higher-order vibration carrier component directly triggers the zero-crossing point capture operator. The measured local time interval fluctuates randomly between 1.2 μs and 78.5 μs, and the calculated local intrinsic modulation transient frequency changes abruptly between 5.0 kHz and 41.5 kHz.

[0033] The digital signal processor (DSP) subtracts the discrete transient slip correction coefficient from the local intrinsic modulation transient frequency, then divides the difference by the inherent rotor cogging constant of the drive motor to calculate the transient shaft rotation frequency. Simultaneously, it calculates the rotation frequency difference between two consecutive historical cycles to determine the real-time angular acceleration trend. Based on this trend, it linearly extrapolates the transient frequency shift of the current cycle and determines the center frequency prediction update increment. The passband cutoff boundary of the adaptive digital bandpass filter moves with the center frequency prediction update increment, following the transient frequency shift trajectory of the higher-order vibration carrier component. The DSP uses a second-order differential operator to differentiate the transient shaft rotation frequency twice in consecutive time to calculate the transient torsional acceleration value. When the transient torsional acceleration value exceeds the torsional vibration safety threshold, the DSP adjusts the step fine-tuning correction amount in fixed 5μs increments. This step fine-tuning correction amount is superimposed on the nominal sampling interval, and the discrete sampling time of each sampling point is adjusted between two adjacent zero-crossing points to construct an angle domain sampling grid. The DSP then adjusts the original... The initial continuous mechanical vibration acceleration signal in the time domain is nonlinearly interpolated and sampled to convert the time-domain waveform into a discrete stationary signal in the angle domain. The characteristic order spectrum analysis engine performs a Fourier transform on the discrete stationary signal in the angle domain to output characteristic order spectra. When the sample group of this invention uses dead zone boundary voltages of 0.05V, 0.20V, and 0.50V respectively, the measured amplitudes of the first harmonic fault order components of the rotating shaft are 0.56m / s², 0.58m / s², and 0.55m / s², respectively, and the order spectrum is in a stable convergent state. In the out-of-range control group, when the absolute values ​​of the positive and negative dead zone boundaries are set to 1.50V, the zero-crossing point capture operator has no data output because the value exceeds the upper limit of voltage fluctuation of the higher-order vibration carrier component, and the transient shaft rotation frequency drops to 0m / s². When the absolute value of the dead zone boundary is set to 0.01V, the zero-crossing point capture operator outputs voltage spikes due to background noise interference, the characteristic order spectrum line experiences energy leakage, and the amplitude of the fault order component fluctuates between 0.12m / s² and 1.45m / s².

[0034] Under three operating conditions with background noise levels of 0.15 m / s², 0.45 m / s², and 1.25 m / s², the fault order component identification accuracy of the sample group in the characteristic order spectrum analysis engine was 99.2%, 98.5%, and 96.8%, respectively. The measurement variance of the fault order amplitude was below 0.05 m / s², and the fault order spectrum showed a monotonically converging trend with the increase of background noise intensity. When the industrial drive motor operation status online monitoring system faces variable speed and variable torque conditions, the non-stationary mechanical vibration acceleration signal collected by the piezoelectric accelerometer on the shell surface is converted and mapped into a discrete stationary signal in the angle domain aligned with the spatial rotation angle of the shaft. The characteristic order spectrum analysis engine discretely extracts the dynamic fault order characteristic amplitude of the mechanical shaft at the first harmonic position and displays abnormal status prompt information.

[0035] Example 3: When a test drive motor with a power of 110kW and 28 rotor cogs is operating under alternating load conditions, and the variable frequency speed control circuit introduces non-stationary electromagnetic interference, a piezoelectric accelerometer fixed to the surface of the test drive motor bearing housing collects a continuous mechanical vibration acceleration signal in the time domain. Because the continuous mechanical vibration acceleration signal contains time-varying noise voltage with a frequency response range of 1Hz to 20kHz and an amplitude linearity within 1.5% in the transmission chain, the cutoff edge of the adaptive digital bandpass filter will produce a spatiotemporal response deviation that fluctuates with the operating conditions, which may easily lead to the risk of false triggering of the zero point of the subsequent high-order vibration carrier component waveform. When the system starts, the digital signal processor calls the reference state calibration rule, selects the original digital voltage sequence containing 10 nominal cycles collected by the piezoelectric accelerometer under the no-load state of the test drive motor as the input source, and configures a temporary sliding buffer interval with a fixed number of data points of 1000 in the internal floating-point memory space to isolate and determine the intrinsic noise floor index of the external electromagnetic environment.

[0036] In the reference state calibration rule, the floating-point unit in the digital signal processor reads the full voltage data within the temporary sliding buffer interval, calculates the arithmetic mean of the absolute values ​​of the full voltage data as a static first-order reference, then calculates the square term of the difference between each voltage data point and the static first-order reference, and accumulates them to generate a sum of squared difference terms. The sum is divided by the degree-of-freedom constant 999 to calculate the noise variance envelope value. The square root of the noise variance envelope value is then taken to output the root mean square value of the background ambient noise. The digital signal processor multiplies the root mean square value term by a fixed safety gain coefficient to dynamically calculate the absolute voltage values ​​of the positive and negative dead zone boundaries, thereby establishing a waveform zero-point cutoff threshold without human intervention. Simultaneously, to eliminate the step response hysteresis of the filter passband cutoff boundary during transient frequency shift, the digital signal processor calculates the rotation axis rotation between two consecutive previous historical cycles. The frequency difference is used to obtain the real-time angular acceleration trend of the mechanical shaft. The real-time angular acceleration trend is multiplied by the square of the nominal sampling interval and then by a scaling factor of 0.5 to obtain the discrete phase displacement compensation amount. The discrete phase displacement compensation amount is then added as an increment to the center frequency prediction update rule to adjust the center frequency of the adaptive digital bandpass filter. When the transient load torque of the test drive motor causes the output speed fluctuation to exceed 30%, the digital signal processor calculates the corresponding step fine-tuning correction amount and adds the step fine-tuning correction amount to the nominal sampling interval to fine-tune each discrete sampling moment between two adjacent zero-crossing points. An angle domain sampling grid aligned with the spatial rotation angle of the shaft is reconstructed. The characteristic order spectrum analysis engine performs a Fourier transform on the reconstructed angle domain discrete stationary signal, and extracts the dynamic fault order characteristic amplitude of the mechanical shaft at the first harmonic position and outputs abnormal status prompt information.

[0037] Example 4: When the piezoelectric accelerometer acquires the continuous mechanical vibration acceleration signal of the housing surface under the stable operation of the drive motor without load, the digital signal processor calls the calibration program, extracts the raw voltage data within the sampling period and imports the data into a temporary sliding buffer interval; the floating-point unit reads the full voltage data in the buffer interval, calculates its absolute average value as a reference, calculates the square term of the difference between each voltage data and the reference and sums them, divides the sum by the difference after subtracting 1 from the total number of data to solve for the noise variance envelope value, and obtains the square root of the noise variance envelope value to output the root mean square value of the background ambient noise, and establishes the initial absolute voltage amplitude of the positive dead zone boundary and the negative dead zone boundary based on the root mean square value.

[0038] When the vibration characteristics of the drive motor housing change due to local wear during operation, the digital signal processor triggers a calibration program through a timed interrupt mechanism. It calculates the root mean square value of the current cycle and compares it with the historical reference benchmark to solve for the deviation increment. When the deviation exceeds the preset tolerance limit for 20 consecutive monitoring cycles, the digital signal processor calls the reconstruction program, uses the real-time variance envelope value of the current cycle to cover the historical reference benchmark, adjusts the center frequency prediction update rule of the adaptive digital bandpass filter, and adjusts the step fine-tuning correction amount used to reconstruct the angle domain sampling grid. This corrects the zero-point axis sway deviation caused by wear and controls the amplitude of the fault order component extracted discretely by the characteristic order spectrum analysis engine to converge within the deterioration alarm threshold range.

[0039] Example 5: When the vibration monitoring method used for the operation status of the drive motor faces the working conditions where the intrinsic stiffness of the drive motor shaft system varies with different manufacturing batches and the initial unbalance of the factory fluctuates randomly, there is an unpredictable slip response between the acceleration signal output by the high-frequency dynamic vibration sensing link and the frequency of the inverter's synchronous magnetic field. This causes the discrete transient slip correction coefficient in the self-demodulation path and the degradation alarm threshold in the characteristic order spectrum analysis engine to have a reference drift risk due to the lack of boundary calibration data.

[0040] To provide a parameter calibration source for the self-demodulation path, the test bench control unit drives the test drive motor to operate stably at the first preset speed node of 300 r / min, the second preset speed node of 600 r / min, the third preset speed node of 900 r / min, and the fourth preset speed node of 1200 r / min under zero load. The digital signal processor receives the time-domain continuous mechanical vibration acceleration signal collected by the high-frequency vibration acceleration sensor, and uses an energy operator transformation with a specific amplitude gating threshold to discretely extract the high-order vibration carrier component, which includes the stator slot passage frequency component. The high-order vibration carrier component is then captured by the zero-crossing point capture operator. The zero-crossing time series at the full dead zone boundary is used to calculate the actual transient shaft rotation frequency. The floating-point unit subtracts the transient shaft rotation frequency from the synchronous electromagnetic speed parameter output by the frequency converter to calculate the intrinsic slip voltage of the mechanical shaft. The intrinsic slip voltage corresponding to each preset speed node is stored in a calibration mapping table in non-volatile memory in discrete form. Before data storage, the processor normalizes the collected slip voltage value into a frequency drift in Hz using a preset voltage-controlled oscillator scaling factor (V / F conversion factor). The calibration table records the corresponding functional relationship between rotational speed and slip frequency for subsequent calculations. The system directly performs algebraic addition and subtraction operations, thereby using first-order linear interpolation rules to calculate the discrete transient slip correction coefficient corresponding to any transient speed point based on the real-time rotational frequency of the current mechanical shaft. Simultaneously, the test bench control unit introduces an eccentric mass block with a preset amplitude to simulate a standard shaft imbalance defect state. The test drive motor is controlled to operate continuously under rated load for 48 hours. The characteristic order spectrum analysis engine periodically extracts the real-time amplitude of the dynamic fault order characteristics at the first harmonic position, calculates the arithmetic mean and standard deviation of all real-time amplitudes within these 48 hours, multiplies the standard deviation by a constant 3, and then adds it to the arithmetic mean, solidifying it into a characteristic order spectrum analysis. The engine's degradation alarm threshold is set by writing the discrete transient slip correction coefficient mapping table generated by offline calibration and the verified degradation alarm threshold into the embedded register of the digital signal processor. Before application in the field environment, the rigid closed calibration parameters of the parameter matrix are latched, so that the reconstructed angle domain sampling grid has self-evident spatiotemporal alignment characteristics when dealing with large physical speed regulation and large fluctuations in the driving motor's operating state. The waveform feature spectrum of the reconstructed angle domain discrete stationary signal is controlled to maintain boundary convergence. The feature order spectrum analysis engine compares the amplitude of the real-time captured shaft dynamic fault order feature with the degradation alarm threshold, and outputs fault alarm prompt information when the deviation is exceeded.

[0041] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0042] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A vibration monitoring method for the operating status of a drive motor, characterized in that, Includes the following steps: Step S101: The continuous mechanical vibration acceleration waveform data of the drive motor housing surface is acquired by a high-frequency vibration acceleration sensor; Step S102: Apply energy operator transformation with a specific amplitude gating threshold to separate the high-order vibration carrier component containing the stator slot passage frequency component from the continuous mechanical vibration acceleration waveform data of the drive motor housing surface, determine its corresponding transient time-varying frequency line, and calculate the transient shaft rotation frequency based on the algebraic division relationship between the transient time-varying frequency line and the inherent rotor slot number constant, thereby establishing a self-demodulation path with adaptive demodulation capability for shaft slip fluctuations; Step S103: Determine the angular acceleration trend of the current period based on the difference in the rotational frequency of the shaft measured in the two adjacent historical periods in the previous period, and calculate the center frequency prediction update increment based on the angular acceleration trend to update the center frequency prediction update rule of the adaptive digital bandpass filter. Control the cutoff boundary of the adaptive digital bandpass filter to move in real time following the transient frequency shift trajectory of the higher-order vibration carrier component to block the zero-point pseudo-triggering caused by sudden changes in rotational speed and torque. Step S104: Perform second-order discrete-time difference calculation on the transient shaft rotation frequency to obtain the transient torsional vibration deviation of the shaft, which characterizes the dynamic flutter intensity of the shaft. Then, based on the transient torsional vibration deviation of the shaft, add a step fine-tuning correction amount on the nominal sampling interval to implement step fine-tuning, thereby dynamically adjusting the discrete sampling time of each sampling point and reconstructing the angle domain discrete stationary signal that is rigidly aligned with the spatial rotation angle of the shaft. Step S105: Import the discrete stationary signal in the angle domain into the characteristic order spectrum analysis engine, and perform Fourier transform on the discrete stationary signal in the angle domain to calculate the characteristic order spectrum of the corresponding drive motor mechanical structure vibration. Step S106: Extract the amplitude of the mechanical dynamic fault order features caused by shaft imbalance or shaft misalignment from the characteristic order spectrum lines, and evaluate the structural health of the drive motor based on the amplitude of the fault order features.

2. The vibration monitoring method for the operating status of a drive motor according to claim 1, characterized in that, After step S101, also The method includes the following: applying a time-series hysteresis damping operator to configure positive and negative dead zone boundaries on both sides of the zero-crossing point axis of the amplitude of the continuous mechanical vibration acceleration waveform data on the surface of the drive motor housing. The continuous mechanical vibration acceleration waveform data on the surface of the drive motor housing is truncated at zero point axis through the positive and negative dead zone boundaries. The absolute values ​​of the positive and negative dead zone boundaries are equal, and the magnitude of the absolute value increases with the increase of the root mean square value of the background ambient noise.

3. The vibration monitoring method for the operating status of a drive motor according to claim 1, characterized in that, In step S102, separating the high-order vibration carrier component containing the stator slot passage frequency component from the continuous mechanical vibration acceleration waveform data of the drive motor housing surface to determine its corresponding transient time-varying frequency line includes the following sub-steps: Step S1021, capturing non-uniform time discrete point signals containing the evolution history of the intrinsic transient geometric phase of the shaft system by a high-frequency dynamic vibration sensing link; Step S1022, extracting the time-frequency ridge of the non-uniform time discrete point signals by an energy operator to separate the high-order vibration carrier component containing the stator slot passage frequency component, and converting the high-order vibration carrier component containing the stator slot passage frequency component into a transient time-varying frequency line.

4. The vibration monitoring method for the operating status of a drive motor according to claim 1, characterized in that, In step S103, updating the center frequency prediction update rule of the adaptive digital bandpass filter includes the following sub-steps: Step S1031, calculate the transient rotational frequency difference between adjacent historical cycles to obtain the real-time angular acceleration trend of the mechanical shaft; Step S1032, linearly extrapolate the transient frequency shift of the current cycle based on the real-time angular acceleration trend, and calculate the center frequency prediction update increment to adjust the cutoff boundary of the adaptive digital bandpass filter, so that the passband of the adaptive digital bandpass filter moves in real time following the transient frequency shift trajectory of the higher-order vibration carrier component containing the stator slot passage frequency component.

5. The vibration monitoring method for the operating status of a drive motor according to claim 1, characterized in that, In step S104, the step-by-step fine-tuning is performed by superimposing a step-by-step correction amount on the nominal sampling interval based on the transient torsional vibration deviation of the shaft, including the following sub-steps: Step S1041, applying the second-order difference operator to differentiate the transient shaft rotation frequency twice in consecutive time, and calculating the transient torsional vibration acceleration value characterizing the transient torsional vibration deviation of the shaft; Step S1042, comparing the transient torsional vibration acceleration value with the determined safety threshold range, if the transient torsional vibration acceleration value exceeds the safety threshold range, then adjusting the step-by-step fine-tuning correction amount according to a fixed step step until the generated angular domain discrete stationary signal satisfies the boundary convergence condition.

6. The vibration monitoring method for the operating status of a drive motor according to claim 1, characterized in that, In step S106, assessing the structural health of the drive motor based on the amplitude of the fault order feature includes the following sub-steps: Step S1061, comparing the amplitude of the extracted fault order feature with a fixed degradation alarm threshold; Step S1062, when the amplitude of the fault order feature continuously exceeds the degradation alarm threshold, outputting a fault warning signal indicating that there is a risk of unexpected shutdown in the mechanical structure of the drive motor.

7. The vibration monitoring method for the operating status of a drive motor according to claim 1, characterized in that, In step S104, during the period when the drive motor is subjected to impact load causing transient mechanical slip fluctuations in the shaft system, the digital signal processor controls the phase deviation of the discrete stationary signal in the angle domain to be maintained within a safe deviation range so that the discrete stationary signal in the angle domain remains in a boundary convergence state.

Images