CN-121632326-B - Rotating motor natural frequency calculation method under operation condition

CN121632326BCN 121632326 BCN121632326 BCN 121632326BCN-121632326-B

Abstract

The method for calculating the natural frequency of the rotating motor under the operating condition discloses a method for acquiring the natural frequency of the rotating motor under the operating condition, and belongs to the technical field of on-line recognition of the structural modal parameters of the motor. The method comprises the steps of uniformly distributing N more than or equal to 4 unidirectional vibration sensors on tooth parts of a stator, carrying out operation by a motor, synchronously collecting vibration signals, carrying out zero-phase band-pass filtering on the signals, calculating a cross-correlation function of a reference point and a response point, modeling a complex mode into a complex exponential series, constructing a matrix equation through single-side cross-power spectrum transformation and multi-starting-time sampling, obtaining coefficient vectors through SVD (singular value decomposition), constructing characteristic polynomial root by the coefficients, mapping to obtain a system continuous pole, and further outputting natural frequency and damping ratio. The whole process does not need to be stopped, excitation is not needed to be externally added, the noise immunity is strong, the frequency identification error is less than 1%, and the mode data under the real boundary condition is provided for motor NVH optimization and structural modification.

Inventors

MA CHENG
ZHANG QIDAN
MENG WEITAO
SHENG ZIHAN
SUN JINGYI
Hong Ziteng
ZHOU QIANG
LI ZHENGDA
SHI ZHAO
GUAN YING
SONG MIN
ZHOU ZHENLIU
SHI XUMEI
Liu Erge

Assignees

沈阳工程学院

Dates

Publication Date: 20260508
Application Date: 20251210

Claims (10)

1. The method for calculating the natural frequency of the rotating motor under the operating condition is characterized by comprising the following steps of: step S1 sensor arrangement N unidirectional vibration sensors are uniformly distributed at the same axial position of the tooth part of the stator core of the rotating motor; step S2, signal acquisition The rotary motor is in a rated load running state, N paths of vibration signals are synchronously collected, and the sampling frequency fs is obtained; step S3 band-pass filtering Zero-phase Butterworth band-pass filtering is carried out on each path of signal, the frequency bands [ fL, fH ] are reserved, fL is the filtering lower limit frequency, and fH is the filtering upper limit frequency; step S4, calculating the cross-correlation function Optionally, one path is a reference point i, the rest N-1 paths are response points j, and a cross-correlation function Rij (tau) is calculated: ; wherein T is the acquisition duration; τ is a time lag variable, the value range is 0- τ - τmax, τmax=1/fL; xi (t) is a vibration signal of the reference point i at the time t; xj (t+τ) is a vibration signal at time t+τ at response point j; step S5, modeling in complex mode The discrete sampled Rij (kDeltat) is expressed as a complex exponential series ; Wherein: k is a discrete sample number, k=0, 1,..k-1, k=τmax/Δt; Δt=1/fs is the sampling time interval; m is the order of the mode to be identified; aij, r is the remainder of the r-th order modality at sensor pair (i, j); λr is the system's r-th order continuous pole (s-1), representing the conjugate; Step S6, pole-mode parameter conversion Decomposing λr into ; The xi r is the modal damping ratio of the r order; ωr is the natural circular frequency of the damping-free mode of the r-th order, and the relation between ωr=2pi fr and the natural frequency fr is set; step S7, constructing single-side cross power spectral density Semi-infinite Fourier transform is carried out on Rij (tau) to obtain a single-side cross power spectrum, and a frequency domain model is established through single-side cross power spectrum density transform: ; wherein Gij (f) is the unilateral cross-power spectral density of the sensor pair (i, j); Take its discrete form ; Δf=fs/n_fft, n_fft being the FFT point number; Step S8, establishing and solving a matrix equation At L different initial sampling moments Where l=0, 1..the Gij (f, t l ) is calculated at L-1, a matrix equation is constructed ; Wherein: g is an L x 1-dimensional observation vector, and the element is Gij (f, tl); phi is an Lx2m-dimensional Vandermonde matrix, and the elements are composed of complex exponential basis functions e { λrtl }; a= [ a0, a1, ], a2m-1} ] T is the coefficient vector to be solved; Solving a least square solution by adopting singular value decomposition to obtain a coefficient vector a: ; step S9, solving the system pole Constructing a characteristic polynomial from the coefficient vector a and taking root: ; obtain a discrete pole zr, map back to a continuous pole λr: ; step S10, outputting natural frequency and damping ratio Substituting λr into the following formula, the natural frequency fr of the r-th order can be obtained simultaneously: ; Modal damping ratio ζr: 。
2. The method for calculating the natural frequency of the rotating electrical machine under the operating condition according to claim 1, wherein in the step S1, N is equal to or greater than 4 and is an even number, and adjacent sensors are circumferentially spaced by an integer multiple of 90 ° in electrical angle.
3. The method for calculating the natural frequency of the rotating electrical machine under the operating condition according to claim 1, wherein in the step S2, the sampling frequency fs satisfies fs not less than 10·fmax, wherein fmax is the highest modal frequency to be identified, T is the acquisition duration and T is not less than 1000/fmin, and fmin is the lowest modal frequency to be identified.
4. A method for calculating the natural frequency of a rotating electrical machine under operating conditions according to claim 1, wherein in said step S3, The filtering order is 4-8.
5. The method for calculating the natural frequency of the rotating electrical machine under the operating condition according to claim 1, characterized in that in said step S5, m=4.
6. The method for calculating the natural frequency of the rotating electrical machine under the operating condition according to claim 1, wherein the vibration signal in the step S2 is an acceleration signal.
7. The method for calculating the natural frequency of a rotating electrical machine under an operating condition according to claim 1, wherein the vibration signal in step S2 is a speed signal.
8. The method for calculating the natural frequency of the rotating electrical machine under the operating condition according to claim 1, wherein in the step S8, L is more than or equal to 200.
9. The method for calculating the natural frequency of the rotating electrical machine under the operating condition according to claim 1, wherein in the step S8, the number of the reserved main singular values is equal to 2m.
10. The method for calculating the natural frequency of the rotating electrical machine under the operation condition according to claim 1, wherein the rotating electrical machine is a permanent magnet synchronous motor, an asynchronous motor or a direct current motor.

Description

Rotating motor natural frequency calculation method under operation condition Technical Field The invention relates to the technical field of online identification of structural modal parameters of a rotating motor, in particular to a method for extracting natural frequencies of a motor by only utilizing stator surface vibration response under the operating condition of motor belt running. Background Rotating electrical machines are widely used in industrial production, but are susceptible to various external and internal factors during their operation, resulting in vibration and noise problems. The motor resonance can lead to equipment structure vibration to aggravate, and long-term probably causes spare part fatigue fracture or not hard up, influences stability and the durability of equipment, causes even that equipment wholeness can decline or damage. Vibration caused by resonance increases friction inside the apparatus or between adjacent parts, thereby increasing energy loss of the system, resulting in waste of energy. The resonance vibration may generate noise, interfere with the surrounding environment and staff, and even damage the health of the human body. Therefore, it is important to develop a natural frequency calculation method based on the running state of the motor. Natural frequency refers to the frequency at which the system naturally vibrates at a particular frequency in the absence of external forces. In a mechanical system, the natural frequency is generally a frequency of a natural vibration mode of the system, and is a characteristic inherent to the system itself. Reflects the rigidity and mass distribution condition of the system structure, and has important guiding significance for structural design and optimization. The structural parameters are reasonably selected, so that the natural frequency is not in the frequency range of the system excited by the outside, the occurrence of resonance phenomenon can be effectively avoided, and the stability and the safety of the system are improved. Natural frequency is one of the important parameters of modal analysis and can be used to analyze the vibration modal characteristics of the system. By modal analysis, the vibration mode distribution condition, the frequency response characteristic and the vibration mode parameters (such as vibration mode shape, damping ratio and the like) of the system can be known, and references are provided for system design and optimization. The existing method for acquiring the natural frequency of the rotating motor mostly adopts a shutdown mode test (EMA), requires additional excitation equipment and cannot reflect real boundary conditions. Operational Modal Analysis (OMA) does not require downtime, but conventional OMA uses only self-power spectrum peak pickup, resolution is highly disturbed by frequency aliasing and noise, and it is difficult to give the damping ratio at the same time. Therefore, there is a need for an online identification method that does not require a shutdown, does not require additional excitation, has high noise immunity, and can simultaneously output a natural frequency to damping ratio. Disclosure of Invention The invention aims at the problems, and provides a method for calculating the natural frequency of a rotating motor under the operating condition, which is realized by the following steps: step S1 sensor arrangement N unidirectional vibration sensors are uniformly distributed at the same axial position of the tooth part of the stator core of the rotating motor. Step S2, signal acquisition And under the rated load running state, the rotating motor synchronously collects N paths of vibration signals, and the sampling frequency fs. Step S3 band-pass filtering Zero-phase Butterworth band-pass filtering is carried out on each signal, the frequency bands [ fL, fH ] are reserved, fL is the lower limit frequency (Hz) of filtering, and fH is the upper limit frequency (Hz) of filtering. Step S4, calculating the cross-correlation function Optionally, one path is a reference point i, the rest N-1 paths are response points j, and a cross-correlation function Rij (tau) is calculated: ; Wherein τ is a time delay variable(s), and the value range is 0- τ - τmax, τmax=1/fL. Xi (t) is the vibration signal (unit: m/s 2 or m/s) of the reference point i at time t; xj (t+τ) is the vibration signal (unit is identical to xi) at time t+τ at response point j. Step S5, modeling in complex mode The discrete sampled Rij (kDeltat) is expressed as a complex exponential series ; Wherein: k is a discrete sample number, k=0, 1,..k-1, k=τmax/Δt; Δt=1/fs is the sampling time interval(s); m is the order of the mode to be identified (dimensionless); Aij, r is the remainder (complex number, unit consistent with Rij) of the r-th order modality at the sensor pair (i, j); λr is the system's order successive pole (s-1) representing the conjugate. Step S6, pole-mode parameter conversion Decomposing λr into ; The xi r is