Search

CN-121977825-A - Dynamic detection system and method for axle gear meshing precision

CN121977825ACN 121977825 ACN121977825 ACN 121977825ACN-121977825-A

Abstract

The invention discloses a dynamic detection system and a dynamic detection method for meshing precision of an axle gear, which relate to the technical field of gear transmission detection and comprise the steps of acquiring vibration noise signals, decomposing meshing components and modulation components, and obtaining a smooth rotating speed curve. And calculating the instantaneous meshing frequency based on the rotating speed, generating a reference phase, and performing angular domain resampling on the meshing component to obtain an angular domain stationary signal. The amount of periodic fluctuation of the meshing order characteristic energy is extracted by order analysis, and at the same time, a modulation depth index is calculated from the modulation component. And (5) building a feature vector by fusing energy fluctuation, modulation depth and temperature gradient trend, inputting a pre-training model, and outputting meshing precision quantization scores. The method overcomes the interference of rotation speed fluctuation on analysis, and realizes direct and online quantitative evaluation of engagement accuracy under variable working conditions.

Inventors

  • CHENG HAIDONG

Assignees

  • 沈阳宝驹汽车传动系统有限公司

Dates

Publication Date
20260505
Application Date
20260403

Claims (10)

  1. 1. The dynamic detection method for the meshing precision of the axle gear is characterized by comprising the following steps of: Collecting multichannel real-time vibration signals and noise signals of a gear pair, decomposing the multichannel real-time vibration signals and noise signals into meshing vibration components directly related to meshing frequency and modulation components related to gear defects, and processing the meshing vibration components and the modulation components to obtain a smooth rotating speed curve and a smooth torque curve; Calculating the instantaneous meshing frequency of the gear pair in a detection period based on the smooth rotating speed curve, and generating a reference phase signal for subsequent analysis according to the instantaneous meshing frequency; angular domain resampling is carried out on the meshing vibration component by utilizing the reference phase signal, and a vibration signal of a time sequence is converted into an angular domain vibration signal which is strictly synchronous with the rotation angle of the gear; performing order tracking analysis on the angular domain vibration signal, extracting characteristic frequency band energy centered on gear engagement order and frequency multiplication thereof, and calculating fluctuation amount of the characteristic frequency band energy in a continuous rotation period; Separating amplitude modulation characteristics and frequency modulation characteristics from the modulation components, and calculating modulation depth indexes of the amplitude modulation characteristics and the frequency modulation characteristics; Fusing the fluctuation quantity of the characteristic frequency band energy, the modulation depth index and the gradient change trend of the temperature signals of the plurality of measuring points to construct a multidimensional meshing state characteristic vector; Inputting the multidimensional meshing state feature vector into a pre-trained gear meshing precision evaluation model, and outputting a specific quantization score representing the meshing precision of the current gear pair by the gear meshing precision evaluation model.
  2. 2. The method for dynamic detection of axle gear engagement accuracy of claim 1, wherein said obtaining a smoothed speed and torque curve comprises: Collecting multichannel real-time vibration signals and noise signals of a gear pair in the load operation process, wherein the gear pair comprises a driving gear and a driven gear; synchronously collecting motor rotation speed signals, torque signals and a plurality of measuring point temperature signals on a gear box shell, wherein the motor rotation speed signals and the torque signals are related to meshing actions of the gear pair; Performing time-frequency joint analysis on the multichannel real-time vibration signal and the noise signal to decompose the multichannel real-time vibration signal and the noise signal into a meshing vibration component directly related to meshing frequency and a modulation component related to gear defects; And filtering and resampling the motor rotating speed signal and the torque signal to eliminate electric noise and pulse interference in the signals and obtain a smooth rotating speed curve and a smooth torque curve.
  3. 3. The dynamic detection method for meshing accuracy of an axle gear according to claim 2, wherein the time-frequency joint analysis of the multichannel real-time vibration signal and the noise signal is performed to decompose the multichannel real-time vibration signal into a meshing vibration component directly related to a meshing frequency and a modulation component related to a gear defect, comprising: Applying wavelet packet transformation to the real-time vibration signal of each channel, decomposing the signal into a plurality of sub-bands preset by the meshing frequency theoretical range; identifying a sub-band with energy most concentrated in the instantaneous meshing frequency and the frequency multiplication vicinity thereof, and reconstructing a signal component of the sub-band to obtain the meshing vibration component; Subtracting the meshing vibration component from the original vibration signal to obtain a residual vibration signal; performing envelope demodulation analysis on the residual vibration signal, calculating an envelope spectrum of the residual vibration signal, and identifying a spectral peak component corresponding to the inherent fault characteristic frequency of the gear in the envelope spectrum; and extracting a time domain envelope waveform associated with the spectral peak component to obtain the modulation component.
  4. 4. A dynamic detection method for axle gear engagement accuracy as claimed in claim 3, wherein calculating an instantaneous engagement frequency of the gear pair in a detection period based on the smoothed rotation speed curve, and generating a reference phase signal for subsequent analysis in accordance with the instantaneous engagement frequency, comprises: performing differential operation on the smooth rotating speed curve to obtain instantaneous angular acceleration data; Multiplying the smooth rotating speed curve by the number of teeth of the driving gear to obtain a theoretical meshing frequency curve; On the basis of the theoretical meshing frequency curve, the instantaneous angular acceleration data is utilized to carry out secondary correction on the frequency change trend, so as to obtain the instantaneous meshing frequency; Performing time integral operation on the instantaneous meshing frequency to obtain an instantaneous phase function which monotonically increases along with time; The instantaneous phase function is equiangularly resampled with fixed phase increments to produce the reference phase signal that is uniformly increased by three hundred sixty degrees each revolution of the gear.
  5. 5. The dynamic detection method for axle gear meshing accuracy of claim 4, wherein angular resampling the meshing vibration component with the reference phase signal converts a time-series vibration signal into an angular vibration signal that is strictly synchronized with the gear rotation angle, comprising: establishing a mapping relation between a time axis of the meshing vibration component and the reference phase signal; calculating the vibration amplitude of the corresponding moment on the time axis of the meshing vibration component by using each fixed phase increment of the reference phase signal as a target angle point through an interpolation algorithm; Sequentially arranging the vibration amplitudes calculated by all target angle points to form the angular domain vibration signal taking the rotation angle as an independent variable; And carrying out multi-period average processing on the angular domain vibration signals, aligning and averaging the angular domain signals of a plurality of continuous rotation periods on the angular domain so as to enhance the periodic meshing characteristics and inhibit random noise.
  6. 6. The dynamic detection method for axle gear engagement accuracy according to claim 5, wherein performing order tracking analysis on the angular domain vibration signal, extracting characteristic band energy centered on gear engagement order and its multiple, and calculating the fluctuation amount of the characteristic band energy in successive rotation periods, comprises: performing angular domain Fourier transform on the angular domain vibration signals subjected to multi-period average processing to obtain an order spectrum of the angular domain vibration signals; in the order spectrum, locating the spectrum peak positions corresponding to the theoretical meshing order, the double meshing order and the triple meshing order; Setting a fixed order bandwidth by taking each spectrum peak position as a center, and calculating integral energy in the order bandwidth as the characteristic frequency band energy of the corresponding order; intercepting a plurality of continuous complete gear rotation period segments from the original angular domain vibration signals which are not subjected to average treatment; Respectively calculating integral energy of each rotation period segment in the fixed order bandwidth to obtain a time sequence related to characteristic frequency band energy; And calculating the ratio of the standard deviation to the mean value of the time sequence as the fluctuation amount of the characteristic frequency band energy in the continuous rotation period.
  7. 7. The dynamic detection method for axle gear engagement accuracy according to claim 6, wherein separating an amplitude modulation characteristic and a frequency modulation characteristic from the modulation component, and calculating a modulation depth index of the amplitude modulation characteristic and the frequency modulation characteristic, comprises: Performing Hilbert transform on the modulation component to obtain an analysis signal thereof; Calculating the instantaneous amplitude of the analytic signal, wherein the waveform of the instantaneous amplitude changing along with time is the amplitude modulation characteristic; Calculating the instantaneous phase of the analytic signal, and deriving the instantaneous phase to obtain an instantaneous frequency, wherein the fluctuation part of the instantaneous frequency around the center frequency is the frequency modulation characteristic; Counting peak values and valley values of the amplitude modulation characteristics in a detection period, and calculating the percentage of the difference value of the peak values and the valley values relative to the average amplitude value to be used as an amplitude modulation depth index; and counting the maximum value and the minimum value of the frequency modulation characteristic in a detection period, and calculating the percentage of the difference value of the maximum value and the minimum value relative to the center frequency to be used as a frequency modulation depth index.
  8. 8. The method for dynamically detecting meshing precision of an axle gear according to claim 7, wherein the step of fusing the fluctuation amount of the characteristic band energy, the modulation depth index and the gradient variation trend of the temperature signals of the plurality of measuring points to construct a multi-dimensional meshing state characteristic vector comprises the steps of: For temperature signals from different measuring points, calculating linear fitting slopes of the temperature signals in a detection period to obtain a temperature gradient value of each measuring point; Carrying out standardization processing on the fluctuation amount of the characteristic frequency band energy, the amplitude modulation depth index, the frequency modulation depth index and the temperature gradient values of all the measuring points so that all the indexes are in the same numerical dimension; All the indexes after the standardization processing are spliced into a one-dimensional array according to a preset sequence, wherein the one-dimensional array is the multi-dimensional meshing state characteristic vector; Recording a corresponding working condition label when the multidimensional meshing state characteristic vector is constructed, wherein the working condition label comprises a load torque range and average rotating speed information.
  9. 9. The method of claim 8, wherein inputting the multi-dimensional meshing state feature vector into a pre-trained gear meshing accuracy assessment model comprises: The pre-trained gear engagement precision evaluation model is constructed based on a deep neural network, and the number of neurons of an input layer is the same as the dimension of the multidimensional engagement state feature vector; The gear engagement precision evaluation model comprises a plurality of fully-connected hidden layers and a regression output layer, wherein the regression output layer outputs a specific quantization score ranging from zero to one hundred; Before the multidimensional meshing state feature vector is input into a model, data standardization parameters corresponding to the current working condition label are required to be called, and secondary standardization is carried out on the multidimensional meshing state feature vector so that the multidimensional meshing state feature vector is consistent with the distribution of model training data; And the gear engagement precision evaluation model finally outputs the specific quantization score through forward propagation calculation according to the input feature vector.
  10. 10. A dynamic detection system for axle gear engagement accuracy comprising a memory, a processor and a computer program stored in the memory and running on the processor, characterized in that the processor, when executing the computer program, realizes the steps of a dynamic detection method for axle gear engagement accuracy according to any one of claims 1 to 9.

Description

Dynamic detection system and method for axle gear meshing precision Technical Field The invention belongs to the technical field of gear transmission detection, and particularly relates to a dynamic detection system and method for axle gear meshing precision. Background In the field of automobile and engineering machinery manufacturing, the meshing precision of an axle gear directly influences the performance and reliability of a transmission system. Existing gear state monitoring techniques rely primarily on vibration signal analysis under steady-state or quasi-steady-state conditions. The conventional method collects signals by adopting fixed time sampling frequency and identifies fault characteristics through spectrum analysis. For the modulation phenomenon present in the signal, the prior art is generally only used for judging the fault type. The prior art solutions have drawbacks. Under the real dynamic working conditions of variable rotation speed and variable load, the fluctuation of the rotation speed of the gear can lead to the diffusion and the blurring of the frequency spectrum of the vibration signal. Analysis based on a fixed sampling rate cannot accurately separate pure meshing vibration components, so that extracted features are severely distorted and lack of consistency. Meanwhile, the traditional modulation analysis is taken as a qualitative auxiliary means, and the modulation depth cannot be established as a direct index for quantifying the comprehensive meshing precision of the gear pair. This results in an inability to accurately evaluate the quality of the assembly of the gears and the meshing smoothness on line. The invention aims to solve the technical problem of precisely quantifying the gear engagement precision under the dynamic working condition. The method has the advantages that interference caused by rotation speed fluctuation on analysis signals is required to be eliminated, vibration data standard which is strictly synchronous with the rotation angle of the gear is obtained, and a quantitative index capable of directly and sensitively reflecting the degree of meshing instability caused by the geometric error and assembly error of the gear is required to be constructed so as to replace qualitative fault judgment. Disclosure of Invention The present invention aims to solve at least one of the technical problems existing in the prior art; Therefore, the invention provides a dynamic detection method for the meshing precision of an axle gear, which comprises the following steps: Collecting multichannel real-time vibration signals and noise signals of a gear pair, decomposing the multichannel real-time vibration signals and noise signals into meshing vibration components directly related to meshing frequency and modulation components related to gear defects, and processing the meshing vibration components and the modulation components to obtain a smooth rotating speed curve and a smooth torque curve; Calculating the instantaneous meshing frequency of the gear pair in a detection period based on the smooth rotating speed curve, and generating a reference phase signal for subsequent analysis according to the instantaneous meshing frequency; angular domain resampling is carried out on the meshing vibration component by utilizing the reference phase signal, and a vibration signal of a time sequence is converted into an angular domain vibration signal which is strictly synchronous with the rotation angle of the gear; performing order tracking analysis on the angular domain vibration signal, extracting characteristic frequency band energy centered on gear engagement order and frequency multiplication thereof, and calculating fluctuation amount of the characteristic frequency band energy in a continuous rotation period; Separating amplitude modulation characteristics and frequency modulation characteristics from the modulation components, and calculating modulation depth indexes of the amplitude modulation characteristics and the frequency modulation characteristics; Fusing the fluctuation quantity of the characteristic frequency band energy, the modulation depth index and the gradient change trend of the temperature signals of the plurality of measuring points to construct a multidimensional meshing state characteristic vector; Inputting the multidimensional meshing state feature vector into a pre-trained gear meshing precision evaluation model, and outputting a specific quantization score representing the meshing precision of the current gear pair by the gear meshing precision evaluation model. Further, the obtaining a smooth rotation speed curve and a smooth torque curve includes: Collecting multichannel real-time vibration signals and noise signals of a gear pair in the load operation process, wherein the gear pair comprises a driving gear and a driven gear; synchronously collecting motor rotation speed signals, torque signals and a plurality of measuring point temperature signals on