Search

CN-121980253-A - Multi-form signal processing method based on neighborhood maximum extraction transformation

CN121980253ACN 121980253 ACN121980253 ACN 121980253ACN-121980253-A

Abstract

The invention discloses a polymorphic signal processing method based on neighborhood maximum extraction transformation, which comprises the following steps of S1, setting proper sampling frequency, utilizing an acceleration sensor to collect vibration signals of a planetary gear box, S2, processing the vibration signals of the planetary gear box by using short-time Fourier transformation, obtaining corresponding time spectrum results, S3, defining a multi-neighborhood maximum detection algorithm based on local energy characteristics of the time spectrum obtained in S2, S4, defining a neighborhood maximum extraction operator by combining the multi-neighborhood maximum detection algorithm in S3 and time-frequency track characteristics, S5, utilizing the neighborhood maximum extraction operator constructed in S4 to adaptively extract time-frequency coefficients on a short-time Fourier transformation time-frequency plane, and S6, judging the working condition state of the planetary gear box according to the time spectrum results of the neighborhood maximum extraction transformation. The invention improves the energy aggregation of the frequency spectrum in the multi-form signal, has stronger noise robustness, and provides powerful guarantee for judging the state of the mechanical working condition of the vibration signal with the multi-form characteristic.

Inventors

  • ZHAO DEZUN
  • Du Shuaicong

Assignees

  • 北京工业大学

Dates

Publication Date
20260505
Application Date
20260118

Claims (5)

  1. 1. A multi-modal signal processing method based on neighborhood maximum extraction transformation, comprising the steps of: Step S1, setting proper sampling frequency, and collecting vibration signals of the planetary gear box by using an acceleration sensor; s2, processing a vibration signal of the planetary gear box by using short-time Fourier transform, and obtaining a corresponding time spectrum result; s3, defining a multi-neighborhood maximum detection algorithm based on the local energy characteristics of the time spectrum obtained in the step S2; s4, defining a neighborhood maximum extraction operator by combining the multi-neighborhood maximum detection algorithm and the time-frequency track characteristics in the S3; S5, adaptively extracting a time-frequency coefficient on a short-time Fourier transform time-frequency plane by using a neighborhood maximum extraction operator constructed in the S4 to obtain neighborhood maximum extraction transformation; And S6, judging the working condition state of the planetary gear box according to the time spectrum result of maximum extraction transformation of the neighborhood.
  2. 2. The method of claim 1, wherein in step S2, the time-frequency spectrum result of the short-time fourier transform is: ; Wherein, the Representing the collected vibration signals of the planetary gear box; the time center is indicated as such, Representing the center of the frequency, Representing imaginary units; is a Gaussian window function in the time domain, and has the expression of 。
  3. 3. The method for multi-modal signal processing based on the neighborhood maximum extraction transform according to claim 1, wherein in the step S3, a multi-neighborhood maximum detection algorithm is defined based on local energy characteristics of the time spectrum, and the expression is: ; Wherein, the And Discrete indexes respectively representing a frequency center and a time center; Is that Is a modulus of the model.
  4. 4. The method for multi-modal signal processing based on the neighborhood maximum extraction transform according to claim 1, wherein in the step S4, a neighborhood maximum extraction operator is defined by combining a multi-neighborhood maximum detection algorithm and a time-frequency trajectory feature: ; Wherein, the Representing a summation function.
  5. 5. The method for processing the multi-modal signal based on the neighborhood maximum extraction transform according to claim 1, wherein in the step S5, the neighborhood maximum extraction transform is obtained based on the time-frequency coefficient on the neighborhood maximum extraction operator adaptive extraction short-time fourier transform time-frequency plane, and the expression is: ; Wherein, the And extracting an operator for the neighborhood maximum.

Description

Multi-form signal processing method based on neighborhood maximum extraction transformation Technical Field The invention belongs to the field of signal processing, and particularly relates to a polymorphic signal processing method based on neighborhood maximum extraction transformation. Background Rotary machines play a vital role in the fields of aviation, transportation, agriculture, etc. Because such equipment operates under heavy load, fatigue and other working conditions for a long time, core parts and key structures of the equipment inevitably suffer from damage to different degrees. The state monitoring of the rotary machine mainly relies on a sensor to collect vibration signals and obtain relevant fault information through signal analysis. As the operating environment of a device becomes increasingly complex, the monitoring signal typically exhibits a polymorphic feature that mixes a harmonic-like signal with a transient-like signal. How to efficiently identify and extract key information in polymorphic signals remains an important technical challenge. Time-frequency analysis methods are widely used to process multi-modal signals because they can simultaneously characterize time-domain information and frequency components. Traditional time-frequency analysis methods cannot provide high-readability time-frequency spectrum due to the hessian-burg uncertainty principle and cross term interference. Therefore, scholars have proposed post-processing techniques to obtain a highly concentrated time spectrum by rearranging the time-frequency coefficients in the time-frequency plane to instantaneous frequencies or group delay trajectories. The reassignment method reassigns the time-frequency coefficient along the frequency and time directions at the same time, thereby improving the energy aggregation of the time spectrum. Because the redistribution method loses the key phase information, the signal cannot be reconstructed back to the time domain, and the time-frequency resolution of the signal still has room for improvement. The time-frequency multiple compression transformation and the time-frequency extraction transformation carry out mode division on each time-frequency position based on a frequency modulation division criterion, and then the time-frequency positions are processed and fused by a unidirectional time-frequency analysis method, so that the resolution of the frequency spectrum of the multi-form signal is improved. The two algorithms need to perform mode division and unidirectional processing on each time-frequency point, so that the overall calculation efficiency is reduced, and meanwhile, the energy aggregation performance still needs to be further improved. In summary, although the prior time-frequency analysis method has been applied to the multi-modal signal processing in the state monitoring of the rotating machine, it is still required to further improve the energy concentration and the calculation efficiency. Disclosure of Invention Aiming at the defects of the prior art, the invention provides a polymorphic signal processing method based on neighborhood maximum extraction transformation, which can remarkably improve the energy concentration of a time spectrum and the calculation efficiency of an algorithm and realize the accurate representation of polymorphic signals. In order to achieve the above purpose, the technical scheme adopted by the invention is that the multi-form signal processing method based on the neighborhood maximum extraction transformation comprises the following steps: Step S1, setting proper sampling frequency, and collecting vibration signals of the planetary gear box by using an acceleration sensor; s2, processing a vibration signal of the planetary gear box by using short-time Fourier transform, and obtaining a corresponding time spectrum result; s3, defining a multi-neighborhood maximum detection algorithm based on the local energy characteristics of the time spectrum obtained in the step S2; s4, defining a neighborhood maximum extraction operator by combining the multi-neighborhood maximum detection algorithm and the time-frequency track characteristics in the S3; S5, adaptively extracting a time-frequency coefficient on a short-time Fourier transform time-frequency plane by using a neighborhood maximum extraction operator constructed in the S4 to obtain neighborhood maximum extraction transformation; And S6, judging the working condition state of the planetary gear box according to the time spectrum result of maximum extraction transformation of the neighborhood. As a further improvement of the present invention, in the step S2, the time-frequency spectrum result of the short-time fourier transform is: Wherein, the Representing the collected vibration signal of the planetary gear box.The time center is indicated as such,Representing the center of the frequency,Representing imaginary units.Is a Gaussian window function in the time domain, and has the expressi