A method for reducing acoustic signal noise for high-speed railway rail corrugation acoustic diagnosis

Through the adaptive noise reduction method of spectrum subtraction, the problem of aerodynamic noise interference in the acoustic diagnosis of high-speed railway rail wave grinding is solved, and the signal-to-noise ratio is significantly improved, achieving the accurate extraction and diagnostic accuracy of the characteristic acoustic signals of the rail wave grinding.

CN114400011BActive Publication Date: 2024-11-26ENERGY SAVING & ENVIRONMENTAL PROTECTION & OCCUPATIONAL SAFETY & HEALTH RES INST OF CHINA ACAD OF RAILWAY SCI CORP LTD +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111438450.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-11-30
Publication Date
2024-11-26
Estimated Expiration
2041-11-30

AI Technical Summary

Technical Problem

The medium and high-frequency vibration noise caused by high-speed railway rail wave grinding is masked by strong aerodynamic noise interference, making it difficult to extract the characteristic acoustic signals of the rail wave grinding, and the existing technology is difficult to effectively reduce noise, affecting the diagnostic accuracy.

Method used

Adaptive noise reduction method of spectral subtraction is adopted, and the discrete Fourier transform, spectral subtraction algorithm and fast inverse Fourier transform are used, and the filter parameters are automatically adjusted to remove background noise interference, highlighting the characteristics of rail wave grinding acoustic signal.

Benefits of technology

It significantly reduces the background noise level and improves the signal-to-noise ratio from 0.21 to 0.46 or 4.14, significantly improving the accuracy and effectiveness of the acoustic diagnosis of rail wave grinding.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114400011B_ABST
    Figure CN114400011B_ABST
Patent Text Reader

Abstract

The invention discloses a method for reducing the noise of an acoustic signal for acoustic diagnosis of high-speed railway rail corrugation, which belongs to the technical field of high-speed railway noise, signal processing and fault diagnosis. The method comprises the following steps: assuming that the time domain signal of the acoustic signal is X(n), performing windowing and frame processing, and obtaining the acoustic signal of the i-th frame as x i (m), frame length is N, any frame sound signal x i (m) perform discrete Fourier transform; it is known that the duration of the acoustic signal without target signal (pure noise segment) is IS, and the corresponding number of frames is NIS, and the average energy value of the noise segment is calculated; finally, perform spectral subtraction algorithm; square root to obtain the amplitude after spectral subtraction #imgabs0# directly use the phase angle information before spectral subtraction to the acoustic signal after spectral subtraction, and perform inverse fast Fourier transform (IFFT) to obtain the time domain sequence of the acoustic signal after spectral subtraction processing #imgabs1# The acoustic signal denoising method for acoustic diagnosis of high-speed railway rail corrugation of the present invention utilizes spectral subtraction to adaptively reduce noise, which can effectively reduce the interference of aerodynamic noise and improve the effectiveness and accuracy of acoustic diagnosis of rail corrugation.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to an acoustic signal noise reduction method for acoustic diagnosis of high-speed railway rail corrugation, belonging to the technical field of high-speed railway noise, signal processing and fault diagnosis. Background Art

[0002] The operating environment of high-speed railways tends to be diversified and complex. The contact between wheels and rails is the key part of the train and track system. Rail corrugation is an important type of wheel-rail damage problem, and its mechanism and detection methods have attracted the attention of many researchers.

[0003] Rail corrugation is a periodic wavy uneven curve that appears on the surface of the rail. The basic mechanism of the formation and development of rail corrugation on high-speed railways is generally believed to be related to the dynamic response of the vehicle-track system. The initial damage and system vibration are coupled to form a feedback loop, which gradually forms and evolves into fixed-wavelength wear of the rail.

[0004] Rail corrugation on high-speed railways causes vibration noise caused by the medium and high frequency vibration response of the vehicle-track system, which directly affects the comfort of passengers and the quality of life of residents along the railway. At the same time, it deteriorates the operating status of various components of the system and aggravates further damage to the rail surface. The initial detection of rail corrugation on high-speed railways is an urgent problem that needs to be solved in the operation of high-speed railways. Acoustic diagnosis methods have the significant advantages of non-contact, convenient and fast. However, the acoustic signal under the high-speed train at a high speed of 300 kilometers per hour contains strong aerodynamic noise interference components, which seriously affects the extraction of characteristic acoustic signals of rail corrugation. Therefore, it is very necessary to study and propose an acoustic signal noise reduction method for acoustic diagnosis of rail corrugation on high-speed railways.

[0005] The aerodynamic noise in the bogie area of ​​a train running at high speed is broadband noise. In the process of microphone signal collection, it always exists as the most important part of the background noise, and there is more or less interference in the whole frequency domain. In the low frequency band, it may even drown out the target signal, so that the signal characteristics are not obvious, which causes troubles for feature extraction. For the aerodynamic noise background interference, since the prior characteristics of the target signal and the background noise are not fixed, it is difficult to achieve the ideal effect by simply using a fixed filter method.

[0006] Therefore, a method for adaptive filter denoising is provided. Under the premise of not much demand on the prior characteristics of the target signal and background noise, the method automatically adjusts its own parameters, and uses the filter parameters obtained at the previous moment to automatically adjust the filter parameters at the current moment, so as to adapt to the statistical characteristics of unknown or random changes of the target signal and background noise, so as to achieve the optimal filtering performance and remove the background noise interference to the greatest extent. According to the prominent characteristics of aerodynamic noise interference of high-speed railway operation noise, the spectral subtraction method and its improved form of adaptive noise reduction are used to perform global adaptive noise reduction on the bogie sound signal, which has become a technical problem that needs to be urgently solved in this technical field. Summary of the invention

[0007] The purpose of the present invention is to provide a sound signal denoising method for acoustic diagnosis of rail corrugation on high-speed railways. By using a spectral subtraction adaptive denoising method, the background noise of the sound signal under the high-speed railway vehicle can be reduced, so that the characteristic sound signal of rail corrugation is highlighted, thereby improving the accuracy of acoustic diagnosis of rail corrugation.

[0008] A method for reducing the noise of acoustic signals for acoustic diagnosis of rail corrugation on high-speed railways, comprising the following steps:

[0009] Assume that the time domain signal of the sound signal is X(n), and after windowing and framing, the sound signal of the i-th frame is x i (m), frame length is N, any frame sound signal x i (m) Perform discrete Fourier transform:

[0010]

[0011] Among them, m is the segmented sound signal, j is the imaginary part, X i (k) is the acoustic signal after discrete Fourier transform;

[0012] X i (k) Find the amplitude and phase angle for subsequent calculations; the amplitude is |X i (k)|, the phase angle is:

[0013]

[0014] It is known that the duration of the sound signal without target signal (pure noise segment) is IS, and the corresponding number of frames is NIS. The average energy value of the noise segment is calculated as:

[0015]

[0016] Finally, perform the spectral subtraction algorithm:

[0017]

[0018] Where a and b are the attenuation factor and gain factor respectively; D(k) is the subtracted signal;

[0019] Take the square root to get the amplitude after spectral subtraction Combined with the previous phase information By combining the phase insensitivity characteristics, the phase angle information before spectrum subtraction is directly applied to the acoustic signal after spectrum subtraction, and the inverse fast Fourier transform (IFFT) is performed to obtain the time domain sequence of the acoustic signal after spectrum subtraction.

[0020] Preferably, the measured sound signal is subjected to a 200 Hz high-pass filter and then subjected to spectral subtraction for noise reduction.

[0021] Beneficial effects:

[0022] The acoustic signal denoising method for acoustic diagnosis of high-speed railway rail corrugation of the present invention utilizes the characteristics that background noise is unrelated to the characteristic acoustic signal of rail corrugation, and uses the noise signal spectrum estimation value measured during the interval without target signal in the statistical stationary stage of the acoustic signal to replace the spectrum of the noise during the target signal period, and subtracts it from the spectrum of the noisy signal to obtain the estimated value of the target signal spectrum. The spectrum subtraction method has the characteristics of simple algorithm and small amount of calculation, which is convenient for realizing fast and efficient processing, and can often obtain a higher output signal-to-noise ratio.

[0023] The present invention is further described below through specific implementation modes and drawings, but it is not intended to limit the protection scope of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a basic form principle diagram of the spectral subtraction adaptive noise reduction of the present invention.

[0025] Figure 2-A This is the noise reduction of the actually measured sound signal before de-noising on a certain high-speed railway line in Application Example 1 of the present invention.

[0026] Figure 2-B This is the noise reduction of the actual measured sound signal after the noise is removed from a high-speed railway line in Application Example 1 of the present invention.

[0027] Figure 3-A It is a time-frequency diagram of the sound signal under the running train before noise reduction in Application Example 2 of the present invention.

[0028] Figure 3-B It is a time-frequency diagram of the noise reduction of the sound signal under the running train in Application Example 2 of the present invention. DETAILED DESCRIPTION

[0029] Unless otherwise specified, the devices used in the embodiments of the present invention are conventional devices available on the market in this field, the connections between the devices are conventional connections, and the methods used are conventional methods in this field.

[0030] Example 1

[0031] like Figure 1 As shown, it is a basic form principle diagram of the spectrum subtraction adaptive noise reduction of the present invention;

[0032] A method for reducing the noise of acoustic signals for acoustic diagnosis of rail corrugation on high-speed railways, comprising the following steps:

[0033] Assume that the time domain signal of the sound signal is X(n), and after windowing and framing, the sound signal of the i-th frame is x i (m), frame length is N, any frame sound signal x i (m) Perform discrete Fourier transform:

[0034]

[0035] Among them, m is the segmented sound signal, j is the imaginary part, X i (k) is the acoustic signal after discrete Fourier transform;

[0036] X i (k) Find the amplitude and phase angle for subsequent calculations; the amplitude is |X i (k)|, the phase angle is:

[0037]

[0038] It is known that the duration of the sound signal without target signal (pure noise segment) is IS, and the corresponding number of frames is NIS. The average energy value of the noise segment is calculated as:

[0039]

[0040] Finally, perform the spectral subtraction algorithm:

[0041]

[0042] Where a and b are the attenuation factor and gain factor respectively; D(k) is the subtracted signal;

[0043] Take the square root to get the amplitude after spectral subtraction Combined with the previous phase information By combining the phase insensitivity characteristics, the phase angle information before spectrum subtraction is directly applied to the acoustic signal after spectrum subtraction, and the inverse fast Fourier transform (IFFT) is performed to obtain the time domain sequence of the acoustic signal after spectrum subtraction.

[0044] Application Example 1

[0045] The acoustic signal denoising method for acoustic diagnosis of high-speed railway rail corrugation of the present invention is verified on-site on a high-speed railway: on a certain high-speed railway line, the acoustic signal under the vehicle running at a speed of 300 km / h is obtained, and the amplitude of the acoustic signal is normalized, and the denoising is performed using the method of the invention patent, and the time history information of the acoustic signal before and after the denoising is obtained; Figure 2-A As shown, it is the actual measured sound signal noise reduction before the noise reduction on a high-speed railway line in the application example 1 of the present invention; Figure 2-B As shown, it is the noise reduction of the measured sound signal after the noise reduction on a high-speed railway line in the application example 1 of the present invention;

[0046] Depend on Figure 2-A and Figure 2-B It can be seen that the normalized instantaneous amplitude of the acoustic signal after denoising using the improved spectral subtraction method is significantly reduced. This is because when the train is running at high speed, the main high-energy level noise is concentrated in the low-frequency aerodynamic noise. It can be seen that the improved spectral subtraction method can remove the background aerodynamic noise to a certain extent.

[0047] After calculation, the signal-to-noise ratio increased from 0.21 to 0.46.

[0048] Application Example 2

[0049] In order to further analyze the aerodynamic noise removal effect of the improved spectral subtraction method on the acoustic signal under the train when the train is running at high speed, the above signal is subjected to short-time Fourier analysis to analyze its noise reduction effect in the frequency domain. In order to highlight the noise reduction effect of the acoustic signal characteristics in the frequency band of wheel-rail disease, the measured acoustic signal is subjected to 200Hz high-pass filtering and then subjected to improved spectral subtraction noise reduction. The time-frequency characteristics of the acoustic signal before and after noise reduction are shown in the figure. Figure 3-A , Figure 3-B As shown. Figure 3-A As shown, it is a time-frequency diagram of the train running car sound signal before noise reduction in application embodiment 2 of the present invention; Figure 3-B As shown, it is a time-frequency diagram of the noise signal under the train running in application embodiment 2 of the present invention after noise reduction.

[0050] By Figure 3-A and Figure 3-B It can be seen that after the acoustic signal is processed by the acoustic signal adaptive denoising method proposed in the present invention, the yellow higher energy area is significantly reduced. In the time-frequency diagram, the contrast between the background area and the characteristic signal energy bright band area is significantly increased. In the entire frequency domain, the interference signal is significantly reduced. Therefore, the improved spectral subtraction method has an obvious effect on removing the broadband interference of aerodynamic noise and highlighting the characteristics of characteristic signals. The signal-to-noise ratio is increased to 4.14 after denoising.

[0051] Signal-to-noise ratio (SNR) indicator:

[0052] For acoustic signals with background noise, the signal-to-noise ratio (SNR) is used to characterize the noise content in the signal, which is defined as:

[0053]

[0054] In the formula, is the energy of the target sound signal, is the energy of the noise.

[0055] Considering that there is strong aerodynamic interference noise in the acoustic signal under the high-speed railway vehicle, it is necessary to reduce the background noise to highlight the characteristic acoustic signal of rail corrugation. The acoustic signal denoising method for acoustic diagnosis of rail corrugation of high-speed railway of the present invention utilizes spectral subtraction to adaptively reduce noise, which can effectively reduce the interference of aerodynamic noise and improve the effectiveness and accuracy of acoustic diagnosis of rail corrugation.

[0056] The acoustic signal denoising method for acoustic diagnosis of high-speed railway rail corrugation of the present invention utilizes high-speed railway undercarriage acoustic signal data collected by a microphone to reduce strongly disturbing aerodynamic background noise through an adaptive denoising method, and then utilizes the acoustic signal after reducing the aerodynamic background noise to highlight the characteristic information of the acoustic signal containing rail corrugation, and finally facilitates the identification of the rail corrugation characteristics.

Claims

1. A method for reducing the noise of acoustic signals for acoustic diagnosis of high-speed railway rail corrugation, the steps of which are as follows: Assume that the time domain signal of the sound signal is X(n), and after windowing and framing, the sound signal of the i-th frame is x i (m), frame length is N, any frame sound signal x i (m) Perform discrete Fourier transform: in, m is the segmented sound signal, j is the imaginary part, X i (k) is the acoustic signal after discrete Fourier transform; X i (k) Find the amplitude and phase angle for subsequent calculations; the amplitude is |X i (k)|, the phase angle is: It is known that the duration of the pure noise segment without target signal is IS, and the corresponding number of frames is NIS. The average energy value of the pure noise segment is calculated as: Finally, perform the spectral subtraction algorithm: Where a and b are the attenuation factor and gain factor respectively; D(k) is the subtracted signal; Take the square root to get the amplitude after spectral subtraction Combined with the previous phase information By combining the phase insensitivity characteristics, the phase angle information before spectrum subtraction is directly used in the acoustic signal after spectrum subtraction, and the inverse fast Fourier transform is performed to obtain the time domain sequence of the acoustic signal after spectrum subtraction. The measured sound signal is subjected to a 200 Hz high-pass filter and then denoised using spectral subtraction.