Noise reduction software makes NVH diagnosis easier

Background introduction

When we diagnose NVH problems, there will be some environmental noises in the audio recorded by microphones or other recording equipment that we don't need to keep and analyze , such as wind noise. This kind of environmental noise will have an impact on NVH analysis. In order to find the really "problematic" noise, we can use Audacity software to reduce noise.

The principle of Audacity software noise reduction is to select a small piece of audio that we don't want to keep and only environmental noise as a sample, and then the software will obtain the characteristics of this environmental noise, and by repeating the noise reduction function, the environmental noise in the entire audio Perform noise reduction.

It should be noted that the selected sample audio can only contain ambient noise , not the "problem" noise that customers complain about. The Audacity software download address is as follows:

https://www.audacityteam.org/download/

Noise reduction steps

The case used in this article is that on an airport bus going to the terminal, a customer recorded a suspicious noise with a mobile phone. The duration is about 40 seconds, of which there are whistling sounds at 0-11s and 27-39s, which are suspected to be produced by the gearbox differential, but no whistling sound is heard in this part of 11-27s.

Please listen to the following audio recorded directly on the bus with a mobile phone before noise reduction:

       Howling of airport bus gearbox (before noise reduction).wav (7.24 MB, Downloads: 0)

Here I suggest that you wear headphones and listen several times, because the environmental noise is very loud, so it will affect the harsh whistling sound to a certain extent. During the period of 11-27s, there were only environmental noises such as wind noise, and no harsh howling was heard.

Next, I imported the audio into Audacity software for noise reduction. First select this small piece of audio from 11-27s as a sample, as shown in Figure 1.

                                                                                            figure 1

Then in the "Effects" menu, click the "Noise Reduction" function, set the corresponding parameters in the pop-up noise reduction dialog box (I choose the default settings, and the parameter settings can be viewed in the software introduction), and then click Get Noise Features, the dialog box will automatically close. At this point, the software has completed the feature acquisition of environmental noise, as shown in Figure 2 and Figure 3.

                                 

                                                                                       figure 2

 

 

                                                                

                                                                                                   image 3

 

Then we use the shortcut key "CTRL+A" to select the entire audio segment, and then in the "Effect" menu, click "Repeat Noise Reduction", so that we have successfully reduced the ambient noise in the entire audio segment, as shown in Figure 4 and Shown in Figure 5. ( Please note that it is not to eliminate the environmental noise , which is related to the parameter setting in Figure 3. In this example, the decibel value of the environmental noise is reduced by 12dB)

                                     

                                                                                                      Figure 4

 

                                    

                                                                                                  Figure 5

 

Then export the noise-reduced audio to .wav format. Please listen to the noise-reduced audio below, we can hear the harsh howling more clearly.

Howling of gearbox of airport bus (after noise reduction).wav (7.24 MB, Downloads: 0)

NVH analysis

The NVH test in PicoDiagnostic software can import .wav format audio, the following forum posts have related introduction: NVH software imports audio files

I first import the pre-noise reduction audio, then set it as a reference waveform, and then import the noise-reduction audio, so that the two can be put together for comparative analysis. As shown in Figure 6, the purple waveform is before noise reduction, and the red waveform is after noise reduction.

                                   

                                                                                                    Figure 6

 

As can be seen from Figure 6, the overall amplitude of the waveform (red) after noise reduction is slightly lower than that before noise reduction. The waveform before noise reduction has a large decibel value due to environmental noise, and spikes appear at many frequencies . There are many suspected frequency points. The waveform after noise reduction has a very obvious spike noise at 375Hz , so it can be suspected that 375Hz is the fault frequency point.

                                     

                                                                                                 Figure 7

 

 

Then I used the filter function of the software, and selected band stop to filter out the 350Hz-400Hz sound, as shown in Figure 7. (The following forum posts have a detailed introduction to the filtering function: Sound filtering function: quickly find the frequency of the faulty sound )

Then I played the audio again, and found that I could not hear the harsh howling, which proved that 375Hz is the fault frequency . After investigation, it was found that one of the gears of the main reducer failed. The shaft rotation frequency of the gear was 8 Hz and the number of gear teeth was 47. The gear meshing frequency is 47*8=376Hz, so there is a harsh whistling sound.

You can download the following two source files to listen to:

     After the unfiltered noise - hear howling .pddata (2.75 MB, Downloads: 0)
     after filtered noise - heard howling .pddata (2.75 MB, Downloads: 0)