# Learning and Teaching Back

This week was very interesting because I had the opportunity to learn amazing things as well as teaching my knowledge in acoustics to younger students! Let’s review this together!

# Learning

## Principal Component Analysis (PCA)

The first thing I learnt was the PCA technique. It is a bit complicated, so I will use an example to explain this. At school, we have a pool of 84 sound pressure signals recorded on an engine. Some of these signals are normal, some of them contain explicit clinkings, and some of them contain little clinkings.

Our first goal was to code an algorithm that can learn when a signal contain jinglings or not. To achieve that, we had to work on half of the pool and identify key characteristics, such as the energy of the signal, its maximum, or a specific pattern on the scalogram (obtained after a Continuous Wavelet Transform). Then, we plot the results on a graph and we determine a rule. For example : All the points of the plane that are in a specific circle correspond to a signal that has clinkings.

Our second role was to use the second half of the pool to verify that our rule was correct. And in our case, it was! We were very proud to see that all the signals with clinkings were represented in the circle that we defined in the first phase! And we were also very proud to be the second best group of the session! I personnally enjoyed implementing this technique to real signals and improving my MATLAB skills!

The PCA technique was very helpful in this case to understand how the get the most from an engine.

## Near-Fild Acoustic Holography (NAH)

On Friday, we worked on near-field acoustic holography! The idea is to measure a sound pressure signal about 4 inches away from different sound sources using an array of 16 microphones. In our case, we had 1 source that was emitting a sine wave, and the other 2 were emitting white noise. The idea is to reconstruct the sound pressure field on the surface of the speakers with the sound pressure field measured 4 inches away.

Our teacher already had implemented an algorithm that was plotting the results on a “sound map” (which is an image with colors that vary as a function of the sound pressure). However, the algorithm wasn’t perfect and our role was to implement a series of filters to improve the results. The most known ones are Veronesi Filter, Modified Veronesi Filter, or Tikhonov Filter. In all, there were 4 filters to implement.

The first one was quite easy to implement and was giving correct results. The second one was a bit trickier but we managed to do it. The third one was very technical, but that was nothing compared to the last one! At the end of the course, I was one of the students who had implemented most of the filters, so I wanted to do the last one! So I stayed with my teacher for about half an hour after the course and we finally did it! It works fine and I am very happy to be the only student of the group that managed to implement everything!

Now I will have to write reports for these two, and that won’t be an easy task!

An example of Near-Field Acoustic Holography. At a specific frequency, we can see that the middle of this system emits the loudest noises, and the top right zone emits almost nothing.

# Teaching back

On Saturday, I wore a new hat. Now that I am almost at the end of my studies, some students from my old high school that are interested in acoustics have a lot of questions to ask me. In particular, there is one student who has to do a school presentation and he chose to study how the vibrations from an engine are trasmitted through the chassis.

It is a very technical subject and I am not a professionnal in automotive, but I have some knowledge about vibration of beams and I think he has to know how to solve simple systems before studying complex ones. So I stayed two hours with him to show him the equations, where they come from, what they mean, and how to solve them.

It was sometimes difficult because on my side, I almost know these equations by heart now, but I forgot the math demonstration behind it! And on his side, he doesn’t know a lot about differential equations to solve them easily. However, he knew everything about cars! Therefore, it was very interesting to talk with him because he had a practical sense of how an engine acts on a chassis and I helped him to mathematically modelize this.

That was a very nice moment where I was in the shoes of a teacher and I will probably do this again a lot when I am on the other side (read when I graduate)!

equations

That’s all for this long post, but this week, I really did amazing things and I wanted to share this with you. Next week should be interesting as well, but that will be for the next post!

Don't be shellfish...