Civil Engineering Brownbag: Dr. Chung Song

Keith McGuffey Author

04/24/2017 Added

48 Plays

Description

Dr. Chung Song presents "Traffic Noise Reduction Using Active Noise Cancellation Technique".

Searchable Transcript

Search:

[00:00:00.862]Let me introduce Dr. Song,
[00:00:02.408]so he's an associate professor in our department.
[00:00:05.655]He was a faculty member
[00:00:06.966]at the University of Mississippi for 11 years.
[00:00:10.445]He's worked on a number of projects
[00:00:12.448]related to geo-technical issues for industry,
[00:00:18.418]he was employed by a number of companies
[00:00:21.511]prior to going to academia.
[00:00:24.567]He's involved with constructing the
[00:00:26.505]Incheon International Airport in Korea,
[00:00:28.768]as one of the main projects.
[00:00:30.583]He got his PHD for LSU,
[00:00:33.305]and a Masters from UT Austin,
[00:00:35.651]so we won't hold that against him will we?
[00:00:38.375]Okay.
[00:00:40.027]Main research focus areas include
[00:00:42.627]evaluation of physical-mechanical properties of soils,
[00:00:46.545]combining noble theories to solve real field problems,
[00:00:50.250]such as bridging nano-particulate continuum mechanics
[00:00:54.457]for soil characterization, he's implementing acoustic
[00:00:57.681]techniques in subsoil exploration,
[00:01:01.596]evaluation of geo-technical structures,
[00:01:03.908]soft soil improvement, improving constitutive models,
[00:01:06.667]and I could go on and on.
[00:01:08.401]But the acoustics component is something that is
[00:01:10.670]specifically related to this project.
[00:01:13.142]So he teaches our soils classes at the graduate
[00:01:15.735]and undergraduate level.
[00:01:17.912]And he's doing a lot of service.
[00:01:20.014]So I asked Dr Song to give this presentation,
[00:01:22.444]he's involved with what I think is a pretty cool project,
[00:01:26.928]that utilizes his acoustics techniques
[00:01:30.112]in a civil engineering domain.
[00:01:34.768]And I'll let him explain,
[00:01:36.835]but I didn't realize he was doing this until I was
[00:01:40.670]giving a tour of the labs in Lincoln
[00:01:43.394]to the current director of the Nebraska Department of Roads,
[00:01:47.673]and we go in his lab and he's explaining he's doing this
[00:01:50.236]and I'm like that's pretty cool.
[00:01:52.213]So what he's doing shows how broadly
[00:01:55.305]our expertise can reach.
[00:01:58.191]So anyways, I'll let him go.
[00:02:00.915]Again, Chung, I apologize for leaving,
[00:02:04.791]and everybody pepper him with questions, right Chung?
[00:02:10.355]All righty.
[00:02:12.256](they laugh)
[00:02:14.803]But anyway, enjoy the presentation.
[00:02:18.762]Thanks, Dan, for the generous introduction.
[00:02:23.129]Again the topic is the traffic noise reduction
[00:02:26.269]using active noise cancellation technique.
[00:02:29.486]I guess probably I can feel what you are thinking right now.
[00:02:33.856]Probably, most of you may feel like:
[00:02:37.240]why active noise cancellation by civil engineering guys?
[00:02:42.847]But you will see here, how come it's related to
[00:02:45.070]civil engineering and my specialty.
[00:02:49.603]I guess some of you are thinking,
[00:02:53.314]why the active noise canceling?
[00:02:56.526]Arrow button or next, please.
[00:02:59.449]Motivation is, actually the traffic noise may cause
[00:03:04.060]psychological discomfort,
[00:03:07.566]and time to time it can even cause physical damages
[00:03:10.894]in human life.
[00:03:13.400]Particularly in metropolitan areas.
[00:03:18.780]And usually they use these kinds of noise barriers.
[00:03:22.155]Problem of noise barriers,
[00:03:28.689]they are mostly effective for the near field condition.
[00:03:33.728]Near field condition only.
[00:03:35.869]This is the highway just right next to the noise radius.
[00:03:43.301]This one maybe the lower stairs, they can feel maybe
[00:03:47.834]some noise reduction, but of course not many of
[00:03:50.755]those residents will feel reduced the noise.
[00:03:56.349]So we need a technique that actually cancel out
[00:03:58.361]the traffic noise, not just reflecting up and out.
[00:04:03.659]And so I went into the modern
[00:04:05.871]active noise cancellation technique.
[00:04:13.721]So primary principle of active noise cancellation technique
[00:04:17.117]is very simple.
[00:04:19.252]Let's say we have a noise source.
[00:04:21.470]And from another speaker we can just supply
[00:04:24.511]so-called anti-wave.
[00:04:26.074]Anti-wave is the exact same image as this one,
[00:04:28.912]but it has different phase.
[00:04:31.250]So when you superimpose this one and this one,
[00:04:35.043]let's say this is positive amplitude vision,
[00:04:38.491]negative amplitude and they will cancel out.
[00:04:42.777]Eventually we will obtain quite low action to the wave.
[00:04:45.650]Or, maybe if everything is really ideal,
[00:04:49.180]we may not hear actually the noise.
[00:04:52.303]That's the principle.
[00:04:55.709]And in reality, the idea was there for a long time,
[00:05:00.318]but in reality it became realistic
[00:05:06.519]when Bose came out with noise-canceling headphones.
[00:05:12.639]This is the exact application of this type.
[00:05:15.812]And then, especially nowadays,
[00:05:18.485]quite good number of the vehicle manufacturers,
[00:05:21.734]Honda, Cadillac, Audi, Porsche, Mercedes-Benz.
[00:05:28.265]Some of those manufacturers advertise that hey we have
[00:05:31.428]the active noise cancellations.
[00:05:33.570]Some manufacturers they don't say anything about
[00:05:35.579]active noise cancellation technique,
[00:05:38.129]but they do incorporate noise cancellation.
[00:05:41.820]The Mercedes-Benz they don't say anything about
[00:05:43.920]noise cancellation technique,
[00:05:46.023]but they do use the noise cancellation.
[00:05:49.064]Porsche is the sports car maker.
[00:05:55.472]When you purchase Porsche you expect some real noise, right?
[00:06:00.162]So they actually use a similar technique to enhance
[00:06:03.700]the motor sound.
[00:06:05.101](they laugh)
[00:06:06.876]That's Porsche, okay?
[00:06:12.214]Especially when we talk about
[00:06:15.013]active noise cancellation techniques,
[00:06:18.502]we've got to really find out what our
[00:06:23.181]noise cancellation techniques can do for our target.
[00:06:28.800]There are some possibilities and challenges.
[00:06:31.510]A big possibility is noise cancellation technique
[00:06:34.600]is especially useful to control low frequency waves.
[00:06:41.291]Probably you hear of that the Honda,
[00:06:44.292]they have their V6 engine, but they have so-called
[00:06:48.073]cylinder de-activation technique.
[00:06:51.363]Initial model of the Honda vehicles with the
[00:06:53.955]cylinder de-activation technique did produce
[00:06:56.545]weird vibration and noise.
[00:07:00.440]Customers complained that it's just weary.
[00:07:04.830]How they solved it?
[00:07:07.094]With the active noise cancellation technique,
[00:07:09.645]and active vibration technique.
[00:07:12.685]The active vibration technique and active noise
[00:07:15.279]cancellation techniques are about the same.
[00:07:18.612]But active noise cancellation technique use the acoustic
[00:07:21.488]wave to control the acoustic wave.
[00:07:23.222]The active vibration technique use the anti-vibration
[00:07:27.241]wave to control source of vibrations.
[00:07:29.628]Just frequencies are different.
[00:07:33.537]And challenges.
[00:07:37.271]When we are talking about vehicle noise, or traffic noise,
[00:07:43.819]there is somewhere outside the highway.
[00:07:46.060]So our target field is really large-scale,
[00:07:48.853]it's not just the headphones, it's larger scale.
[00:07:52.970]And typically the traffic noises
[00:07:56.545]have multiple different frequencies.
[00:07:59.635]This, about 200 hertz is the typical tire noise.
[00:08:05.058]About 400 is a kind of engine rumble noise.
[00:08:10.023]But 500 is for tire noises.
[00:08:13.556]About 1000 to 1500 is wind noise.
[00:08:18.001]So when we say we've got to control the noise,
[00:08:21.742]we've got to control all those different frequencies.
[00:08:24.913]And sometimes it may not be very easy.
[00:08:28.239]Then also in the field, our sound or our noise can
[00:08:32.289]propagate all different directions, so 3-D conditions.
[00:08:36.686]So 1-D condition, that is very similar to the acoustic
[00:08:40.885]condition in our ear canals.
[00:08:44.377]It's very easy to control.
[00:08:45.896]That's the reason why those noise canceling headphones
[00:08:49.967]are properly working.
[00:08:51.858]In in three-dimensional conditions,
[00:08:54.163]we got to predict where we have the reflections,
[00:08:56.667]it will be reflected this way, it will be reflected this way
[00:09:00.241]or some other waves will be refracted and may arrive
[00:09:03.202]at our ears later, so we go to consider those things.
[00:09:18.688]Okay, this one is, I would say,
[00:09:21.854]active noise cancellation 101.
[00:09:24.686]It's not my material, this is the material from Kuo.
[00:09:28.262]I guess his name is as famous as Timoshenko
[00:09:32.606]in civil engineering.
[00:09:35.609]This showing us the noise source,
[00:09:37.959]it can be a fan or it can be anything.
[00:09:40.745]This side is generating the noise.
[00:09:43.379]Then we have something that can detect the source noise.
[00:09:47.073]So we use the microphone.
[00:09:49.251]So this microphone is detecting input noise.
[00:09:54.064]And this microphone will supply the signal X(n),
[00:09:58.548]this is just input signal,
[00:10:00.522]it will supply input signal to,
[00:10:02.900]and here it just said ANC controller,
[00:10:08.819]but the easier language is just DSP analyzer.
[00:10:12.193]Digital signal processor and analyzer.
[00:10:15.282]That means this guy will analyze source noise
[00:10:19.345]and try to generate anti-wave.
[00:10:22.143]Then supply to the anti-wave speaker or canceling speaker,
[00:10:27.376]then the wave propagated from here,
[00:10:30.983]and anti-wave propagated from here,
[00:10:33.378]will meet at somewhere here,
[00:10:36.418]and those two guys will try to cancel out each other.
[00:10:40.794]So when everything is ideal, usually what we can have
[00:10:44.492]is we have somewhere about 10 decibel reduction,
[00:10:47.825]somewhere here.
[00:10:49.590]10 decibel reduction is really substantial reduction.
[00:10:57.603]But in real life, this lowering is not very easy to achieve.
[00:11:01.794]There are a lot of uncertainties.
[00:11:05.156]Or technical challenges.
[00:11:07.864]One of the biggest technical challenges is,
[00:11:11.231]sound wave is recorded from here,
[00:11:13.864]we go to process it by somehow some magic,
[00:11:16.951]and when we just output the anti-wave,
[00:11:22.382]our calculations field and all those manipulations field
[00:11:25.348]should be fast enough so that we can complete
[00:11:28.714]all those things when those waves arrive here.
[00:11:34.515]So from here to there, let's say that is about 10 meters,
[00:11:38.296]about 30 feet, if there is about 30 feet,
[00:11:42.856]then we got to finish all those manipulations
[00:11:45.862]in somewhere 1/30 second.
[00:11:50.796]Just a matter of calculation time.
[00:11:52.274]Do we have really 1/30 second to do all those
[00:11:56.340]calculations and spit out the anti-wave?
[00:12:00.078]Yes, we can do that without too much problems
[00:12:02.780]with modern computers and modern DSP.
[00:12:08.286]That's the reason why actually the active noise
[00:12:11.116]cancellation technique is effective only maybe about
[00:12:14.314]five or six years ago.
[00:12:16.491]Before that, calculation speed was not up to that level.
[00:12:20.683]Even though idea was there,
[00:12:22.370]active noise cancellation technique was not
[00:12:24.392]very successful to actually control the noise.
[00:12:30.895]Then I show here another microphone called error microphone.
[00:12:35.889]This error microphone will detect the sound pressure level
[00:12:39.008]and compare the sound pressure level in here.
[00:12:43.111]But when this guy detect hey,
[00:12:49.009]your noise was not properly reduced,
[00:12:50.826]then it will give a signal to this ANC controller
[00:12:53.082]and the ANC controller will do the calculation again
[00:12:55.743]and come out with the different noise cancellation function.
[00:13:00.287]All those things, they are fractions and those things
[00:13:02.958]got to happen usually in a matter of,
[00:13:06.495]in the real life in a matter of 1/1000 second.
[00:13:10.893]Not 1000, one thousandth second.
[00:13:18.054]Okay, then with this one I guess you can understand
[00:13:21.746]why this guy just dive in to
[00:13:24.540]active noise cancellation technique.
[00:13:27.411]This one is the source wave.
[00:13:29.672]Function of the source wave.
[00:13:31.481]When we have the source wave,
[00:13:33.123]we just take a small portion of the source wave,
[00:13:36.297]and we divide it into different frequencies.
[00:13:41.932]I guess we follow the Fourier series.
[00:13:45.628]Using the Fourier function, we figured out
[00:13:48.172]first dominant frequency of the wave distribution.
[00:13:51.499]Second dominant on the wave distribution.
[00:13:53.881]So we usually pick slightly less than 10 different
[00:13:57.495]dominant frequencies.
[00:14:00.859]Then, this mathematical equation is somewhere here.
[00:14:06.954]This equation is how we can find out anti-wave to those
[00:14:12.383]dominant frequency waves.
[00:14:16.452]Those are all math.
[00:14:18.427]Then this one is the error function.
[00:14:21.789]So it'll use the error microphones and receive
[00:14:25.120]the combined signal and try to find out
[00:14:30.332]whether the error is acceptable or not,
[00:14:33.654]then when error is not acceptable,
[00:14:36.116]then it have to go back to this one,
[00:14:38.746]then through all those calculations again.
[00:14:43.016]So really the active cancellation nowadays,
[00:14:46.343]that is primarily math problem.
[00:14:50.740]Whether you have good math, good algorithm,
[00:14:53.786]or bad math or bad algorithm,
[00:14:56.008]the efficiency of your active noise cancellation system
[00:14:59.578]really depends.
[00:15:06.883]The books from Kuo, and those are my writings
[00:15:12.723]to really understand what those things means.
[00:15:20.045]Then you may wonder, hey, where we can get the
[00:15:23.583]DSP software?
[00:15:27.899]In my case, I went to
[00:15:33.811]Google and I typed active noise cancellation software
[00:15:38.948]and it took me to the MATLAB's webpage.
[00:15:44.585]The printed one in here is the MATLAB code
[00:15:51.005]from their webpage.
[00:15:53.146]Might find this rightly then,
[00:15:54.585]I tried to debug it,
[00:15:55.987]those are my memo to debug the MATLAB code.
[00:16:04.387]And the good thing about the MATLAB code,
[00:16:06.646]actually this one was my first time to use the MATLAB code,
[00:16:10.392]I never learned MATLAB code,
[00:16:11.709]but after the initial learning time,
[00:16:15.488]it was really fun to use the MATLAB.
[00:16:19.144]In here I just highlighted H for F-I-R,
[00:16:25.225]that is all hold FIR filter function.
[00:16:30.888]FIR filter is, some people are selling FIR filters.
[00:16:35.644]They are selling.
[00:16:37.289]But when you use the MATLAB you can just
[00:16:40.003]define FIR filter by those two lines.
[00:16:45.068]See this is a very good and you can actually
[00:16:46.834]change the characteristics of the FIR filter
[00:16:50.216]based on your taste.
[00:16:52.846]And there are some other things.
[00:17:01.029]I cut it out to make it short.
[00:17:04.650]But based on this one we try to code our own DSP
[00:17:10.499]for the active noise cancellation logics.
[00:17:20.290]I will come back to this one a little bit later.
[00:17:28.679]Then we try to compare performance of our filter
[00:17:32.171]and the sample filter in MATLAB.
[00:17:36.702]This is the return from the sample in the MATLAB webpage,
[00:17:43.577]so you can see here.
[00:17:46.419]I mean those waves are kind of artificially generated waves
[00:17:49.619]to mimic a kind of noisy fan.
[00:17:53.650]You can see the signal is,
[00:17:56.238]pretty high total decibel here,
[00:17:58.913]and here it is negative 20.
[00:18:02.237]Negative 20 does not mean sound pressure is negative,
[00:18:05.445]it's just a stationary scale.
[00:18:07.504]From here to arrange the range.
[00:18:10.011]And they put the, I guess, eight different
[00:18:12.519]predominant frequencies.
[00:18:14.409]So they tried to generate anti-wave for those
[00:18:18.239]eight dominant frequencies.
[00:18:22.839]Then they combined the source wave and anti-wave,
[00:18:27.105]then obtained blue ones.
[00:18:31.298]That means noise level was reduced from here
[00:18:33.935]to somewhere here.
[00:18:36.720]Somewhere here is about minus five,
[00:18:38.783]somewhere here,
[00:18:40.465]so in this figure we know that the anti-wave function
[00:18:46.218]reduced the noise level somewhere 35 decibel,
[00:18:49.502]or about 30 decibel.
[00:18:51.504]That is really, really substantial reduction.
[00:18:57.089]But that is literally a bit freaky.
[00:18:59.846]This much reduction is in the electronic domain.
[00:19:03.944]Or in the calculation domain.
[00:19:05.995]When it come out for the real world,
[00:19:09.533]then we have all those different reflections, refractions.
[00:19:13.441]And let's say if somebody in this room walked out,
[00:19:15.613]it will change the acoustic environment,
[00:19:18.278]then the actual result will be different.
[00:19:20.548]If, let's say, two people are coming in,
[00:19:23.221]then the same thing will happen,
[00:19:24.536]the acoustic environment will be different,
[00:19:26.466]then we're gonna have to change the ANC functions.
[00:19:31.854]And then for our case, this is the output from our case.
[00:19:37.311]Original noise was blue one.
[00:19:41.781]Then, we supplied anti-wave that is the red one.
[00:19:49.498]Then finally we obtained the green one,
[00:19:54.055]this is the reduced sound.
[00:19:58.494]So you can see in here,
[00:20:00.705]maybe from somewhere here to somewhere there,
[00:20:04.783]about 40 or whatever,
[00:20:06.542]but in general, including those things, in general
[00:20:09.293]in the electronic domain we could reduce
[00:20:14.069]somewhere 43 decibel.
[00:20:18.634]But for this one in the real acoustic field,
[00:20:22.372]when we measured it by the sound pressure meter,
[00:20:25.136]we could reduce about 10 decibels.
[00:20:27.880]But 10 decibels is still very effective noise reduction.
[00:20:33.925]So what I heard from NDOR is, actually the FHWA,
[00:20:39.070]they will accept whatever noise reduction technique
[00:20:42.482]that can reduce six decibel actual acoustic environment.
[00:20:48.725]If you can reduce the noise level six decibel,
[00:20:50.865]they will just take the idea.
[00:20:53.285]That means it's very difficult to reduce the noise level,
[00:20:56.451]even up to six decibel.
[00:21:01.051]Go back, back.
[00:21:03.228]So in real life, to just increase the effectiveness
[00:21:09.480]of noise reduction with active noise cancellation technique,
[00:21:15.767]the actual system will get really complicated.
[00:21:20.826]All those things, this figure is much more complicated
[00:21:24.076]than the previous diagram.
[00:21:25.922]The reason is, let's say we have the microphone.
[00:21:30.235]Microphones are not perfect, we know that, right?
[00:21:33.116]There are some good microphones,
[00:21:35.175]and not very good microphones.
[00:21:37.397]Not very good microphone means,
[00:21:39.742]it may record the signal but that signal
[00:21:43.078]may not be very clear.
[00:21:46.197]The microphone itself will just
[00:21:49.197]cancel out some of the signals
[00:21:50.884]or it may add some noise signals.
[00:21:53.970]In that case, we got to calibrate our microphone
[00:21:58.702]and we got to generate what should be the proper function
[00:22:01.120]to calibrate the microphone.
[00:22:03.319]So in here, so this one is the calibration function
[00:22:06.356]for the microphone.
[00:22:07.942]So we got to add additional functions to calibrate it.
[00:22:16.513]And this one is the amplifier - to operate the speakers
[00:22:20.548]we should have the amplifier.
[00:22:22.972]And we know that there are some good amplifiers
[00:22:25.399]and not very good amplifiers.
[00:22:26.713]What's the difference?
[00:22:28.193]Distortion.
[00:22:29.884]If some amplifiers may just change
[00:22:31.739]the tone of the sound completely, in that case
[00:22:35.143]we should have the proper correctional function.
[00:22:38.134]By the way, all those functions are called
[00:22:40.598]transform functions in active noise cancellation technique.
[00:22:46.729]So whenever we have some uncertainties,
[00:22:48.863]speaker itself, there are some good speakers
[00:22:51.532]and not very good speakers.
[00:22:54.252]We intended to generate let's say a 100 hertz wave.
[00:22:58.864]Well some speakers may generate
[00:23:00.507]100 hertz and some other noises too.
[00:23:03.508]Then we have to apply transform functions so that
[00:23:09.758]that speaker does not generate any other frequencies.
[00:23:13.696]Many different things.
[00:23:16.198]And some other things is the reflection by echo.
[00:23:20.186]So we generated the active noise cancellation signal
[00:23:23.274]by this speaker but it'll cause the reflection
[00:23:26.969]and that reflection will be fed
[00:23:30.497]back to the source microphones.
[00:23:34.781]And these source microphone should have some function
[00:23:37.774]which can filter out the reflected signals.
[00:23:40.401]So there are really many different transform functions
[00:23:44.141]to enhance the efficiency or effectiveness of
[00:23:47.559]active noise cancellation system.
[00:23:50.060]Again all those said are mathematical calculations.
[00:23:56.843]If we have, let's say, three-dimensional system,
[00:24:01.605]our sound source can be from many different places.
[00:24:05.264]In that case we got to install many different microphones
[00:24:09.656]to detect the different sound sources.
[00:24:12.450]And we got to use multiple speakers.
[00:24:15.203]The problem of using multiple speakers is
[00:24:17.584]this one will be designed to cancel out the signal
[00:24:21.158]detected by this guy,
[00:24:22.560]but we know that sound will propagate this way here,
[00:24:25.765]there and there.
[00:24:27.738]So when sound propagates to this way,
[00:24:30.042]then we need to have a function that
[00:24:34.488]turn off ...
[00:24:38.507]create another transform function take care of the
[00:24:41.797]interference of this one to there.
[00:24:45.126]Same thing from here to there we need another function.
[00:24:49.070]So for the sophisticated active noise cancellation,
[00:24:53.589]active noise cancellation system,
[00:24:55.723]we usually have more than 10 different transform functions.
[00:25:00.259]So that's the reason why there are good active noise
[00:25:03.875]cancellation headphones, and not very good ones.
[00:25:07.901]Whenever you purchase the active noise cancellation
[00:25:10.569]headphones, don't buy the cheapest one.
[00:25:17.026]Next.
[00:25:19.694]Next.
[00:25:21.222]And actually here we did quite good amount of tests
[00:25:23.807]after we developed all those different transform functions.
[00:25:28.236]We did the tests in the lab.
[00:25:30.701]We the tests outside the lab.
[00:25:33.493]I guess you guys would love to have,
[00:25:37.112]those here at work.
[00:25:38.269]Six, actually I have nine, nine sub-woofer system
[00:25:42.984]to mimic the traffic noise.
[00:25:50.383]This computer is the one, it can control
[00:25:53.552]those transform functions.
[00:25:55.321]We also tested it in a tunnel.
[00:25:58.742]The major problem of the blasting wave in the tunnel
[00:26:02.980]is blast wave has very low frequency.
[00:26:06.145]Usually get dominant frequency of the tunnel blasting wave
[00:26:10.456]is somewhere around 20 hertz or lower than 20 hertz.
[00:26:14.559]The problem is, our speakers
[00:26:17.026]cannot generate that low frequency like this.
[00:26:28.119]20 hertz wave is this one.
[00:26:30.832]Did you hear that?
[00:26:33.951]I just did a trick.
[00:26:35.437](they laugh)
[00:26:37.241]Even though the wave is there, we may feel it with our body
[00:26:41.143]but our ears may not hear anything.
[00:26:46.488]So that's the main difficulties.
[00:26:48.388]I mean if our speakers cannot produce that low wave,
[00:26:51.426]there is no way to cancel out those low frequency waves.
[00:26:59.465]And this one is a test in the so-called anechoic chamber.
[00:27:04.142]Those are designed to absorb most of the reflections.
[00:27:10.225]And to my surprise, test result from this
[00:27:14.296]anechoic chamber is not any better than the test result
[00:27:16.358]from the real environment.
[00:27:18.125]I don't know the reason why.
[00:27:21.126]But it seems like in real life,
[00:27:24.292]all those microphones and speakers work properly,
[00:27:27.827]we need some reflections and refraction.
[00:27:31.153]Next, please.
[00:27:34.401]Okay this one is some of the results.
[00:27:38.146]We used the Excel project fun way.
[00:27:41.061]200 hertz sinusoidal wave, source wave was 82.5 decibels,
[00:27:48.603]and when we applied the anti-wave,
[00:27:52.125]this one is actually a big, real reduction in
[00:27:55.696]real acoustic field.
[00:28:00.919]For all those here, single-frequency sinusoidal waves,
[00:28:04.610]we got pretty good noise reduction.
[00:28:07.978]When we combined several different hertz frequencies
[00:28:12.959]then we got quite good reduction but not as effective
[00:28:17.896]as the single-frequency waves.
[00:28:20.812]And again, for the tunnel blast wave.
[00:28:26.928]We tried to reduce it but our speakers generated
[00:28:29.274]some strange waves so it actually increased the noise level.
[00:28:34.907]Fan noise, we did not have very successful noise reduction,
[00:28:42.763]but what we know is,
[00:28:45.595]we had to try to record the fan noise,
[00:28:49.243]but I believe our recording was not very accurate.
[00:28:54.390]Again, when you record the noise,
[00:28:57.138]the microphone shouldn't be really that much microphone,
[00:29:00.835]but it seems like this one was not very good source.
[00:29:05.115]But for those things we used the signal generator.
[00:29:08.280]So we know that our way was very clean,
[00:29:11.030]200 was very clean, 550.
[00:29:16.539]So the number one rule is:
[00:29:18.554]okay, source signal should be clear.
[00:29:24.263]Next.
[00:29:26.898]They told me that I'm supposed to talk about 20 minutes,
[00:29:30.137]so it's already 30 minutes so I will just wrap up.
[00:29:34.330]Active noise cancellation technique
[00:29:36.100]for traffic noise control is challenging,
[00:29:38.893]but a promising new technique.
[00:29:42.378]It is known that California DoT is interested in ANC
[00:29:47.799]for traffic noise control.
[00:29:50.308]Actually, a couple of weeks ago, I heard that they actually
[00:29:56.299]are doing the research for using active noise cancellation
[00:29:58.725]technique to control the traffic noise.
[00:30:01.729]But they did not publish any results yet,
[00:30:04.774]I don't know the reason why.
[00:30:06.744]Anyhow.
[00:30:08.426]But to apply it we need really substantial amount
[00:30:11.507]of lab and field test.
[00:30:14.179]And I said in Omaha area.
[00:30:16.396]The reason why I put the Omaha area is this idea
[00:30:19.358]was proposed to NDOR,
[00:30:22.148]and because NDOR liked the idea,
[00:30:24.362]that they tried to persuade NCHIP.
[00:30:32.266]I felt like NCHIP was not appreciate it.
[00:30:37.119]So that's the reason why I put the Omaha area in there.
[00:30:40.859]Next please.
[00:30:44.232]Actually, again, another reason why I brought up this one
[00:30:47.357]is, again, you may still wonder why I stepped into
[00:30:50.970]the acoustics area.
[00:30:53.599]Before actually I did active noise cancellation
[00:30:56.150]research, I did the research to send acoustic wave
[00:31:00.832]through the soil to a long distance.
[00:31:06.212]The good thing about using the acoustic wave
[00:31:08.401]for soils is we can control the frequencies.
[00:31:12.183]Using the traditional seismic technique,
[00:31:15.477]the frequency is almost unlimited.
[00:31:17.610]We usually use the original frequency of the soils itself.
[00:31:21.643]But using the speakers and wave generators
[00:31:25.462]we can control the wave frequency,
[00:31:27.600]whatever frequency we want,
[00:31:29.944]but usually when you generate the opposite wave in the soil
[00:31:36.793]then when you install your microphones
[00:31:40.494]usually we call it ...
[00:31:42.753]For my case I usually ...
[00:31:47.200]Hydrophones, that is usually designed to be used
[00:31:50.769]under the water.
[00:31:53.029]Within about a meter distance,
[00:31:55.409]I could hear the wave that I sent out,
[00:31:58.287]but if I exceeded about a meter distance
[00:32:00.956]I could not hear the sound that
[00:32:02.518]I sent out through the soils.
[00:32:05.593]But anyway, anyhow, you can here measure the wave,
[00:32:09.501]measure the host wave in the soil.
[00:32:12.379]Believe it or not we can compute
[00:32:14.308]hydraulic conductivity of soils.
[00:32:18.386]It's actually more accurate than any kind
[00:32:21.669]of permeability test that we use in the lab.
[00:32:25.972]So before I did this active noise cancellation
[00:32:28.396]I did the research for using the acoustic wave
[00:32:31.933]under the ground,
[00:32:33.208]and I did really a lot of different things
[00:32:35.473]to send it to far distance.
[00:32:37.612]Another guy, somewhere in Florida,
[00:32:42.833]he used some special technique,
[00:32:45.291]so he could measure host wave that he sent out
[00:32:48.421]even miles away.
[00:32:56.069]That technique is not open to the public,
[00:32:59.187]and that technique was funded by Navy.
[00:33:03.669]So you can imagine Navy can hear the sound through the soil
[00:33:08.099]even miles away.
[00:33:10.275]Navy needed that for their sonars.
[00:33:15.360]However I'm switching to the active noise cancellation
[00:33:18.449]technique.
[00:33:19.557]I guess that's about myself and my material for the
[00:33:22.181]active noise cancellation.
[00:33:26.640]So that will actually complete my presentation.
[00:33:32.712]I have a quick question.
[00:33:34.151]Yes.
[00:33:35.049]So you said acoustics can measure soil permeability.
[00:33:38.902]How about concrete permeability?
[00:33:40.920]Can you measure concrete permeability?
[00:33:43.845]I did not try it, but I think yes.
[00:33:48.333]You can measure it.
[00:33:50.058]Because when we cause the, let's say, agitation
[00:33:55.198]in porous material, we would cause the vibration, right?
[00:34:00.631]Then what happen is, the liquid in the pore,
[00:34:04.330]and the solid materials, they vibrate with different traits.
[00:34:08.942]Because they are two different materials,
[00:34:10.582]they don't move together.
[00:34:12.512]There is the relative deformation.
[00:34:15.591]So I use the information of that relative deformation
[00:34:18.265]to compute the permeability of soils.
[00:34:22.414]As far as that is the porous material, it will work.
[00:34:35.149]So when you said that you weren't able to send
[00:34:39.084]acoustic waves a long distance in the soil,
[00:34:43.569]was it in the audible frequency range?
[00:34:47.305]'Cause I know that the reason that, that's what I think,
[00:34:53.021]the reason that those acoustic waves cannot travel that long
[00:34:58.166]may be because of attenuation.
[00:35:01.452]Because shear wave, or a rarely wave, that can travel
[00:35:04.486]pretty far compared to compressional waves.
[00:35:08.102]But what if you tried different frequencies?
[00:35:11.836]They may not be in the audible frequency range,
[00:35:14.879]but would that help to have more preservation
[00:35:20.437]of waves in long distance.
[00:35:26.863]Yes and no.
[00:35:29.101]That may work, but essentially what's happening is,
[00:35:32.466]bending coefficient of soil is much higher than
[00:35:34.924]bending coefficient in the air,
[00:35:37.386]that's the reason why the acoustic wave
[00:35:40.299]just attenuate much quicker than in the air
[00:35:43.556]or in the water.
[00:35:45.613]Actually the primary principle of the technique
[00:35:49.314]that the guys in Florida used,
[00:35:52.758]it's very similar to our cellphone technique.
[00:35:56.954]Digital cellphones.
[00:35:59.468]Digital smartphones use the digital technique, right?
[00:36:05.590]What they do is,
[00:36:08.051]in the silver.
[00:36:10.356]They supply and they send out the cellphone signal.
[00:36:17.017]The cellphone as strong as the cellphone signal,
[00:36:19.770]maybe about 20 years ago, it's not very strong.
[00:36:24.547]But what they do is, they mix special noise.
[00:36:28.374]It has two purposes.
[00:36:29.859]Special noise so that other guys does not know
[00:36:32.535]what kind of signal it has.
[00:36:34.877]Even though when other guys just record the signal,
[00:36:36.895]when they see it, they don't see anything.
[00:36:39.523]Because they mixed the artificial noise.
[00:36:42.324]But their artificial noise they call it pseudo-random noise.
[00:36:46.682]Random means random, but pseudo is the key word.
[00:36:51.045]It's not exactly the random noise.
[00:36:54.501]The guys who put the pseudo-random noise
[00:36:56.999]knows the function of the pseudo-random noise.
[00:37:00.408]So in our cellphone, it has a small chip.
[00:37:03.580]When they receive the cellphone signal,
[00:37:06.494]their signal will go through their small chip.
[00:37:09.294]The small chip will remove that cellphone signal,
[00:37:15.092]and that chip will take the noise
[00:37:17.100]that has that artificial way.
[00:37:20.354]If it does not have the pseudo-random noise,
[00:37:22.455]they don't take the signal.
[00:37:24.463]So the chip takes the one with the noise,
[00:37:27.295]and then it will remove the noise.
[00:37:32.718]That means you can have the sound
[00:37:36.465]that another guy tried to say.
[00:37:39.965]Does that make sense?
[00:37:41.897]So sender send it,
[00:37:44.036]then this guy will add pseudo-random noise,
[00:37:46.883]then receiver will receive it,
[00:37:48.696]then that chip, actually that chip is usually
[00:37:51.248]manufactured by Qualcomm.
[00:37:53.842]Qualcomm take the signal data pseudo-random noise,
[00:37:56.799]then flush out all other signals,
[00:37:59.794]then it will remove the pseudo-random signal
[00:38:02.754]from the original signal so that it can just
[00:38:07.400]pass out the original signal.
[00:38:11.346]I mean by theory it's easy,
[00:38:14.094]for real application, it's not very easy.
[00:38:17.686]But it seems like Navy figured out the technique
[00:38:20.231]by some reason.
[00:38:25.697]Other questions?
[00:38:29.311]Actually the pseudo-random signal was initially used
[00:38:34.897]for military communication.
[00:38:39.990]It was used for a long time, but some time in the 80s,
[00:38:45.207]there is a law that even though some have some
[00:38:48.536]communication information
[00:38:50.226]about opening to the general public.
[00:38:52.199]So sometime 80s that information was open to general public
[00:38:56.967]and actually Qualcomm, I don't know how they did it but,
[00:39:01.652]they just bought that technique.
[00:39:06.290]That's the reason why, I guess all of
[00:39:09.279]our cellphones, they pay some royalty to Qualcomm,
[00:39:15.081]because of their small chip that handles the
[00:39:18.372]pseudo-random signal.
[00:39:25.078]Any other questions, from the Omaha side?
[00:39:31.731]Chung, when you do this for that traffic for a highway,
[00:39:36.879]noise cancellation, do you plan to then have
[00:39:41.240]like 20 miles of speakers on the highway?
[00:39:46.271]Yeah, that is actually
[00:39:48.350]another technical challenge.
[00:39:51.218]Theoretically yes, we can handle let's say 20 miles,
[00:39:55.867]I'm not gonna say 20 miles, maybe 2 miles.
[00:39:59.123]We can handle it,
[00:40:00.976]but in that case there is an interference
[00:40:03.241]between several different speakers
[00:40:05.089]and several different microphones.
[00:40:07.924]So far actually, there are nobody try it yet.
[00:40:14.182]But usually,
[00:40:17.656]when I talk to the client,
[00:40:20.992]the target is the very noisy highway system
[00:40:25.362]somewhere in the downtown.
[00:40:27.833]Usually someone in the downtown, even in the big cities,
[00:40:33.045]usually the noise level is somewhere 80 or 85 decibels.
[00:40:39.173]They probably we can install active noise cancellation
[00:40:42.507]system only where we have real high noise level.
[00:40:48.129]Okay.
[00:40:48.962]When you do that, do you send the canceling waves
[00:40:53.478]toward the source, or do you send the canceling waves
[00:40:57.378]in the same direction as the source waves
[00:40:59.805]are being propagated?
[00:41:03.496]Based on my experience,
[00:41:05.878]whenever we install microphones close to the source,
[00:41:09.529]then we can capture the very clean wave,
[00:41:12.692]that is really important.
[00:41:16.468]I tried to capture the source wave
[00:41:20.125]somewhere let's say,
[00:41:22.095]let's say our source wave is outside the door,
[00:41:26.775]and we are trying to capture the source wave inside
[00:41:28.956]the door, it's more convenient but even though
[00:41:31.583]it was corrupted.
[00:41:34.920]But theoretically it should not make a difference,
[00:41:37.761]but technically, I mean our microphones will capture
[00:41:41.905]cleaner signal when the microphone is placed
[00:41:44.420]close to the source.
[00:41:46.889]I wasn't asking how you catch the source wave,
[00:41:49.725]but your canceling speakers, you can either direct those
[00:41:55.673]toward the source or you could direct those
[00:41:57.843]in the same direction as the source,
[00:42:00.224]both toward the receptor.
[00:42:04.918]Okay, thank you for the clarification.
[00:42:08.164]It really does not matter.
[00:42:11.900]Because when we just program the transform function,
[00:42:17.340]we program the transform function with the direction
[00:42:21.815]of the speakers.
[00:42:26.044]Does that make sense?
[00:42:28.714]Probably.
[00:42:29.944]Yeah. Usually when we have the speaker facing the audience
[00:42:34.007]it'll provide compressional wave to the audience.
[00:42:38.029]Right.
[00:42:39.547]Then when we face the speaker the other way,
[00:42:44.156]it still generates the compressional wave,
[00:42:46.666]but on the back side it generates a little bit sideways.
[00:42:52.212]So it makes a difference, but yeah
[00:42:54.714]our transform function can take care of those things.
[00:43:04.630]Any other things?
[00:43:06.062]I guess ANC is the kind of common topics.
[00:43:09.932]And I convinced you that is not completely civil.
[00:43:14.418]But the thing is, no discipline is very specialized
[00:43:19.142]in active noise cancellation.
[00:43:23.867]Can you think of any specialties out there
[00:43:27.485]dealing with active noise cancellation?
[00:43:30.325]That essentially my point,
[00:43:31.970]is nobody's specializing in active noise cancellation,
[00:43:35.138]however but the proper background for doing that
[00:43:38.887]is skills in our area.
[00:43:43.925]Again, I'm not very good in the active noise cancellation,
[00:43:47.178]but when I saw those transform functions I thought
[00:43:51.361]okay I can handle it,
[00:43:53.663]and consequence is, I could handle some of those things,
[00:43:56.209]but there are still a lot of things
[00:43:58.134]that I could not handle.
[00:43:59.907]Especially two-dimensional speaker arrangement,
[00:44:02.451]or three-dimensional speaker arrangement.
[00:44:10.009]Okay, then.
[00:44:12.481]I guess that should be it.
[00:44:16.634]Jing Kao may have some question.
[00:44:19.145](they laugh)
[00:44:21.357]You guys are taking instrumentation class, right?
[00:44:23.984](they laugh)
[00:44:25.052]Yes.
[00:44:26.202]Microphones are actually one of the sensors.
[00:44:29.291]In geo-tech we use a lot of sensors called geo-phones.
[00:44:34.263]Geo-phones are actually microphones,
[00:44:37.720]that is essentially it's the same as microphones
[00:44:39.821]but we use it to monitor vibration of soils.
[00:44:44.129]Pick-ups,
[00:44:46.015]and other sensors,
[00:44:47.496]we can call it actuator.
[00:44:50.740]Actuator is something that we supply,
[00:44:53.286]from geo-tech program for hydraulic motion.
[00:44:57.315]And it'll move something.
[00:45:00.695]Speakers are forging the vibration
[00:45:02.794]and it will generate the sound wave.
[00:45:06.172]So really, I mean, to make it simple I would say
[00:45:08.973]active noise cancellation is a problem
[00:45:11.896]of sensing, DSP analyzing and controller.
[00:45:19.867]Just one of those and it happens.
[00:45:24.266]Okay, thank you everyone.
[00:45:27.601](they applaud)

The screen size you are trying to search captions on is too small!

You can always jump over to MediaHub and check it out there.

Comments

0 Comments

Civil Engineering Brownbag: Dr. Chung Song

Description

Searchable Transcript

Comments icon comment

Related Channels

Comments