Transcript Limitations of earthquake detection with the E-puck`s

Limitations of earthquake detection
with the E-puck's accelerometer
?
Par Rodrigo De Pablo Peña et Fabian Baerenbold
Introduction


Theoretical part
–
Functioning
–
Measure limitations
–
Earthquakes
Experimental part
–
Choose a test-frequency and design a filter
–
Try to make the E-puck find a source of vibrations
Accelerometer – How it works




Accelerometer consists of 3 beams in series where the
middle one is attached to a movable mass. The 3 beams
form 2 capacitors in series.
An applied force changes position of the mass and
therefore the value of the capacitors.
The capacitors values are linked to an output voltage
between 0 and 3300[mV], which on the E-puck is a binary
code between 0 and 4096
The range of the E-puck's accelerometer is set to +/- 2[g],
g=gravity
Measure limitations and noise
• Resolution:
3300[mV ]
 0.001[ g ]
600[mV / g ]  4096
• Noise RMS in data sheet : 4.7 [mV]
5.0
 3300[mV ]  4.0[mV ]
4096
4.0[mV ]
3

amplitude: 600[mV / g ]  0.02[ g ]
• Noise RMS calculated :
• Minimum detectable
• Maximum detectable amplitude : 2[g], (up to 2.75[g], but less
accurate above 2[g])
• Sampling frequency: Up to 11[kHz]
Earthquakes
•
E-puck can recognize frequencies up to 11/2[kHz]
•
Frequencies of an earthquake between 0.1 and 20[Hz]
=> E-puck is able to measure this
•
E-puck detects earthquakes having amplitudes bigger
than 0.02[g] (=0.2[m2/s]) if it is right by the epicenter
•
0.2[m2/s] corresponds to an earthquake where “dishes in
a cupboard rattle”!
=> E-puck is not very useful in this case
Experimental part
Signal of the razor
Frequency Detection - FIR Filter
• Razor frequencies: 12-14[Hz]
• FIR equiripple bandpass filter
• Sampling frequency : 200[Hz] (to minimize the filter order)
Band: 10 -16 [Hz], Filter order: 143
Unfiltred vs. Filtred Signal
Shaking by hand (2-3[Hz]) --- Razor (13[Hz]) --- Cell phone (25[Hz])
Filter with attenuation of low
frequencies of 20dB
Filter with attenuation of low
frequencies of 40dB
Unfiltred vs. Filtred Signal
Is it possible to find the source
of the « earthquake » ?
“Bacteria” Algorithm
Mobile experiment I
•Goal: E-puck finds the source of vibrations
•Assumption: There is a continuous drop of amplitude with
distance to the source.
•How: E-puck moves with constant speed. As long as the amplitude
measured by the E-puck increases, it continues to move forward. If
not, it changes direction.
•Amplitude computation: Take the sum of the filtered signal
elements squared and divide by an appropriate constant to avoid
overflow.
•Random term: A random term is added to the constant speed to
make the E-puck able to move randomly in any direction.
“Bacteria” Algorithm
Mobile experiment I
But: Although the E-puck sometimes finds the source, it often gets stuck far away
from it….
Why ?
=>>> Idea: Scan the whole surface with the E-puck to have an idea of the
vibration pattern of it. Then repeat the experiment.
Material for experiments
Relative amplitude response
Static experiment
Relative amplitude response
Mobile experiment II
Relative amplitude response
Mobile experiment II
+
Relative amplitude response
Mobile experiment
References
•
http://www.naun.org/journals/geology/20-052.pdf
•
http://en.wikipedia.org/wiki/Japan_Meteorological_Agency_sei
smic_intensity_scale
•
http://pdf1.alldatasheet.com/datasheetpdf/view/103487/MOTOROLA/MMA7260Q.html