By Mirza Adrian NP*
ABSTRACT
Pavement structural evaluation plays an important role in road
maintenance. One method to perform pavement structural evaluation is by performing
Non-Destructive Test, such as Falling Weight Deflectometer, on pavement
structures and performs backcalculation
to calculate the pavement’s layer moduli using the acquired FWD deflection
data. One approach to perform backcalculation is by using a database of
pavement deflection. Using this approach, the calculation of pavement moduli is
done by interpolating between enveloping moduli saved within the database. Thus,
the accuracy of the results from this approach greatly depends on the data and
the method of interpolation. Also, the problem of backcalculation is a complex
multi-dimensional problem that has many local minima within the solution space.
To avoid that, Genetic Algorithm is used because of its ability to perform
global search within the solution domain.
This
report studies the database approach for backcalculation and found that backcalculation
can be done accurately using
a database containing 37.240 pavement deflection data. This database is
acquired from combining 49 asphalt moduli (E1), 40 sub-base moduli
(E2), and 19 subgrade moduli (E3). This report also found
that the semi-logarithmic method of interpolation is a better method to
interpolate pavement deflection compared to the linear method of interpolation.
The database and the
interpolation method are then used to develop Genetic Algorithm based software
to perform backcalculation on pavement deflection. Comparison with programme GA
BackCalc for 27 real deflection data shows that the results compares favourably
well even though the calculation time is quite lengthy.
Keywords: genetic algorithm,
backcalculation, pavement evaluation, database
* This is an abstract for my undergraduate thesis. If you would like to see the complete text, you can contact me through my email mirza.adrian@gmail.com
No comments:
Post a Comment