| Summary: | 碩士 === 國立成功大學 === 交通管理(科學)學系 === 81 === Generally speaking,the TRANSYT model is one of the most
important and popular network signal timing design models in
the world. It comprises a traffic flow simulation model and
another signal timing optimization model.The former one is to
macroscopically simulate the plotoon progression behavior in
the road segment. The latter one,then,is used to find the
optimal signal timing pattern. Since 1967,the traffic flow
simulation model has been revised several times so that it can
sa- tisfactorily interprete the real-world flow phenomena. But
a so called "Hill-Climbing Method" has been employed in its
timing optimization model for 25 years and can only obtain a
local optimum timing solution. The Genetic Algorithm is
currently developing to act as a new optimization method that
can find the global optimum timing solution.Based on this
understanding,a new network timing design model,i.e. TRANSYT-
GA, is proposed to improve the performance of TRANSYT-7F Model
which is a widely used version in the TRANSYT models family,
All the important features which are derived from the Genetic
Algorithm,i.e.Crossover Method、Rate of Match、 Preventing
Incest、Probability of Mutation.Fitness and K value of the
performance index,are carefully defined in this thesis study.
For the simulation results with no-limited computation time
cases,the performance improvements of TRANSYT-GA than TRANSYT-7
F are between 0.62% and 17.6% on the artificial networks. And
for the actual network cases,it is betwween 3.29% and 15.61%.
In the other studies of fixed-time simulation,evidances show
that TRANSYT-GA is better than TRANSYT-7F around -7.25% to
16.46% in pseudo network cases and -5.87% to 9.34% in the
actual network cases.
|
|---|
