Downloaded by UNIV. OF ARIZONA on March 14, 2017 | http://arc.aiaa.org | DOI: 10.2514/1.G000691
we look at how we arrived at the optimal solution using multiobjective
optimization. The optimal solution identified may not have been the best
theoretical solution. It was, however, the best of the set of solutions
proposed by the conflict-resolution model, and was therefore an effective
optimal solution that the air traffic controllers were able to apply.
Statement of the Multiobjective Problem
Each of the potential resolutions to a conflict has to balance
competing demands. On the one hand, there is a desire to minimize
fuel consumption (which is of interest to the airlines); on the other
hand, there is the wish to maximize the likelihood of the proposed
conflict-resolution maneuver being implemented by controllers.
Those solutions with a higher likelihood (of the resolution being
implemented by the controller) are the most appropriate.
Figure 10 shows all the possible temporal and vertical resolutions,
proposed by the model, for an individual conflict. The y axis indicates
the additional fuel consumed by each resolution, and the x axis gives
the percentage likelihood of the resolution being implemented by the
controller. This is labeled
“Likelihood of ATC resolution (%).” The
optimal solution is one that maximizes the value of x and minimizes
y. Figure 10 gives the potential resolutions to one specific conflict.
However, a typical flight-planning scenario will involve a number of
conflicts and, in such cases, the search for Pareto-optimal solutions
must be automated.
The first objective is to reduce the cost of the resolution, thereby
satisfying the airline. The second objective is to solve the conflict by
using, as far as possible, the maneuvers normally applied by the
controller. In this way, we will get a more homogeneous flow
Chia sẻ với bạn bè của bạn: