[Zurück]

G. Raidl:
"Hybrid Metaheuristics for Optimization Problems in Public Bike Sharing Systems";
2017.

Hybrid Metaheuristics for Optimization Problems in Public Bike Sharing
Systems

I will start with an introduction of the Algorithms and Complexity
Group of TU Wien and an overview of our research topics and recent
projects. Two of our larger research projects of the past four years
address optimization problems occurring in the setup and maintenance of
public bicycle sharing systems (BSS). This talk will give some insight
on the algorithms we developed on the one hand for planning
transportation tours for balancing a BSS and on the other hand for
deciding where to build new stations of which size in the task of
setting up a new or extending an existing BSS.

Operators of BSSs have to regularly redistribute bikes across the rental
stations in order to prevent them getting overly full or empty. This is
usually achieved with a fleet of vehicles with trailers. We will
consider hybrid PILOT, GRASP and Variable Neighborhood Search approaches
for an effective transportation tour planning.

When establishing a new BSS or extending an existing one, one of the
core questions is at which locations rental stations of which size should be
built. We model this station planning problem on the basis of a given
expected customer traveling demand over the considered geographical area and
locations where stations can potentially be built. The objective is to
maximize the actually satisfied demand under budget constraints. In order
to deal with the huge amount of input data for a larger city, we apply a
hierarchical clustering based approach. The optimization
problem is then solved by a multilevel refinement metaheuristic making use of
mixed integer linear programming and local search techniques.