20Multi-Directional Comprehensive Disaster Response System Based on Optimiza...

杨利霞

Multi-Directional Comprehensive Disaster ResponseSystem Based on Optimization

According to the actual situation of Puerto Rico, we designed a disaster response system
from the perspective of disaster area demand, company cost, realizability and security.
First, we identified the number and type of UAVs(unmanned aerial vehicle) in the UAV
fleet based on the geographical location and needs of Puerto Rican hospitals. Minimum UAVs
are used to save costs. Solving this optimization problem, we get two schemes: scheme one
needs four UAVs (B, C, D and H), the number of which is 1B, 1C, 1D and 3H; scheme two
needs four UAVs (B, C, G and H), the number of which is 2B, 1C, 1G and 3H. Each scheme
needs three containers.
Second, we designed the packaging configuration for containers. The number of medical
packages is large, so the heuristic algorithm is not effective. We propose a one-dimensional
maximum utilization packing scheme of “medical package first, UAV later”. It can not only
realize the greater use of container space, but also be easy to achieve when loading containers.
The maximum space utilization rate is 93.22% and the minimum utilization rate is 68.14%.
Third, we gridded the main roads in Puerto Rico's main disaster areas and transformed
the continuous problems into discrete ones. We identified the optimal location of the disaster
response system by using grid search method. The three containers’ locations are as follows:
,
,
.
Fourth, the payload packaging configuration of UAV is designed by using optimization
methods. Drone B load 2MED1,drone C load 1MED1+1MED3, drone D load 4MED1+2MED3
or 3MED1+3MED2 or 2 MED1+1MED2+2MED3. UAV flight delivery routes need to avoid
mountain and high buildings, so we use Voronio Diagram and Dijkstra algorithm to get
delivery route. The flight schedule of UAV is obtained according to the delivery route.
Fifth, in order to make the UAV reconnaissance the road as wide as possible, flight
schedule of the UAV are obtained by using ant colony optimization (ACO). It can use the
limited flight time to reconnoitre the road as much as possible.
To sum up, we considered many factors to design DroneGo system.
Keywords: Optimization; ACO; Gridding; Voronio; Dijkstra