International Core Journal of Engineering 2020-26 | Page 158

T ABLE V. C LUSTER A NALYSIS
First, we standardize the above four attributes, then
multiply the weights and add them to get the corresponding
values. After sorting these values, we can know the priority
level of these five points. Through the cluster analysis
method, after analyzing the distance between these five
target points and the maximum flight distance of the drone
full load, it can be known that the DL1 point is a single type,
DL2, DL3, and DL4 belong to one type, and DL5 is a type.
Combined with the map, we can see that this classification
is very reasonable, so we need to establish Cargo container
points in DL1 and DL5.Then the question becomes how to
choose a point to position Cargo Container in the remaining
three points DL2, DL3, and DL4. No matter at which point
the Cargo Container can be used to complete the most basic
medical package transportation, we only need to follow the
priority order obtained in the beginning, hence we select
DL3 as the final location.
a crucial position. We define the target functions based on
these:
Then we consider the restrictions:
1) The drone should arrive the destination;
2) The delivery items must meet the requirements of the
delivery location;
Based on these, we calculate the max distance the drone
can fly (Distance = speed * time) as a new distribution and
give different weight for each distribution.
The locations of position 1 and position 5 are too far
from the others. So the drones in them just do the job of
surveying road conditions, and based on these, we tend to
choose the drones with the greatest endurance. Drone B will
be the best choice. Then we should consider the “DroneGo”
in position 3. In this situation, we have to consider the
optimal solution of target function. We use the idea of
genetic algorithm to analyze the problem step by step
Combing the transport capacity of the drone, and we have
the final results: DroneG, DroneB, DroneF will perform
great, listing as follows:
IV. D RONE F LIGHT P LAN
Based on the best locations we selected and the scores
drones have got, we make a shipping route and flight plan.
We can infer that in position 1 and position 5 we don’t need
to consider the delivery mission, thus the distance will be
the most important thing we will consider. Before starting,
we should consider the target and the restrictions. For the
delivery location, waiting time is the most important thing.
And for the “DroneGo” system, the amount of drones takes
T ABLE VI.D RONE F LIGHT P LAN
B1 B2 F2 G2 B3
6:00~7:00 Road survey Delivery to DL2 Delivery to DL2 Delivery to DL4 Road survey
7:00~8:00 Charge Charge Charge Charge Charge
8:00~9:00 Charge Charge Charge Charge Charge
10:00~11:00 Road survey Road survey Road survey Road survey Road survey
Charge
11:00~12:00 Charge Charge Charge Charge 12:00~13:00 Charge Charge Charge Charge Charge
13:00~14:00 Road survey Road survey Road survey Road survey Road survey
14:00~15:00 Charge Charge Charge Charge Charge
15:00~16:00 Charge Charge Charge Charge Charge
16:00~17:00 Road survey Back to DL3 Back to DL3 Back to DL3 Road survey
17:00~17:20 Back to DL1 ˉ ˉ ˉ Back to DL5
Note: H-type drones are used to establish emergency communication network platform, and will be positioned in
LD1,LD3,LD5, working for all day long without interruption
maximum load, the cost and the capacity of the drone. The
fuzzy weights during calculation have a great impact on the
model. To reduce this effect, the PSO algorithm is used to
correct the weight matrix. The PSO algorithm searches for
the corresponding weight based on the existing information
V. C ONCLUSION
For the design of the "Dronego" system, an Fuzzy
AHP-based evaluation model is used to score the
performance of each drone under three criteria of the
136