10-06-2013, 03:52 PM
Improved Ant Colony Optimization for Multiobjective Route Planning of Dangerous Goods
Improved Ant Colony.pdf (Size: 287.84 KB / Downloads: 37)
Abstract
Dangerous goods (DGs) can significantly affect the
human and nature if they are exposed to the environment
without any protection. This situation is likely to occur when
accidents happen during the transportation process. Especially in
large cities, due to high population density and complex traffic
network, the transportation of GDs has to pass through densely
populated areas or other sensitive districts. So only considering
one traditional objective in routing planning, such as the shortest
length of route or lowest cost, can no longer meet our needs.
There is an urgent need to review and improve the way of route
optimization for DGs transportation. This paper develops a
multi-objective model for the determination of optimal routes. In
this model, three conflicting objectives are considered. They are
total travelling time, accident probability and population
exposure risk. For settling this model, an improved ant colony
optimization (ACO) is introduced with a novel multi-objective
decision method named MAXMIN. With the support of
geographical information system (GIS), a case study of Hong
Kong is carried out for the transportation of DGs. The
experimental results show the proposed approach is feasible and
effective.
INTRODUCTION
Dangerous goods (DGs) are a set of materials which are
hazardous to the environment, especially to the health of people
that are within a certain distance of them. Besides storage, the
most dangerous thing is the transportation of DGs. In this
process, DGs will be exposed to public for a long time and face
unexpected cases on the road. All of these increase the
population risk and endanger the public safety. So no matter in
which country or city, the safe transportation of DGs is a
subject of considerable interest. According to the research
conducted by Zhijun Chen et al. [1], the main factors of road
transportation system for DGs include DGs itself, the vehicles,
the behavior of relevant people, the situation of environment
and road, and safety supervision and emergency rescue. In
terms of objective factors, it is difficult to change. But for
subjective factors, besides to require drivers to be careful in
driving process, another thing we can do is selecting a feasible
route in order to minimize damages in case of accident.
MAXMIN method
For MOPs, all objectives usually are not able to achieve
their optimal values at the same time. The improvement in one
objective is often made at the expense of others [10].
Therefore, to evaluate solution’s fitness becomes the key to
solve MOPs. The most commonly used method is to present
relative importance as weighting factors for these objectives,
and to produce a single value to evaluate solutions. However,
although this method is easy to use, one of the biggest
drawbacks is that numerically quantifying the weights is a
difficult task. Other evaluation methods, such as multiobjective
and compromise programming approaches [9], have
their own drawbacks as well. In 1999, Balling et al. [11]
presented the MAXMIN fitness and introduced a method for
computing it.
CASE STUDY
Hong Kong is a large city with high population density and
narrow streets (especially on the island). Due to the land
constraints, vehicles carrying DGs inevitably have to pass
through densely populated areas or their vicinities. In an
attempt to ensure public and environmental safety, DGs
transportation is strictly controlled by the laws of HK. The
government also has issued rules and regulations for DGs
transportation. Each DG transport company is required to
advise their drivers to follow major routes and to avoid heavy
traffic and densely populated areas as much as possible. As a
result, routing selecting becomes an important task before
transportation.
In this paper, the proposed optimization model and
modified ACS integrated with MAXMIN method will be
applied to DGs route planning on the HK island. In addition,
we also use GIS to display and analyze the optimization results.
With the support of GIS, the optimal routes will be selected
effectively [20]. At the beginning, there are some explanations
to illustrate. Firstly, all date used in this experiment are not
actual values, but normalized ones from actual values. The
reason is that various objectives have different units. This is not
helpful to calculate and compare results together. The
computation process of three objectives in proposed model has
explained in Li’s paper [9]. Secondly, it is necessary to
determine the parameters of ACS. One of them is the number
of ants.
CONCLUSION
In this paper, we have proposed a multi-objective,
constrained optimization model for DGs route planning, and
applied an improved ACS to settle it. Because route planning
for DG transportation is a typical kind of MOP, the MAXMIN
method is used to assist ACS to judge which candidate solution
is better. With the support of GIS, decision makers are allowed
to choose optimal routes visually.