IEEE Computational Intelligence Magazine - May 2022 - 95

Ho-Huu et al. [148] constructed a multiobjective
model that considered noise
impact and fuel consumption simultaneously
and proposed a two-stage
approach to solve the model. In the first
stage, MOEA/D [149] is used to determine
a set of Pareto optimal routes.
Then, in the second stage, NSGA-II is
used to assign airplanes to the routes
obtained in the first stage.
In addition, the flight route of airplanes
should remain away from dangers
such as poor weather, crowding traffic,
and no-fly zones. González-Arribas et al.
[150] constructed a dynamic model to
represent thunderstorms in airplane
route design. Han et al. [151] planned
the routes of multiple airplanes simultaneously.
The flight speed of each airplane
is assumed to be the same. A DEbased
algorithm is proposed to adjust
the flight direction of each airplane in
each time window, guaranteeing that the
minimal distance between any two airplanes
will not be smaller than a certain
threshold to avoid collisions. Wang [152]
proposed a hybrid algorithm of the artificial
immune algorithm and ACO to
design a route with minimal distance
while avoiding no-fly zones.
V. EC for Intelligent Sea
Transportation
For coastal and port cities, sea transportation
is a significant part of the intelligent
transportation in smart cities. The
application of EC algorithms for intelligent
sea transportation is classified into
the following three categories: berth
allocation, path planning, and container
stowage. These three categories all
considered the profit and service quality
of either shipping companies or
port operators, and are thus considered
from the business perspective. Details
of these three categories are introduced
as follows.
A. Intelligent Berth Allocation
A number of ships arrive at a given port
every day. Containers must be loaded or
unloaded by quay cranes in the berths of
the port, which typically requires dozens
of hours. Therefore, due to the limited
number of berths in the port, arriving
ships may not be able to dock at the
berths for container loading/unloading
immediately. This situation could benefit
from an intelligent berth allocation
approach that reduces ship waiting time
and avoids delays of subsequent trips
[153]. Berth allocation includes assigning
ships to the berths and determining
the handling sequence of each berth. In
the literature, certain exact algorithms
(e. g., branch-and-cut [154] and branchand-price
[155]) have been applied to
the berth allocation and have obtained
great performance on small- and medium-scale
instances. However, they may
suffer from poor time efficiency in solving
large-scale instances. In comparison,
EC algorithms can obtain optimal or
near-optimal solutions on instances of
various scales within an acceptable execution
time. Ting et al. [156] used PSO
to optimize the time-in-port, which
includes the waiting and handling times
of all ships. Cheong and Tan [157] constructed
a multiobjective model where
waiting and handling times were set as
the two optimization objectives. A
multi-colony ACO is proposed to solve
the model. S¸ahin and Kuvvetli [158]
used DE to minimize the penalty of
delay and non-optimal berthing position.
Hu [159] used NSGA-II to minimize
delays and night work. Night work
is not preferred by employees and is also
less efficient than daytime work.
Heterogeneous berths are considered
in [160] and [161]. The difference
among berths is described in size (i.e.,
length and width) and water depth.
Large ships cannot dock at a small berth,
while several small ships may dock at the
same large berth and be handled simultaneously.
Tsai et al. [160] used GA to
minimize total waiting time. Ji et al.
[161] considered the total time-in-port
as the optimization objective, and the
constraints of the problem are considered
to be the other optimization objective.
Thus, a multiobjective model is
constructed. NSGA-II is used to solve
this model.
In addition, some studies created
more detailed berth allocations by considering
additional resources in ports
(e.g., quay cranes and yard trucks). Quay
cranes can be flexibly allocated to the
berths based on the handling workload
[162]. Hsu [163] proposed an improved
PSO algorithm to solve berth allocation
and quay crane assignment simultaneously.
De et al. [164] used a chemical
reaction optimization algorithm to optimize
the time-in-port of ships and the
additional rental cost of quay cranes.
Liang et al. [165] considered two
objectives. One is the total time-inport
of ships, and the other is the
movements of quay cranes. Specifically,
if two consecutive ships at the same
berth require different numbers of quay
cranes, some quay cranes must be
moved. The proposed algorithm is a
multiobjective GA. Yard trucks are used
to transfer containers between the
berth and warehouse. Li et al. [166]
considered the time-in-port of ships
and the transfer distance between the
berth and warehouse. A PSO-based
algorithm is proposed to solve this
multiobjective optimization problem.
B. Intelligent Path Planning
When planning the navigation path of
ships, feasibility, safety, and economic
benefit are three critical issues. Feasibility
issues refer to the avoidance of various
obstacles, such as reefs and shallow
seas (i.e., where ships easily run aground).
Liang et al. [167] used ACO to
construct a feasible path with minimized
length.
Safety issues refer to the handling of
weather influences and the control of
multi-ship traffic. Maki et al. [168] considered
the influence of winds and
waves, and used GA to optimize fuel
cost and path safety. Lin et al. [169] proposed
a ship weather routing system that
consisted of four modules for ship
motion, ocean environment, navigation,
and routing optimization. PSO is used
in the routing optimization module to
optimize fuel consumption and travel
time. For the control of multi-ship traffic,
researchers have focused on planning
the path of multiple ships in the same
area for maintaining a safe distance
between ships and avoiding collisions
[170]. Zlapczynski and Szlapczynska
[171] used GA to generate paths, where
MAY 2022 | IEEE COMPUTATIONAL INTELLIGENCE MAGAZINE 95
IEEE Computational Intelligence Magazine - May 2022

Table of Contents for the Digital Edition of IEEE Computational Intelligence Magazine - May 2022
