IEEE Computational Intelligence Magazine - May 2022 - 88

of charging stations to minimize investment
for constructing charging stations
and to maximize the service quality.
Sadeghi-Barzani et al. [42] optimized the
location and charging capacity of fastcharging
stations, where the objective
included investment, grid loss, and traveling
cost of EVs to reach fast-charging
stations. The above three studies [40]-
[42] all used GA to solve their corresponding
problem. Ko et al. [43] considered
a special charging facility for
electric public buses. These charging
facilities support wireless power transmission
and are deployed underground
along a bus route; a bus can thus be
charged while driving. The deployment
of such charging facilities must be able to
maintain the buses' battery level within a
normal range. A PSO-based algorithm is
proposed to determine the location of
charging facilities and the battery capacity
of buses to optimize the overall cost
of such a public bus charging system.
In addition to the infrastructure
deployment mentioned above, certain
studies have focused on developing
charging strategies from the citizen perspective
based on the existing charging
infrastructure. Moghaddam et al. [44]
and Liu et al. [45] investigated the selection
of charging stations and options,
and planned the route for each EV to
reach the corresponding charging station.
Moghaddam et al. [44] proposed an
ACO-based algorithm to optimize the
traveling time, queueing time, and
charging cost of EVs, while Liu et al.
[45] proposed a DE-based algorithm to
optimize the traveling time, charging
cost, and difference between the real
battery state and the expected battery
state. Unlike refueling petrol vehicles
that only require several minutes, the
charging of EVs may require much
more time (e.g., several hours). To
reduce the time cost, battery swapping is
a promising charging strategy. In battery
swapping, an EV unloads the nearempty
battery and replaces a full battery
from the exchanging station [46]. The
near-empty battery is then charged in
the exchanging station and can be provided
for the other EVs. Kang et al. [47]
considered battery swapping from the
perspective of exchanging stations. By
using a hybrid algorithm of PSO and
GA to schedule the charging time and
location of the batteries, the charging
cost, power loss, and voltage deviation
are optimized. Wu et al. [48] investigated
the selection of charging options for the
batteries in the exchanging station. Different
options have different features in
terms of charging speed, charging cost,
and battery damage. The proposed algorithm
is also a hybrid algorithm of PSO
and GA.
For energy companies, Yang et al.
[49] built a probabilistic model to
describe an EV's charging load on a
power grid, in which an ACO-based
algorithm was proposed for parameter
identification. Amini et al. [50] used GA
and PSO to allocate distributed renewable
resources to parking lots to minimize
power grid loss. Rahmani-Andebili
et al. [51] simultaneously investigated
two optimization problems: deployment
optimization and charging scheduling.
The first optimization problem refers to
the selection of parking lot locations,
which aims to optimize the initial
investment and operation cost over the
next 30 years. The second optimization
problem aims to schedule the charging
time of EVs to avoid using expensive
electricity generation units to maximize
the profit of energy companies. A simulated
annealing algorithm is used to
solve the first optimization problem,
while GA is used to solve the second.
Wind farms produce power from wind
energy, which is clean and renewable.
Both wind power and EVs are considered
environmentally friendly. Some
researchers have thus focused on the
economic dispatch problem involving
the charging of EVs and wind power,
which aimed to determine the optimal
electricity generation output to meet
the load demand. Zhao et al. [52] used
PSO to minimize the electricity generation
cost of wind farms. Qiao et al. [53]
proposed a nondominated sorting-based
DE to solve a multiobjective model that
minimized the electricity generation
cost and pollutant emissions simultaneously.
In this multiobjective model, thermal
power is also considered to cooper88
IEEE COMPUTATIONAL INTELLIGENCE MAGAZINE | MAY 2022
ate with wind power to meet the load
demand, and the pollutant emissions are
caused by the usage of thermal power.
D. Intelligent Metro Transportation
An increasing number of cities, particularly
large cities, have built metro
systems. Metro transportation is a convenient
and efficient choice for commuting
because it rarely has delays. It
can also reduce the traffic congestion on
roads. There are mainly two research
topics within intelligent metro transportation:
timetable design and speed control.
These topics are related to the
expenditure of metro corporations and
the travel convenience for citizens; thus,
they are considered from both the business
and citizen perspectives.
First, timetable design includes the
design of arrival and departure times of
trains in each station along their route.
The most widely considered optimization
objective is to minimize the waiting
time of transfer passengers [54]. Zhong
et al. [55] proposed a dual-population
DE, in which two populations focused
on search diversity and convergence,
respectively. Hassannayebi et al. [56] used
PSO and proposed an adaptive parameter
control strategy to enhance search
efficiency. In addition to the waiting
time of transfer passengers, Nitisiri et al.
[57] further considered the operating
cost of metro corporations and proposed
a parallel multiobjective GA to optimize
these two objectives. Certain studies
focused on the timetable design for
reusing energy. In detail, when a train
brakes, the generated kinetic energy can
be reused by another train in the same
substation concurrently for acceleration.
Lin et al. [58] used GA to schedule the
dwell time of a train in each station for
efficient reuse of the regenerative energy
from braking. However, the classic version
of GA is used and there is no specific
strategy to enhance the algorithm's
performance. Liu et al. [59] proposed an
artificial bee colony-based algorithm to
schedule the headway and dwell time of
trains for optimizing the utilization of
regenerative energy.
Second, speed control handles
the operational status of trains. The
IEEE Computational Intelligence Magazine - May 2022

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