IEEE Computational Intelligence Magazine - May 2022 - 91

models that set the primary and secondary
objectives are formulated in [92],
[94], and [95]. In [93], [97], and [98],
multiobjective models are directly formulated,
and nondominated sorting
GA-based algorithms are proposed to
obtain a set of Pareto optimal solutions.
The number of privately owned cars
has grown rapidly in recent years. By the
end of 2019, the number of privately
owned cars is projected to exceed 200
million in China [99]. This large number
of cars leads to difficulties in finding
parking space. Typically, in the central
business district of a city, the supply of
parking spaces cannot meet demand.
The deployment of smart parking systems
is a promising way to relieve the
parking difficulty, which has been
widely studied in recent years [100].
Parking reservations are an important
part of smart parking systems [101],
which allow drivers to reserve a parking
space before arriving at a parking lot.
However,
if all parking spaces are
reserved, other drivers on the road cannot
park in parking lots. Therefore, some
parking spaces should be made available
only to non-reservation drivers. Mei
et al. [102] used GA to optimize the
proportion of parking spaces for reservation
in parking lots. Two optimization
objectives are considered: 1) social benefits,
which include travel time, travel distance,
and queuing time of drivers for
parking; and 2) parking lot revenue,
which primarily refers to parking fees
paid by drivers. Liu et al. [103] introduced
a parking guidance system. In
their proposed system, a wavelet neural
network is used to predict the available
parking spaces for drivers, and PSO is
used to optimize the network configurations.
The selection of parking lots for
users is based on a multivariate logit
model, and an adaptive GA is used to
provide a route with a short distance
and time for drivers to reach the
selected parking lot.
G. Intelligent Trip Recommendation
With economic development and
increasing income levels, people increasingly
travel for vacations. A tourism city
always has various trip items (e.g., tourist
attractions and entertainment activities).
However, tourists typically have
limited time and budgets. How to efficiently
plan the itinerary of a trip to
obtain the best experience is the most
important concern of tourists, which
creates a large demand for trip recommendation
services [104]. Thus, this
application is considered from the citizen
perspective. A general trip recommendation
scheme includes the
following three parts: a set of trip items
to be visited, the visit sequence of the
trip items, and the traveling route/vehicle
between two trip items.
Luan et al. [105] proposed an ACObased
algorithm for trip recommendation.
The solution from this algorithm is
constructed by adding trip items one by
one until all items are included or the
user's time or budget has been exceeded.
The selection of trip items is based on
two factors: trip relevance and trip
diversity. Trip relevance represents the
interest of a tourist in a kind of trip item
or a specific trip item, which is obtained
based on users' scores. However, if only
trip relevance is considered, the trip recommendation
scheme may include too
many similar trip items, which is not
what tourists expect. To address this
problem, trip diversity is incorporated,
which represents the difference level
among the selected trip items. Huang
et al. [106] also considered trip diversity.
In addition, the rating of trip items and
transportation time are further considered.
Since some tourists do not have
specific interests, they may prefer highrated
trip items (e.g., must-see attractions).
In addition, with limited time,
tourists prefer to minimize transportation
time. A niching GA is proposed to
generate multiple candidate itineraries
for users. Migliorini et al. [107] considered
crowding. For example, during a
public holiday, many tourists will flock
to tourism cities, creating dense crowds
at tourist attractions. Dense crowds have
a poor impact on the trip experience
and induce longer visiting time at
attractions. Thus, the balance of tourists
among the attractions is incorporated
into the trip recommendation. A simulated
annealing algorithm is proposed to
solve this problem. Dotoli et al. [108]
focused on planning vehicles during the
itinerary, including bikes, buses, metros,
and cars. Three optimization objectives
are considered: cost, time, and gas emissions.
GA is used to solve this problem,
in which the fitness of the chromosome
is set to the weighted sum of the three
optimization objectives above.
IV. EC for Intelligent Air
Transportation
Air transportation is the most modern
mode of transportation. For the construction
of smart cities, the transportation
issues related to the airport are
essential, especially for those cities with
airline hubs. The application of EC algorithms
for intelligent air transportation is
classified into the following three categories:
landing scheduling, airline management,
and route design. These three
categories are introduced as follows.
A. Intelligent Landing Scheduling
Air traffic is typically busy in large cities.
For example, the total traveler throughput
of the Beijing Capital International
Airport reached more than 100 million
in 2019 [109]. However, the limited
number of runways in an airport may
not meet the demand of take-off and
landing (i.e., an airplane may not take
off or land immediately when it is
ready). Constructing new runways is a
potential solution but is expensive and
may not be possible due to insufficient
free space. An alternative solution is
scheduling the take-off/landing
sequence of airplanes to reduce delays.
Most studies have focused on landing
scheduling. However, the take-off and
landing of airplanes both refer to the
occupation of runways so these studies
are also applicable to hybrid take-off and
landing scenarios. Landing scheduling is
related to the service quality of airlines
from the business perspective.
A simple example of landing scheduling
on a runway is shown in Fig. 5.
The arrival time of airplanes is typically
known in advance based on the timetable
of the airline. The decision variable is
the landing sequence of airplanes. An
important constraint is that the landing
MAY 2022 | IEEE COMPUTATIONAL INTELLIGENCE MAGAZINE 91
IEEE Computational Intelligence Magazine - May 2022

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