Benchmark
Problems for
H. Kim, B. Kim, and D.Noh
¡°The Multi-profit Orienteering
Problem,¡±
Computers and Industrial Engineering,
2020. https://doi.org/10.1016/j.cie.2020.106808.
1.
Introduction
The orienteering problem (OP) has recently
drawn the attention of many researchers. It is also known as the selective
traveling salesman problem (STSP), in which a vehicle maximises the profits
collected from customers within some threshold (Gendreau et al., 1998; Laporte
and Martello, 1990). In this study, we introduce the multi-profit orienteering
problem (MPOP), which is another variant of the OP.
In the MPOP, every customer vertex has multiple
profits, and the profit is determined when the vehicle visits a customer
vertex. We assume that there are multiple pre-defined time slots for each
vertex and that the profits are assigned to these time slots. Hence, the
vertices to visit, the visiting times of the chosen vertices, and the routing
order of these vertices must be determined to maximise the collected profits of
the MPOP.
The MPOP is motivated by the concert promotion
problem. A manager who wants to advertise a performer¡¯s concert has to
determine the schedule for visiting some sites and promote the concert to
people near that site. Because the moving population of a given region varies
during the day, the promotion efficiency depends on the visiting time. Election
campaigns can be an application area of the MPOP. The times and resources for
campaigning are usually limited; thus, a campaign manager has to delicately
determine the visiting priorities of the candidate regions. It is very
important to plan campaign routes considering the floating population.
This paper presents a mathematical model for
the problem. For a solution procedure, a hybrid heuristic algorithm consisting
of a vertex sequence setting algorithm and an optimal visiting-time setting
algorithm for a given sequence. A simulated annealing (SA) algorithm is used
for the first part, and a visiting-time optimisation algorithm (VTOA) is used
for the second part. Computational results on 98 benchmark problem instances
show the effectiveness of the proposed solution.
2.
Benchmark
Problems
To the best of our knowledge, no available
benchmark data for MPOP exist. Thus, 98 benchmark instances were developed
based on OP benchmark instances and VRP benchmark instances. Table 1 shows the
summary of data sets.
To generate benchmark instances, we used the
well-known benchmark data of the OP. The data were generated by Chao et al.
(1996b) and Tsiligirides (1984) and can be downloaded from http://www.mech.kuleuven.be/en/cib/op. Chao et al. (1996b) provided two sets of
data. The first set (Set 64) is composed of 14 benchmark instances with
different time limits Tmax, each of which has 64 vertices including
starting and ending vertices. The second set (Set 66) consists of 26 instances
with 66 vertices. Tsiligirides (1984) created three datasets with 18, 11, and
20 benchmark instances (Tsiligirides 1, Tsiligirides 2, and Tsiligirides 3).
The numbers of vertices in these datasets are 32, 21, and 33, respectively. We
converted these datasets into MPOP data. In addition, to generate more
large-sized instances, we used Solomon¡¯s VRP instances C101, R101, and RC101,
each of which has 101 vertices, including depot vertex. These VRP benchmark
data can be downloaded from http://web.cba.neu.edu/~msolomon/problems.htm. By converting the demands into profit, the
VRP benchmark instances can be converted into OP instances.
To convert
OP data into MPOP data, Qi (the
number of time slots for vertex i)
should be determined. We set Qi
to 3 for all i because a day is
naturally divided by morning, afternoon, and evening. For the profit value, we
first take the profit value S of a
vertex from an OP instance and generate three profits [0.7S, S, 1.3S]. These
profits are randomly allocated to the time slots of the same vertex in the MPOP
instance. Thus, many combinations of profits such as {Pi1 = S, Pi2
= 0.7S, Pi3 = 1.3S}, {Pi1 = S, Pi2 = 1.3S, Pi3
= 0.7S}, {Pi1 = 0.7S, Pi2 = 1.3S, Pi3 = S}, {Pi1 = 1.3S, Pi2 = 0.7S, Pi3 = S} are possible. The
same procedure for all of the vertices in each instance is performed. For the
starting and ending vertices, zero profits are allocated. The travel distances
are calculated as the Euclidean distance and rounded to the first decimal.
Table 1.
Benchmark data summary
|
Name |
# of |
# of |
|
T1 |
32 |
18 |
|
T2 |
21 |
11 |
|
T3 |
33 |
20 |
|
S64 |
64 |
14 |
|
S66 |
66 |
26 |
|
VRP |
100 |
9 |
The 98 benchmark data sets can be downloaded by
click here.
Required
data can be calculated or retrieved as follows:
ü Travel
time from vertex i to vertex j (tij)
= Euclidean distance between the coordinates of the vertices, rounded
to the first decimal
ü
Each vertex has 3 time slots :
where each boundary time is rounded to
the first decimal (in OptionFile)
The
algorithm exe and CPLEX code files can be downloaded below:
The
solution files obtained from the algorithm ¡°DP + SA + ng-route + update
incumbent¡± can be downloaded
below:
ü Solution_files_from_the_MPOP_Exact_algorithm_DP
+ SA + ng-route + update incumbent
ü ¡°Hybrid Dynamic Programming with Bounding Algorithm
for the Multi-profit Orienteering Problem,¡± working
paper.
If you
have any questions, please contact us via the email below.
ü e-mail
: dnwjd2355@postech.ac.kr
3.
References
[1]
Chao, I., Golden, B. L., & Wasil, E. A. (1996b). A
fast and effective heuristic for the orienteering problem. European Journal of Operational Research, 88, 475-489. https://doi.org/10.1016/0377-2217(95)00035-6.
[2]
Gendreau, M., Laporte, G., & Semet, F. (1998). A
tabu search heuristic for the undirected selective traveling salesman problem. European
Journal of Operational Research, 106, 539-545. https://doi.org/10.1016/S0377-2217(97)00289-0.
[3]
Laporte, G., & Martello, S. (1990). The selective
traveling salesman problem. Discrete Applied
Mathematics, 26, 193-207. https://doi.org/10.1016/0166-218X(90)90100-Q.
[4]
Tsiligirides, T. (1984). Heuristic Methods Applied to
Orienteering. Journal of the Operational
Research Society, 35, 797-809. https://doi.org/0.2307/2582629.