Information Technology Reference
In-Depth Information
The first multi-objective optimisation problem appears to have been tackled in
2006 [12] for the design and operation of a heat pipe on a satellite. One of the few
mixed variable problems was also reported in 2006 [13].
Applications related to Information Technology (IT), including data clustering [14,
15] and information routing in computer networks [16], have begun appearing in re-
cent years.
Since the introduction of the original HS algorithm, researchers have been investigat-
ing ways to improve its performance or adapt it to new types of problems. Even the two
original studies [1, 7] present the basic method and two modifications. Figure 2 shows
only a few of the many innovations that have been reported. Extensive modifications to
the original algorithm were made to solve a type of ecological conservation problem for
Oregon, USA, termed the Maximal Covering Species Problem (MCSP) [17]. Mahdavi,
Fesanghary and Damangir [18] developed the Improved Harmony Search (IHS), which
has been used in several subsequent studies. The highly modified algorithm by Cheng,
Li and Chi [19] is another noteworthy development. Around 45% of studies to date
(Figure 1) use the original HS algorithm as presented by Yang in Section 2.2 of Chapter
1, with the remainder using modified versions.
Modified methods have also been developed by hybridising HS with other meta-
heuristic optimisation methods, such as GA or Particle Swarm Optimisation (PSO).
Approximately half of the modified HS algorithms may be classified as hybrid meth-
ods. Section 4 discusses hybrid and other modified HS algorithms.
A little theoretical analysis of the HS algorithm has been conducted. Studies in-
clude HS convergence [1, 7], the 'stochastic partial derivative' [20] and a population
variance analysis for Harmony Memory [21].
Some of the larger problems studied to date using original or modified HS methods
include
•
MCSP for ecological conservation with 441 binary variables [17], and a pipe net-
work layout problem with 112 binary variables [22, 23];
•
Balerma water distribution network with 454 discrete variables [24], and university
course timetabling with 450 discrete variables [25];
•
Benchmark optimisation problems with up to 100 continuous variables [26, 27],
and soil stability analysis with up to 71 continuous variables [28];
•
In terms of mixed variable optimisation, structural optimisation problems with up
to 8 discrete and 13 continuous variables [29], and 8 discrete and 5 continuous
variables [13].
2.3 Areas of Activity
Until around 2005 about half the published HS studies were devoted to the design of
water distribution networks. Benchmark optimisation, structural design and route
finding problems comprised the remaining portion of the literature. Since 2005 or so
the range of applications areas has expanded. Figure 3 shows the current approximate
distribution of HS applications, as measured by the number of publications devoted to
that topic. Section 3 goes through each of these discipline areas, describing the typical
problems addressed and citing specific publications.
Search WWH ::
Custom Search