Bio-inspired Evolutionary Computing with Context-Awareness and Collective-Effect - Technologies and Applications of Artificial Intelligence - page 103

3 Guidable Bat Algorithm is Applied in Solving Optimization

Problems

In BA, the frequency of bat is generated by a random vector with uniform distribu-
tion. This randomly generated frequency effortlessly makes aimless search of bats to
deteriorate the evolution quality of algorithm. And, there are no exclusive rules to
steer the bats toward the correct direction during movement procedure. These situa-
tions deteriorate the evolution quality to search the global optimal solution for algo-
rithm. In order to overcome these disadvantages to enhance the solving efficiency, a
guidable bat algorithm with frequency shift based on Doppler Effect is proposed in
this study.
The principles of Doppler Effect and conflict behavior are applied in BA to invent
an innovative bio-inspired EC approach called as Guidable Bat Algorithm (GBA) as
shown in Fig. 1. This conception mainly provides a regular rule based on Doppler
Effect to guide the bats toward the correct direction in guidable search. The bats go-
verned in GBA are able to fast and accurately discover the global optimal solution. A
regular rule is able to move the step of bats according to the distance between the bats
and the current best bat appropriately.
In the traditional EC such as particle swarm optimization (PSO) [14, 15], the par-
ticles learn and communicate experience with each other. The particles only use the
fitness value to evaluate the quality of position. As well as, the fly direction of all
particles depend on the best particle. Hence, the particles are very similar so that the
particles easily fall into local optimal solutions. In order to overcome this disadvan-
tage. In divers search of GBA, the conflict behavior is considered to design the search
strategy to strengthen the ability of global search for bats. The bats search new loca-
tion by considering the similarity between their own location and another location.
Therefore, the population is comprised by diversity bats to search wider region.

Start

Initialization

Refined sesarch

N

If f(l i )<f(x * )
and r A <A i

Guidable search

Y

Update the
current best

Y

Update bat
information

If r p >r i

N

Divers search

N

If termination is
satisfied

Y

End

Fig. 1. Flowchart of a paradigm of GBA with bio-inspired evolutionary computing