Abstract
Metaheuristic algorithms based on the collective behavior of nature
social groups, such as ants and bees, have been widely explored to solve
many optimization problems in engineering and other sciences. The
processing time and the chance to end up in a local optimal solution are
drawbacks of these algorithms, and none has proved to outperform the
others. In this paper, an improved swarm optimization technique, named
Grand Tour Algorithm (GTA), based on the behavior of a peloton of
cyclists, is introduced and applied to sixteen benchmarking optimization
problems in the literature to evaluate its performance in comparison to
the original particle swarm optimization (PSO) algorithm. Most of these
benchmarking problems are tackled with a number of 20,000 variables, a
really huge number inspired in the human genome. Under these conditions,
GTA clearly outperforms the classical PSO. In addition, various
sensitivity analyses are performed to verify the influence of the
initial parameters in the GTA efficiency. It can be demonstrated that
the GTA fulfils such coveted main aspects of an optimization algorithm
as ease of implementation, speed of convergence, and reliability, thus
confirming GTA’s improved performance.