No abstract

No abstract

This paper presents a hybrid spiral dynamic algorithm with a super-opposition spiral dynamic algorithm (SOSDA) strategy. An improvement on the spiral dynamic algorithm (SDA), this method uses a concept centered on opposition-based learning, which is used to evaluate the fitness of agents at the opposite location to the current solution. The SDA is a simple-structured and deterministic type of algorithm, which also performs competitively in terms of solution accuracy. However, its deterministic characteristic means the SDA suffers premature convergence caused by the unbalanced diversification and intensification during its search procedure. Thus, the algorithm fails to achieve highly accurate solutions. It is proposed that adopting super-opposition into the SDA would enable the deterministic and random techniques to complement one another. The SOSDA was tested on four benchmark functions and compared to the original SDA. To analyze the result statistically, the Friedman and Wilcoxon tests were conducted. Furthermore, the SOSDA was applied to optimize the parameters of an interval type-2 fuzzy logic control (IT2FLC) for an inverted pendulum (IP). The statistical results produced by the SOSDA for both benchmark functions and the IP show that the proposed algorithm significantly outperformed the SDA. The SOSDA-based IT2FLC scheme also produced better IP responses than the SDA-based IT2FLC.

No abstract

No abstract