Playground Algorithm as a New Meta-heuristic Optimization Algorithm

Abstract

This paper presents a new meta-heuristic algorithm called Playground algorithm. The Playground algorithm is designed to model social interaction amongst children, and the mechanisms and operators of the algorithm are inspired by the model of child interaction and engagement in games during a child's stay at the playground. In order to evaluate the performance of the algorithm, a series of tests were performed over a class of functions selected so that they possess properties such as: multimodality / unimodality, (non) separability, (non) differentiability, (non) convexity, existence of ridges and valleys and multidimensionality. During testing, the values of the algorithm parameters are varied, in order to determine their recommended values. The analysis was carried out with an overview of the effects of the algorithm parameters on the performance of the algorithm in the problem area, performance in the criterion domain, and the execution time.