A new evolutionary algorithm based on Alopex and Harmony Search algorithm

Abstract

Alopex (Algorithm of Pattern Extraction) was a correlation-based algorithm which possessed characteristics of both gradient descent and simulated annealing. It had been proved to be an effective tool for engineering optimization. The HS (Harmony Search) algorithm was a meta-heuristic algorithm proposed in recent years and had been shown several advantages compared with traditional optimization methods such as GA (genetic algorithm). In this paper, HS was embedded into the Alopex-based evolutionary algorithm (AEA) to form an improved evolutionary algorithm HS-Alopex. In the HS-Alopex, with the help of the random nature of HS, the diversity of population was improved and the prematurity problem was alleviated to a certain extent. The proposed algorithm is investigated on ten commonly used benchmark functions. Simulation results demonstrate that the new algorithm can obtain a better solution quality and faster convergence speed, comparing with the single AEA and HS algorithm.