A parallel harmony search algorithm with dynamic harmony-memory size

Inspired by the concept of swarm intelligence, a parallel harmony search algorithm with dynamic harmony-memory size (PDHS) is presented to solve unconstrained problems. The PDHS algorithm has two significant operations: Harmony Memory (HM) grouping and Harmony Memory Size (HMS) updating. In the former operation, the harmony memory is divided into several groups named sub-HMs which help decreasing the optimization time because of their parallel searching. In the latter operation, the HMS value is dynamically changed according to the given rules during the iteration, which helps keeping the diversity of harmonies and supporting greater exploration in search space. Various benchmark optimization problems are tested to demonstrate the effectiveness of the proposed algorithm. The simulation results reveal that the proposed PDHS algorithm can get better solutions and is more efficient than the original HS algorithm in most cases.