A Hybrid Flower Pollination with p-Hill Climbing Algorithm for Global Optimization

In this paper, the -hill climbing optimizer is hybridized with the flower pollination algorithm (FPA) as a local refinement operator for global optimization problems. The proposed method is called HyFPA-hc. Such hybridization aims to enhance the balance between exploration and exploitation processes...

Full description

Saved in:
Bibliographic Details
Main Author: Abou Doush, Iyad (author)
Other Authors: Al-Betar, Mohammed Azmi (author), Awadallah, Mohammed (author), Alyasseri, Zaid Abdi Alkareem (author), Abasi, Ammar (author), Makhadmeh, Sharif (author), Alomari, Osama Ahmad (author)
Published: 2021
Online Access:https://dspace.auk.edu.kw/handle/11675/8251
https://www.sciencedirect.com/science/article/pii/S1319157821001609?via%3Dihub
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In this paper, the -hill climbing optimizer is hybridized with the flower pollination algorithm (FPA) as a local refinement operator for global optimization problems. The proposed method is called HyFPA-hc. Such hybridization aims to enhance the balance between exploration and exploitation processes during the search, thus improving the quality of the outcomes. -hill climbing optimizer is a recent trajectory-based algorithm with a powerful digging the niche to search and find the local optimum, while FPA is a recent population-based algorithm with robust mining several niches in the search space without proper concentration. The proposed HyFPA-hc is evaluated using 15 unimodal and multimodal test functions established in IEEE-CEC2015. The results show significant improvement in the convergence behaviour of the proposed HyFPA-hc over FPA using different dimensions of the test function. The comparative evaluation is also conducted against 26 state-of-the-art methods. The experiments consider three problem sizes (with dimensions 10, 30, and 50) to show the proposed HyFPA-hc performance against all comparative methods, where the proposed method outperformed all compared methods in optimizing 8, 7, 4 out of 15 test functions for 10, 30, 50 dimensions, respectively. Accordingly, the achieved results prove the efficiency of the proposed HyFPA-hc in optimizing various problem dimensions. In conclusion, the proposed hybrid metaheuristic method can search powerfully in the niches of optimization problems search space and produces very fruitful outcomes.