Improving Gravitational Search Algorithm Performance with Artificial Bee Colony Algorithm for Constrained Numerical Optimization

TL;DR

This paper introduces GSABC, a hybrid algorithm combining gravitational search and artificial bee colony techniques, to enhance constrained numerical optimization by avoiding local minima and improving global search performance.

Contribution

The paper presents a novel hybrid optimization algorithm, GSABC, that integrates GSA and ABC to improve solution accuracy and robustness in constrained numerical problems.

Findings

GSABC outperforms five state-of-the-art methods on benchmark functions.

The hybrid approach effectively avoids local minima.

Experimental results demonstrate improved convergence speed.

Abstract

In this paper, we propose an improved gravitational search algorithm named GSABC. The algorithm improves gravitational search algorithm (GSA) results improved by using artificial bee colony algorithm (ABC) to solve constrained numerical optimization problems. In GSA, solutions are attracted towards each other by applying gravitational forces, which depending on the masses assigned to the solutions, to each other. The heaviest mass will move slower than other masses and gravitate others. Due to nature of gravitation, GSA may pass global minimum if some solutions stuck to local minimum. ABC updates the positions of the best solutions that has obtained from GSA, preventing the GSA from sticking to the local minimum by its strong searching ability. The proposed algorithm improves the performance of GSA. The proposed method tested on 23 well-known unimodal, multimodal and fixed-point multimodal benchmark test functions. Experimental results show that GSABC outperforms or performs similarly to five state-of-the-art optimization approaches.