Fitness Varying Gravitational Constant in GSA

Bansal, Jagdish Chand and Joshi, Susheel Kumar and Nagar, Atulya K. (2018) Fitness Varying Gravitational Constant in GSA. Applied Intelligence (APIN), 48 (10). pp. 3446-3661. ISSN 0924-669X (Accepted for Publication)

[thumbnail of FVGGSA.pdf]
Preview
Text
FVGGSA.pdf - Accepted Version

Download (1MB) | Preview

Abstract

Gravitational Search Algorithm (GSA) is a recent metaheuristic algorithm inspired by
Newton's law of gravity and law of motion. In this search process, position change is based on the calculation of step size which depends upon a constant namely, Gravitational Constant (G). G is an exponentially decreasing function throughout the search process. Further, inspite of having different masses, the value of G remains same for each agent, which may cause inappropriate step size of agents for the next move, and thus leads the swarm towards stagnation or sometimes skipping the true optima.
To overcome stagnation, we first propose a gravitational constant having different scaling characteristics for different phase of the search process. Secondly, a dynamic behavior is introduced in this proposed gravitational constant which varies according to the fitness of the agents. Due to this behavior, the gravitational constant will be different for every agent based on its fitness and thus will help in controlling the acceleration and step sizes of the agents which further improve exploration and exploitation of the solution search space. The proposed strategy is tested over 23 well-known classical benchmark functions and 11 shifted and biased benchmark functions. Various statistical analyses and a comparative study with original GSA, Chaos-based GSA (CGSA), Bio-geography Based Optimization (BBO) and DBBO has been carried out.

Item Type: Article
Additional Information and Comments: This is the author's version of an article accepted for publication in Applied Intelligence. The final version will be published at http://www.springer.com/computer/ai/journal/10489
Keywords: Gravitational Search Algorithm (GSA), Swarm Intelligence, Gravitational Constant, Exploration, Exploitation.
Faculty / Department: Faculty of Human and Digital Sciences > Mathematics and Computer Science
Depositing User: Atulya Nagar
Date Deposited: 08 Jan 2018 09:31
Last Modified: 05 Mar 2019 01:15
URI: https://hira.hope.ac.uk/id/eprint/2331

Actions (login required)

View Item View Item