An enhanced class topper algorithm based on particle swarm optimizer for global optimization
- PDF / 2,719,444 Bytes
- 19 Pages / 595.276 x 790.866 pts Page_size
- 87 Downloads / 232 Views
An enhanced class topper algorithm based on particle swarm optimizer for global optimization Alfred Adutwum Amponsah 1,2
&
Fei Han 1,2 & Qing-Hua Ling 3 & Patrick Kwaku Kudjo 4
# Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract Class topper optimization (CTO) algorithm divides the initial swarm into several sub-swarms, and this causes it to possess a strong exploration ability throughout optimization. It however randomly selects best-ranked particles as section toppers (ST’s) and class topper (CT), and the inability of every particle to directly learn from the CT causes slow convergence during the latter stages of iterations. To overcome the algorithm’s deficiency and find a good balance between exploration and exploitation, this study proposes an enhanced CTO (ECTPSO) based on the social learning characteristics of particle swarm optimization (PSO). We created an external archive called the assertive repository (AR) to store best-ranked particles and employed the Karush-Kuhn-Tucker (KKT) proximity measure to assist in the selection of STs and CT. Also, the intensive crowded sorting (ICS) is developed to truncate the AR when it exceeds its maximum size limit. To further encourage exploitation and avert particles from getting trapped in local optimum, we incorporated an adaptive performance adjustment strategy (APA) into our framework to activate particles when they are stagnated. The CEC2017 test suite is employed to evaluate the effectiveness of the proposed algorithm and four other benchmark peer algorithms. The results show that our proposed method possesses a better capability to elude local optima with faster convergence than the other peer algorithms. Furthermore, the algorithms were applied to economic load dispatch (ELD), of which our proposed algorithm demonstrated its effectiveness and competitiveness to address optimization problems. Keywords Adaptive performance adjustment . Class topper optimization . Intensive crowded sorting . Particle swarm optimization
1 Introduction Finding the global optimum solution requires an effective optimization algorithm, which is balanced in terms of exploitation and exploration [1]. Exploration pertains to covering the entire search space to achieve a high level of diversification among the solutions acquired. However, there is the need to converge as quickly as possible when a potential optimum solution is discovered in a promising area. In this regard,
* Fei Han [email protected] 1
School of Computer Science and Communication Engineering, Jiangsu University, Zhenjiang 212013, Jiangsu, China
2
Jiangsu Key Laboratory of Security Technology for Industrial Cyberspace, Zhenjiang 212013, Jiangsu, China
3
School of Computer Science, Jiangsu University of Science and Technology, Zhenjiang 212013, Jiangsu, China
4
Department of Information Technology, University of Professional Studies, Accra, Ghana
exploitation is achieved and it increases the likelihood of converging on a very good solution. Although the traditional optimization algori
Data Loading...
