Improved flower pollination optimization algorithm based on swap operator and dynamic switch probability selection

Iqbal, Muhammad (2023) Improved flower pollination optimization algorithm based on swap operator and dynamic switch probability selection. Doctoral thesis, Universiti Tun Hussein Onn Malaysia.

	Text 24p MUHAMMAD IQBAL.pdf Download (1MB)
	Text (Copyright Declaration) MUHAMMAD IQBAL COPYRIGHT DECLARATION.pdf Restricted to Repository staff only Download (1MB) | Request a copy
	Text (Full Text) MUHAMMAD IQBAL WATERMARK.pdf Restricted to Registered users only Download (4MB) | Request a copy

Abstract

Meta-heuristic algorithms have become popular in finding optimal solutions for nonlinear complex problems. These algorithms belong to stochastic and nondeterministic classes, have some problems with exploration or exploitation. The researchers used different strategies to tackle these issues. The flower pollination algorithm (FPA) is one of the more popular nature-inspired search algorithms. It has powerful global searchability to find the best optimal solution of real-world problems by using levy flight to explore the search space. Moreover, its single tuning parameter and simple mathematical model make it easier to implement. However, the flower pollination algorithm has a few drawbacks which are partially addressed by the practitioners. The first issue is regarding the balance between global pollination and local pollination, which may negatively affect the optimality of solution. Secondly, the FPA has a diversification problem which may leads to premature convergence of the optimal solution. This research proposed an algorithm which is based on dynamic switch probability to control the balance between exploration and exploitation which increases its searchability. The swap operator has been added in local pollination to enhance the exploitation behavior of pollens during the pollination process. Furthermore, it is hybridized with the Pattern Search algorithm to ensure the optimality of the solution. The performance of the proposed algorithm (IFPDSO-PS) has been evaluated on seventeen standard test functions and compared with the stated metaheuristic algorithms. The statistical tools (Absolute Mean Error & Fried Man) are used to rank the performance of all algorithms. The proposed algorithm is ranked first among the stated algorithms with respect to its performance in getting the optimal solution

Item Type:	Thesis (Doctoral)
Subjects:	T Technology > T Technology (General)
Divisions:	Faculty of Computer Science and Information Technology > Department of Information Security
Depositing User:	Mrs. Sabarina Che Mat
Date Deposited:	29 May 2024 02:26
Last Modified:	29 May 2024 02:26
URI:	http://eprints.uthm.edu.my/id/eprint/11046

Actions (login required)

View Item