1. UKM Learning and Research Repository
2. PHD Thesis / Tesis PHD
3. Faculty of Information Science and Technology / Fakulti Teknologi dan Sains Maklumat

Title:	Plunge-diving seabirds inspired optimization algorithm in polar coordinates with levy flight
Authors:	Moaath Mohammad Amin Shatnawi (P63458)
Supervisor:	Mohammad Faidzul Nasrudin, Assoc. Prof. Dr.
Keywords:	Coordinate system Cartesian Particle Levy flight Algorithm Universiti Kebangsaan Malaysia -- Dissertations
Issue Date:	4-Apr-2017
Description:	The coordinate system is a system of representing points in a space of given dimensions by coordinates, such as the Cartesian coordinate system or the system of celestial longitude and latitude. Swarm-based algorithms are appearing as a family of nature-inspired optimization algorithms that are able to produce fast, low cost and robust solution to various optimization problems. Particles in most of algorithms move in the Cartesian coordinate system. However, in Cartesian coordinate system, a small disturbances change of the particle will result in a great angle change of particle in D dimensional space, which make particle be far away from the optimal point. One solution is to compute the algorithm in polar coordinate system such as what has been accomplished by the Polar Particle Swarm Optimizer (PPSO). However, PPSO is suffering from a distortion of particles of initialization, and imprecisions in floating points from sine and cosine multiplications in high dimensions. During transformation from Cartesian to polar coordinates the distance will stretch the search in the space. Besides that, the concept of the modified velocity update equation in PPSO is still in the Cartesian coordinated system which may distort the movement of its particles in the search space. Therefore, improving the PPSO or proposing a new algorithm that has velocity and position update totally in polar coordinates is needed. The first objective of this work is to improve the drawbacks from the PPSO. Particles initialization in polar coordinates is a novelty. It is done by improving particle distribution and adding a constraint to enforce the outbound particle positions to be in a valid region. The second objective is to propose a new algorithm, operates entirely in polar coordinates, which mimics plunge-diving behavior of sea birds during foraging. The algorithm consists of three parts. Firstly, most of the birds move toward the global best bird. Secondly, some birds move toward near the global best with some randomization. Thirdly, few birds move randomly to discover the rest of the search space. The third objective is to add the Levy flights to the new algorithm. This is to improve particles diversification by the step-lengths with a heavy-tailed probability distribution. This version of the algorithm is then tested to two engineering problems, the welded beam design and spring design optimization. Throughout the work, eight benchmark test functions such as DeJong, are used to evaluate all algorithms with various settings of particles, dimensions, and algorithm parameters. The results of the new algorithm are compared with the results from the original PSO and PPSO. Experimental results from the new initialization technique show little enhancements in some benchmark test functions of both the original PSO and PPSO, but statistically, there are no significant differences by the analysis of variance (ANOVA). The new algorithm outperformed the Polar Particle Swarm Optimizer in most cases and statistically it is performed better and significantly. At the same time, it does not outperform the original Particle Swarm Optimization (PSO) algorithm in all cases of settings. Results from the new algorithm with Levy flights outperform significantly both the original PSO and PPSO in almost all cases. Besides that, the new algorithm successfully finds the optimum point of the both engineering problems. The new algorithm proved to be a better algorithm compared to the original PSO and PPSO. Its also a new algorithm runs truly in the polar coordinate system.,Certification of Master's/Doctoral Thesis" is not available
Pages:	210
Publisher:	UKM, Bangi
Appears in Collections:	Faculty of Information Science and Technology / Fakulti Teknologi dan Sains Maklumat

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.