Adapting And Hybrid Ising Harmony Search With Metaheuristic Components For University Course Timetabling

Al-Betar, Mohammed Azmi (2010) Adapting And Hybrid Ising Harmony Search With Metaheuristic Components For University Course Timetabling. PhD thesis, Universiti Sains Malaysia.

[img]
Preview
PDF
Download (1MB) | Preview

Abstract

Masalah Penjadualan Waktu Kursus Universiti (MPWKU) merupakan suatu masalah penjadualan kombinatorik yang rumit. Algoritma Gelintaran Harmoni (AGH) ialah suatu kaedah metaheuristik berdasarkan populasi. Kelebihan utama algoritma ini terletak pada keupayaannya dalam mengintegrasikan komponen-komponen utama bagi kaedah berdasarkan populasi dan kaedah berdasarkan gelintaran setempat dalam satu model pengoptimuman yang sama. Disertasi ini mencadangkan suatu AGH yang telah disesuaikan untuk MPWKU. Penyesuaian ini melibatkan pengubahsuaian terhadap operator AGH. Hasil yang diperoleh adalah dalam lingkungan keputusan terdahulu. Tetapi beberapa kelemahan dalam kadar penumpuan dan eksploitasi setempat telah dikesan dan telah diberikan tumpuan menerusi penghibridan dengan komponen metaheuristik yang diketahui. Tiga versi terhibrid dicadangkan, di mana, setiap hibrid merupakan peningkatan daripada yang sebelumnya: (i) Algoritma Gelintaran Harmoni yang Diubah suai; (ii) Algoritma Gelintaran Harmoni dengan Kadar Penyesuaian Berbagai Nada, dan (iii) Algoritma Gelintaran Harmoni Hibrid. Semua hasil yang diperoleh dibandingkan dengan 21 kaedah lain menggunakan sebelas dataset piawai de facto yang mempunyai saiz dan kekompleksan yang berbeza-beza. University Course Timetabling Problem (UCTP) is a hard combinatorial scheduling prob- !em. Harmony Search Algorithm (HSA) is a recent metaheuristic population-based method. The major thrust of this algorithm I ies in its abiiity to integrate the key components of populationbased methods and local search-based methods in the same optimisation model. This dissertation presents a HSA adapted for UCTP. The adaptation involved modifying the HSA operators. The results were within the range of state of the art. However, some shortcomings in the convergence rate and local exploitation were identified and addressed through hybridisation with known metaheuristic components. Three hybridized versions are proposed which are incremental improvements over the preceding version: (i) Modified Harmony Search Algorithm (MHSA); (ii) Harmony Search Algorithm with Multi-Pitch Adjusting Rate (HSA-MPAR), and (iii) Hybrid Harmony Search Algorithm (HHSA). The results werecompared against 21 other methods using eleven de facto standard dataset of different sizes and complexity.

Item Type: Thesis (PhD)
Subjects: Q Science > QA Mathematics > QA75.5-76.95 Electronic computers. Computer science
Divisions: Pusat Pengajian Sains Komputer (School of Computer Sciences) > Thesis
Depositing User: Mr Noorazilan Noordin
Date Deposited: 09 Mar 2017 02:02
Last Modified: 12 Apr 2019 05:26
URI: http://eprints.usm.my/id/eprint/32357

Actions (login required)

View Item View Item
Share