Tay, K.M and Kok, Chin Chai and Chee, Peng Lim (2015) A New Sign Distance-Based Ranking Method for Fuzzy Numbers. Proceedings of the 18th Asia Pacific Symposium on Intelligent and Evolutionary Systems-Volume 2, 2. pp. 55-64. ISSN 2363-6084
|
PDF
A new sign distance-based ranking method for fuzzy numbers.pdf Download (379kB) | Preview |
Abstract
In this paper, a new sign distance-based ranking method for fuzzy numbers is proposed. It is a synthesis of geometric centroid and sign distance. The use of centroid and sign distance in fuzzy ranking is not new. Most existing methods (e.g., distance-based method [9]) adopt the Euclidean distance from the origin to the centroid of a fuzzy number. In this paper, a fuzzy number is treated as a polygon, in which a new geometric centroid for the fuzzy number is proposed. Since a fuzzy number can be represented in different shapes with different spreads, a new dispersion coefficient pertaining to a fuzzy number is formulated. The dispersion coefficient is used to fine-tune the geometric centroid, and subsequently sign distance from the origin to the tuned geometric centroid is considered. As discussed in [5-9], an ideal fuzzy ranking method needs to satisfy seven reasonable fuzzy ordering properties. As a result, the capability of the proposed method in fulfilling these properties is analyzed and discussed. Positive experimental results are obtained.
Item Type: | Article |
---|---|
Uncontrolled Keywords: | Geometric centroid, sign distance, fuzzy numbers, reasonable ordering properties, dispersion coefficient, unimas, university, universiti, Borneo, Malaysia, Sarawak, Kuching, Samarahan, ipta, education, research, Universiti Malaysia Sarawak |
Subjects: | Q Science > Q Science (General) |
Divisions: | Academic Faculties, Institutes and Centres > Faculty of Engineering Faculties, Institutes, Centres > Faculty of Engineering |
Depositing User: | Meng |
Date Deposited: | 01 Jul 2015 03:56 |
Last Modified: | 02 Jul 2015 05:44 |
URI: | http://ir.unimas.my/id/eprint/8145 |
Actions (For repository members only: login required)
View Item |