Detail Fuzzy neural applications in group technology Bibliografi Author: Lee, E. Stanley (Advisor); Pai, Ping-Feng Frank Topik: ENGINEERING; INDUSTRIAL|ARTIFICIAL INTELLIGENCE Bahasa: (EN )    ISBN: 0-599-12932-8     Penerbit: KANSAS STATE UNIVERSITY     Tahun Terbit: 1998     Jenis: Theses - Dissertation Fulltext: 9914223.pdf (0.0B; 5 download) Abstract The most basic operation in group technology is the formation of different parts or machines into groups. One of the difficulties in this operation is the fact that the specifications of the characteristics of the parts or machines are usually vague and linguistic in nature. The use of fuzzy set, which can represent vague or linguistic expressions easily, forms an ideal approach to overcome this problem. Furthermore, since fuzzy representation retains all the original knowledge, more efficient operation can be obtained by exploring the vagueness of the original problem through this exact but vague representation. Due to the vagueness of the fuzzy representation, which is principally caused by the characteristics of the original problem, some forms of learning or up-dating are desirable. Neural network can be used to serve this purpose. In addition to its learning ability, neural network can and has been used for clustering or classification purposes in group technology. Thus, the recent developments in neural-fuzzy networks, which combine the abilities of learning and classification of neural network and the ability of representation of fuzzy set theory, are ideal approaches to form a more efficient algorithm for group technology. Several fuzzy-neural approaches are applied to group technology in this dissertation and both the supervised and unsupervised learning are investigated. The fuzzy self-organizing maps network, an unsupervised learning network, is first investigated with fuzzy weights and fuzzy input data. The influences of the various parameters on the system performances are analyzed by observing the convergent behavior and the grouping performance. Different input data are also generated and the performances are compared. For the supervised learning approach, the adaptive fuzzy-neural systems are employed to handle the part-machine grouping problem. Both the backpropagation training algorithm and the radio basis function approach are used to train the adaptive fuzzy systems. Comparison of the convergent behaviors of these two approaches are discussed. The handling of the newly coming parts assignment problem is also investigated. Finally, the adaptive fuzzy system with different rules are analyzed to observe the mapping and generalization abilities of the system. Opini Anda Klik untuk menuliskan opini Anda tentang koleksi ini! Lihat Sejarah Pengadaan  Konversi Metadata   Kembali

Copyright © 2006, 2007 Unika Atma Jaya, all rights reserved