PARALLEL AND MULTISTAGE FUZZY INFERENCE BASED ON FAMILIES OF ALPHA-LEVEL SETS

Authors
Citation
K. Uehara et K. Hirota, PARALLEL AND MULTISTAGE FUZZY INFERENCE BASED ON FAMILIES OF ALPHA-LEVEL SETS, Information sciences, 106(1-2), 1998, pp. 159-195
Citations number
37
Categorie Soggetti
Computer Science Information Systems","Computer Science Information Systems
Journal title
ISSN journal
00200255
Volume
106
Issue
1-2
Year of publication
1998
Pages
159 - 195
Database
ISI
SICI code
0020-0255(1998)106:1-2<159:PAMFIB>2.0.ZU;2-5
Abstract
In this paper, methods for parallel fuzzy inference and multistage-par allel fuzzy inference are studied on the basis of families of alpha-le vel sets. The parallel fuzzy inference is characterized by the unifica tion of inference consequences obtained from a number of conditional p ropositions. Thus, in this paper, the methods for the unification of i nference consequences via alpha-level sets are presented first. It is found that the unification approximated by using fuzzy convex hull is efficient in the case where the unification is performed by the maximu m operation. The methods for defuzzification are also examined via alp ha-level sets for the unified consequences. The computational efficien cy is evaluated in order to show the effectiveness of the unification and defuzzification via alpha-level sets. Moreover, it is studied by c omputer simulations how the approximation by fuzzy convex hull affects the performance in fuzzy control. The results indicate that this appr oximation does not degrade the control performance. Next, the multista ge-parallel fuzzy inference is considered from the operational point o f view via alpha-level sets, The multistage-parallel fuzzy inference i s characterized by passing the unified consequence of parallel fuzzy i nference in each stage to the next stage as a fact. Hence, the studies are focused on this consequence passing in this paper. It is larified that the straightforward way of inference operations via alpha-level sets is time consuming because of the nonconvexity in the unified infe rence consequence in each stage. In order to solve the problem, the mu ltistage-parallel fuzzy inference is formulated into a form of linguis tic-truth-value propagation. As a result, the inference operations in middle stages can be conducted by convex fuzzy sets and then efficient computations for inference is provided. The computational efficiency is also evaluated to show the effectiveness of the formulation. Finall y, this paper concludes with some brief discussions. (C) 1998 Elsevier Science Inc. All rights reserved.