Parametric design with neural network relationships and fuzzy relationships considering uncertainties

Authors:
Dong Zhao;Deyi Xue
Affiliations:
Department of Mechanical and Manufacturing Engineering, University of Calgary, 2500 University Drive N.W., Calgary, Alberta T2N 1N4, Canada;Department of Mechanical and Manufacturing Engineering, University of Calgary, 2500 University Drive N.W., Calgary, Alberta T2N 1N4, Canada
Venue:
Computers in Industry
Year:
2010

Citing 3
Cited 1

Bayesian methods for adaptive models

Bayesian methods for adaptive models
Learning in Neural Networks: Theoretical Foundations

Learning in Neural Networks: Theoretical Foundations
A First Course in Fuzzy Logic, Third Edition

A First Course in Fuzzy Logic, Third Edition

Lightweight design of vehicle parameters under crashworthiness using conservative surrogates

Computers in Industry

Quantified Score

Hi-index	0.00

Visualization

Abstract

This research introduces a new parametric design approach with neural network relationships and fuzzy relationships considering uncertainties. In this work, parameters are associated by a hybrid parameter relationship network. In addition to deterministic parameters and relationships, non-deterministic parameters (e.g., random parameters and fuzzy parameters) and non-deterministic relationships (e.g., neural network relationships and fuzzy relationships) can also be modeled in this network. Changes of parameter values and their uncertainties are propagated through this network. Two types of optimization methods, reliability based design optimization and possibility based design optimization, are employed to identify the optimal design considering objective random uncertainties and subjective fuzzy uncertainties. A computer system has been implemented and used for the optimal design of a solid oxide fuel cell (SOFC) system considering uncertainties.