Bayesian Network Classifiers for Set-Based Collaborative Design

Abstract

Complex design problems are typically decomposed into smaller design problems that are solved by domain-specific experts who must then coordinate their solutions into a satisfactory system-wide solution. In set-based collaborative design, collaborating engineers coordinate themselves by communicating multiple design alternatives at each step of the design process. The goal in set-based collaborative design is to spend additional resources exploring multiple options in the early stages of the design process, in exchange for less iteration in the latter stages, when iterative rework tends to be most expensive. Several methods have been proposed for representing sets of designs, including intervals, surrogate models, fuzzy membership functions, and probability distributions. In this paper, we introduce the use of Bayesian networks for capturing sets of promising designs, thereby classifying the design space into satisfactory and unsatisfactory regions. The method is compared to intervals in terms of its capacity to accurately classify satisfactory design regions as a function of the number of available data points. A simplified, multilevel design problem for an unmanned aerial vehicle is presented as the motivating example.

Pattern Search Algorithm for Blackboard-Based Multidisciplinary Design Optimization Frameworks

“…drag and mass. Of course one can always combine these into the Breguet range equation and use that as an objective function as demonstrated by [7]. But in order to keep the objective function in the format given in Section IV.A.1, we deemed the weighted function to be sufficient in capturing the compromise between the two domains.…”

“…Previously, other authors have used a similar concept to improve the practicality of their proposed MDO methods. Shahan and Seepersad [7] used a Bayesian Network to capture knowledge from individual domain analyses. The shared design variables were mapped on probability distributions to enable a chief engineer to select designs that 1 meet the preferences of multiple domains.…”

New Solutions for Energy Absorbing Materials

“…In future work, NS research of this nature should focus on the implementation of the models validated during this project to perform an iterative exploration of the micro-and mesoscale design space to improve the elastic and absorptive properties of viscoelastic composites. Multiscale design model should be wrapped by a nonlinear optimization scheme, such as kernel-based Bayesian network mappings to identify optimal microscale designs that lead to amplified stiffness and loss performance through the use of NS [46].…”

Sequential Sampling With Kernel-Based Bayesian Network Classifiers

“…Two particularly simple classifiers, the Parzen window classifier and the naïve Bayes classifier, have been shown in prior work to have significantly lower classification errors than interval classifiers [1]. In this prior work, we introduced the kernel-based Bayesian network (KBN) classifier framework, as described in Section 3, that includes both the Parzen window and naïve Bayes classifiers (and other types as well) in one representation.…”

“…In previous implementations of our classifier approach to set-based design [1], we relied upon deterministic space filling sampling methods-Hammersley [2] and Halton [3] sequences in particular.…”