Use of Bayesian Networks for Qualification Planning: Early Results of Factor Analysis

Abstract

This paper discusses the factor analysis that provides the basis for development and use of Bayesian Network (BN) models to support qualification planning in order to predict the suitability of Six Degrees of Freedom (6DOF) vibration testing for qualification. Qualification includes environmental testing such as temperature, vibration and shock to support a stochastic argument about the suitability of a design. Qualification is becoming more complex because it involves significant human expert judgment and relies on new technologies that have often never been fully utilized to support design assessment. Technology has advanced to the state where 6DOF vibration tests are possible, but these tests are far more complex than traditional single degree of freedom tests. This challenges systems engineers as they strive to plan qualification in an environment where technical and environmental constraints are coupled with the traditional costs, risk and schedule constraints. BN models may provide a framework to aid Systems Engineers in planning qualification efforts with complex constraints. Previous work identified a method for building a BN model for the predictive framework. This paper discusses validation efforts of models derived from the factor analysis and summarizes some recommendations on the factor analyses from industry subject matter experts.