Applying bayesian networks to test-field cost optimization decisions

Numerous papers and texts have been written in the reliability literature regarding the determination of the optimum test duration for a production stress or burn-in test. The techniques presented have largely been based on the identification of the change point at which infant mortality has largely been removed from the units. The time-on-test is typically the only factor that influences this decision. Few of these models have attempted to integrate the field performance or the influence of warranty costs into this decision. Bayesian Networks (BNs) are a probabilistic methodology used to model and predict the behaviour of a particular system based on its observed stochastic phenomena called information variables. The observations are the quantitative part of the methodology. The main purpose of building a BN is to estimate the certainty of unobservable (or high-cost observable) events called as Hypothesis variables. The other part of a BN which is the qualitative part, is a directed acyclic graph (DAG) used to study the causality relationship between the variables. This paper utilizes a BN as a statistical technique to integrate the influence of the production test parameters and the field performance including their respective costs into a single unified model. The objective is the identification of a production test duration that minimizes the overall cost.