SBIR-STTR Award

A software product will be developed that will enhance common engineering analysis tools with a probabilistic capability. This capability will involve the probabilistic quantification and management of confidence in the model-based predictions. The quantification of the confidence will be achieved by relying on Polynomial Chaos representations of stochastic variables and fields. This will permit the development of the software product as a library that integrates, with the least amount of intrusion, with existing software packages. In addition to characterizing the probabilistic content of the predictions, this formulation also permits the computation of the sensitivity of the probabilistic statements regarding the predictions with respect to probabilistic statements on the random data. This information can be used either to identify those probabilistic statements that are consistent with available information, or to design data acquisition efforts aimed at achieving a target confidence in the predictions,

Keywords:
Validation Of Predictive Models, Probabilistic Sensitivity, Polynomial Chaos, Probabilistic Error Estimation, Probabilistic Model Adaptation.