Bayesian Inference Modeling of Total Ionizing Dose Effects on System Performance

Z. J. Diggins, Nagabhushan Mahadevan, E. Pitt, D. Herbison, Rebecca M. Hood, Gábor Karsai, Brian D. Sierawski, Eric J. Barth, Robert A. Reed, Ronald D. Schrimpf, Robert A. Weller, Michael L. Alles, Arthur F. WitulskiIEEE Transactions on Nuclear Science 2015 · DOI

Abstract

A probabilistic Bayesian modeling method for determining the effects of radiation-induced component-level parameter shifts on system-level performance is described. The modeling method incorporates information about the system design and component-level degradation into a Bayesian network and performs inference on the constructed network using Markov chain Monte Carlo approaches, producing distributions for the range of component responses. Deterministic simulations use the results of the Bayesian inference to determine the combined impact of multiple degraded components on system performance quantities. The goal of the modeling approach is to turn uncertain information into actionable knowledge. The utility of this approach is demonstrated using a case study based on total ionizing dose degradation of line-sensor components in a simple line-tracking robot system.
BibTeX
@article{diggins2015bayesian,
  author = {Z. J. Diggins and Nagabhushan Mahadevan and E. Pitt and D. Herbison and Rebecca M. Hood and Gábor Karsai and Brian D. Sierawski and Eric J. Barth and Robert A. Reed and Ronald D. Schrimpf and Robert A. Weller and Michael L. Alles and Arthur F. Witulski},
  title = {{Bayesian Inference Modeling of Total Ionizing Dose Effects on System Performance}},
  journal = {IEEE Transactions on Nuclear Science},
  year = {2015},
  doi = {10.1109/tns.2015.2493882},
}

Loading related content...

Figure