Multiple Sensor Data Integration (PNNL Scope # 90001)
EMSL Project ID
5116
Abstract
Development of a mathematical/statistical framework for nuclear, chemical, and biological (NCB) sensor integration and decision analysis is proposed. In particular, a statistical model for integrating disparate instrument signatures will be developed and incorporated into a triage tree approach to identification of unknowns. A Bayesian decision framework will be used for data integration and unknown identification. A key benefit of the Bayesian framework is that it allows users to place relative importance on each piece of information through prior distributions and costs associated with incorrect decisions. The statistical models developed from this framework will then be combined with a companion effort being funded by the HSI initiative for nuclear threat detection, and propagated through a triage approach to NCB signature detection and identification. Successful completion of this research will result in robust, effective algorithms that can be used in development and deployment of integrated sensor systems. While the proposed research is being developed for NCB threat detection, potential applications for this work are very broad and include automated analyte detection/identification in field deployable and high throughput analytical laboratory applications. In order to due this, we need more transmissive FTIR data on a number of vegetative bacteria.
Project Details
Project type
Exploratory Research
Start Date
2003-10-06
End Date
2003-10-06
Status
Closed
Released Data Link
Team
Principal Investigator
Team Members