Technical Approach

The Underwater Munitions Expert System (UnMES) will be a prototype software tool demonstrating the methodology for providing knowledge required to effect a more efficient deployment of the likely constrained resources available for underwater munitions detection and classification.

The expert system approach employs a Bayesian network whose probabilistic basis provides a means of accounting for the inherent uncertainties given unknown initial conditions and the long time intervals. UnMES will draw on the researchers' extensive experience in the successful development of a Bayesian network for the prediction of sea mine burial. The knowledge base for UnMES will be derived from several sources. The primary resource consists of the extant physics-based process models of UXO mobility, burial, and reexposure. These available models, however, do not sufficiently cover the range of conditions present at remediation sites of interest. To extend these models, the researchers will conduct limited, focused laboratory studies designed to address crucial gaps in the phenomenological understanding of munitions mobility and sediment‐object interaction. Additional knowledge resources are the numerous SERDP field studies that have been completed in the past several years, which can be analyzed to infer characteristic relationships between environment classification and munitions location and burial.