Shipboard Chem-Bio Agent Spread Modeling for Damage Control

The overall objective of this project is to develop a system-level prognostic modeling tool for predictions of the spread of chemical/biological agents in shipboard air systems. The intent is to integrate the tool with existing DC systems. The proposed system will be modeled as a set of interacting software components coupled through a simulation environment. To achieve execution speed and accuracy, innovative use of Artificial Neural Nets (ANNs) will be made to model the different system components. The ANN models will directly interface with sensors and allow real-time updates of shipboard conditions, including sensor data validation. Phase I work will involve adaptation of an existing computational environment for multiple-thread parallel and series execution of the ANNs. Feasibility of the concept will be demonstrated on a representative ship section. Training of ANNs models will be done using CFD generated data. Treatment of sensor data validation and definition of the system architecture will be completed. Phase II project will involve adaptation of a network solver using ANNs as components for full ship system analysis. The software will be demonstrated and validated on a representative shipboard airflow system selected in collaboration with Navy, with ANNs trained for specific system components using CFD-generated and available experimental data. The work will oversee development of a GUI for effective data display to the operators, and development of strategies for interfacing with the ship control systems.