NEUROENGINEERING ANALYSIS AND DESIGN TOOL FOR COMPLEX COOLANT SYSTEMS

DEVELOPMENT OF HYPERSONIC AIRCRAFT REQUIRES A HIGH DEGREE OF SYSTEM INTEGRATION. DESIGN TOOLS ARE NEEDED THAT CAN PROVIDE RAPID, ACCURATE CALCULATIONS OF COMPLEX FLUID FLOW PATTERNS. THE APPLICATION OF NEURAL NETWORKS IS PROPOSED TO MODEL COOLANT FLOW PATTERNS. FIRST, THE ABILITY OF NEURAL NETWORKS TO MODEL COOLANT FLOW IS BEING DEMONSTRATED AT THE COMPONENT LEVEL. A SINGLE HET EXCHANGER COMPONENT IS MODELED BY A NEURAL NETWORK AND TRAINED USING DATA GENERATED BY THE EXISTING FLOW EQUATIONS. THEN, MULTIPLE COMPONENTS IN SERIES WITHIN A FLOW LEG ARE MODELED USING A SERIES OF NEURAL NETWORKS. NEXT, THESE LEGS ARE JOINED IN PARALLEL TO FORM A HEAT EXCHANGER PANEL. FINALLY, VARIOUS COMBINATIONS OF PANELS IN SERIES AND PARALLEL ARE MODELED USING HIERARCHICAL NEURAL NETWORKS.