ADAPTIVE NEURAL NETWORK ARCHITECTURE FOR ROBUST RECONFIGURABLE FLIGHT CONTROL

THE PROPOSED WORK ADDRESSES SPECIFIC ISSUES RELATED TO AIRCRAFT STABILITY DURING CONFIGURATION CHANGES, DAMAGE CONTROL AND CONFIGURATION MANAGEMENT WHICH HAVE RECEIVED INCOMPLETE TREATMENT IN EXISTING AIRCRAFT FLIGHT CONTROLLER SYSTEMS DESIGN. THE APPROACH WE HAVE SELECTED UTILIZES NEURAL NET TECHNOLOGY TO IDENTIFY AND COMPENSATE FOR ADVERSE DYNAMICS WHICH CAN BE ENCOUNTERED DURING OPERATIONAL CONFIGURATION CHANGES CONSEQUENT TO (1) DAMAGE CONTROL, (2) MISSION PROFILE SEGMENT TRANSITION OR (3) TACTICAL MODE CHANGE. ADVANCED SIMULATION TECHNIQUES WILL BE EMPLOYED TO EVALUATE THE POTENTIAL OF SEVERAL ALTERNATIVE NEURAL NET ARCHITECTURES IN IDENTIFICATION AND CONTROL OF ADVERSE PLANT CHARACTERISTICS RESULTING FROM CONFIGURATION CHANGE. EXPERIMENTAL RESULTS WILL BE USED TO SPECIFY A DESIGN STRATEGY FOR A RECONFIGURATION MANAGER WHICH IS SENSITIVE TO MISSION OBJECTIVES AND SITUATIONAL VARIABLES WHILE MAXIMIZING AIRCRAFT PERFORMANCE AND SAFETY DURING RECONFIGURATION.