DEMONSTRATION OF NEURAL NETWORK PERFORMANCE FOR ON-LINE HEALTH MONITORING AND INSPECTION OF AIRCRAFT

INNOVATIVE DYNAMICS IS PROPOSING TO IMPLEMENT NEURAL NETWORK SIGNAL PROCESSING SOFTWARE WITH OUR AIRCRAFT HEALTH MONITORING SYSTEM (HMS) TESTBED. HMS IS A UNIQUE DISTRIBUTED SYSTEM BASED ON "SMART STRUCTURES" TECHNOLOGY. IT HAS RECENTLY SUCCESSFULLY COMPLETED A FLIGHT TEST PROGRAM AT NASA LEWIS THAT DEMONSTRATED THE FEASIBILITY TO DETECT RIVET LINE CORROSION AND LEADING EDGE ICE ACCRETION. BY MONITORING THE RESULTANT STRUCTURAL VIBRATION SIGNATURE, HMS DETERMINES STRUCTURAL ABNORMALITIES USING A NETWORK OF ATTACHED SENSOR ARRAYS, DISTRIBUTED PROCESSORS, AND INDEPENDENT SIGNAL PROCESSING SOFTWARE MODULES. USING CONVENTIONAL MINIMUM DISTANCE CLASSIFICATION ALGORITHMS, A HIGH CLASSIFICATION ACCURACY HAS BEEN ACHIEVED. THE OBJECTIVE OF THE PROPOSED PROGRAM IS TO EXTEND THIS RESEARCH BY IMPLEMENTING NEURAL NETWORK SOFTWARE FOR PERFORMANCE COMPARISON WITH OUR PATTERN RECOGNITION BASED ON-LINE MONITORING/ INSPECTION SYSTEM. THIS SOFTWARE WILL BE IMPLEMENTED ON EXISTING FLIGHT QUALIFIED HARDWARE AND DEMONSTRATED DURING A PHASE I GROUND TEST PROGRAM. UPON SUCCESSFUL CONCLUSION OF PHASE I, A TEST PLAN WILL BE SUBMITTED TO THE NAVY FOR A PHASE II DEMONSTRATION ON A SPECIFIC AIRFRAME USING A VARIETY OF NONDESTRUCTIVE MATERIALS MEASUREMENTS. A NEURAL NETWORK BASED SYSTEM HAS THE POTENTIAL FOR LIFETIME MONITORING OF STRUCTURAL PROPERTIES AS WELL AS PROVIDING A REAL-TIME DAMAGE DETECTION CAPABILITY CRITICAL TO NAVAL FIGHTER/ATTACK AIRCRAFT.