A Novel Self-diagnosis Approach to Nondestructive Inspection of Composite Structures

High strength and low weight composite materials have been widely used in many critical, high-value military systems. The military environment is very harsh due to large changes in temperature, pressure, and loading conditions. As a result, delamination,fiber breakages, matrix cracks, disbonds, and other failures creep in once the system is in service for some time. Conventional methods are either costly and bulky, or have the potential to weaken the structure.Here we propose a novel system to detect and classify faults in composite structures. The system combines novel sensing ideas called Electric Resistance Change Method (ERCM) and a robust software for fault diagnosis. In the ERCM, we use the intrinsicresistance property of the composite materials. On the surface of composite structures, multiple thin (0.02 mm) and extremely light weight copper foil electrodes are mounted to measure the resistance. This approach has several key advantages: 1) embeddedsensing; 2) applicable to new and existing structures; 3) no strength reduction; 4) able to detect delamination, matrix crack, etc. without loading, and 5) low cost and portable. The second element of the system is an automatic fault diagnosis tool, whichconsists of Principal Component Analysis (PCA) and Learning Vector Quantization (LVQ). The proposed composite component failure diagnosis system is novel by itself and will have many practical applications in other structural diagnostics applications. The system combines latest development in sensor technology with most recentin neural networks to detect and classify composite failures and hence will provide valuable information that could save costs and will also help to make structures safe. The system has low computation burden and hence is suitable for real-timeapplications. An automatic tool for diagnosis will be very useful for structural integrity applications. We expect the market for this system will be at least 50 million.