Neural network method based on a new damage signature for structural health monitoring

Abstract

Adopting wide-band Lamb wave based active monitoring technology, this study focuses on a neural network method based on a new damage signature for on-line damage detection applied to thin-walled composite structures. Honeycomb sandwich and carbon fiber composite structures are studied. Two kinds of damage are considered: delamination and impact damage. A new damage signature is introduced to determine the presence and extent of damage in composites, while eliminating the influence of different distances between the actuator and sensor. Neural network method is researched to take advantage of this new damage signature combined with several other signatures to decide the damage mode. Kohonen neural network is developed. The proposed method is shown to be effective, reliable, and straightforward for the specimens considered in the present study, which are composed of different materials and suffer various levels of damage.