Intelligent computational methods for corrosion damage assessment

Corrosion is one of the damage mechanisms affecting the structural integrit y of aging aircraft structures. Various nondestructive inspection (NDI) tec hniques are being used to obtain images of corroded regions on structures. A computational approach using wavelet transforms and artificial neural net works to analyze and quantify the extent of corrosion damage from the NDI i mages is described. The wavelet parameters obtained from the images were fi rst used to classify between corroded and uncorroded regions using a cluste ring algorithm. The corroded regions were further analyzed to obtain the ma terial loss due to corrosion using an artificial neural network model. Expe riments were carried out to investigate the developed methods for aircraft panels with engineered corrosion obtained from the Federal Aviation Adminis tration Validation Center in Albuquerque. The results presented indicate th at the computational methods developed for corrosion analysis seem to provi de reasonable results for estimating material loss due to corrosion damage.