Damage location in a stiffened composite panel using Lamb waves and neural networks

Neural networks have proved to be very powerful tools in pattern recognition and machine learning and have consequently seen a great deal of applications in Structural Health Monitoring; a field where Pattern Recognition is one of the main lines of attack. The current paper presents a case study of damage detection and location in a stiffened composite panel interrogated by ultrasonic Lamb waves. Rather than work directly on features extracted from the wave profiles, the proposed approach derives secondary features in the form of a vector of novelty indices for the plate. This can be used to train both neural network classifiers and regressors and the use of both for damage location is demonstrated in the paper.