Damage detection in an aluminium plate using outlier analysis

This work presents a damage detection study of a curved aluminium panel using the statistical technique of outlier analysis, where the structure was investigated using ultrasonic Lamb waves. These waves were generated by large custom-made piezoceramic patches and were sensed by smaller piezoceramic sensors surrounding each patch. The patches were mounted using the viscoelastic material Araldite 2027 and damage was added to the plate in the form of drilled holes and saw cuts increasing in length. Outlier analysis was successfully able to distinguish between the damaged and undamaged cases with a very high classification rate of over 99%, and was also able to approximate the direction from the actuator patch in which the damage lay.