This paper investigates the feasibility of multilayer perceptron (MLP) for the classification of structural delamination and transducers debonding in smart composite laminates. Structural vibration response is employed to extract the discriminative features for multiple damages. The dynamic model of the smart structure with inter-ply delaminations and partially debonded piezoelectric sensor and actuator is developed by incorporating improved layerwise theory, higher order electric potential field and finite element method. The developed model is solved in the time domain to obtain the transient response of the healthy and damaged structures through a surface bonded
piezoelectric sensor for random input excitations applied through a piezoelectric actuator. The input-output information is fed into a system identification algorithm to identify damage sensitive features for the healthy and
damaged state of the smart composite laminate. The discriminative features are classified through MLP in a supervised manner and its classification accuracy is evaluated in terms of true positive (TP) rate, false positive (FP) rate, precision and area under the receiver operating characteristic curve (ROC area).
Kim, H.S., Chattopadhyay, A. and Ghoshal, A., (2004), "Dynamic Analysis of Composite Laminates with Multiple Delamination Using Improved Layerwise Theory," AIAA Journal, Vol. 41, No. 9, pp. 1771~1779
Phan, M.Q., J.A. Solbeck, and L.R. Ray (2004). A direct method for state-space model and observer/Kalman filter gain identification. in AIAA guidance, navigation, and control conference and exhibit, Rhode Island.
Sung D.-U., Oh J.-H., Kim C.-G., Hong C.-S., (2000) Impact monitoring of smart composite laminates using neural network and wavelet analysis, Journal of intelligent material systems and structures, 11 180-190.