Wavelet Transform And Neural Network Damage Vibration Detection In Composite Structures

In this paper, online detection of initial crack damage for a cantilever laminated composite plate is studied using the energy spectrum of the decomposed dynamic response signal by wavelet transform and the Artificial Neural Network ( ANN) .The finite element dynamic models of the cantilever intact and damaged plates are established. The piezoelectric smart materials bonded on the laminated composite plate are used as both actuator and sensor. Introducing plate crack into FEM model, variations of physical and modal parameters due to crack are discussed. The dynamic response signals are calculated.The sub-signal energy spectrum of dynamic response signal by wavelet transform and energy spectrum variations of intact plate and damaged plates are numerically simulated.Taking the energy spectrum of the decomposed wavelet signals of dynamic responses as the inputs of the artificial neural network, neural networks are designed for structural damage detection. The trained network is used to detect the placement and size of crack damage.The online damage detection of laminated composite plate can be implemented using the bonded piezoelectric materials and signal processing based on vibration analysis. The results show that it is much more sensitive than the existing methods for taking the energy spectrum of the decomposed wavelet signals of dynamic response signal by wavelet transform as the characteristics of damage detection, and it can detect extremely small cracks in composite plates.The methods for crack damage detection and the conclusions presented in this paper are useful for engineering practice.