Download files
Abstract
This research work deals with aspects concerned with delamination detection in composite structures as revealed by an approach based on vibration measurements. Variations in vibration characteristics generated in composite laminates indicate the existence of delaminations. This is because degradation due to delamination causes reduction in flexural stiffness and strength of a material and as a result, vibration parameters like natural frequency responses are changed. Hence it is possible to monitor the variation in natural frequencies to identify the presence of delamination, and assess its size and location for online structural health monitoring (SHM).
The approach in this thesis typically depends on undertaking the analysis of structural
models implemented by finite element analysis (FEA). FE models also known as the simulator are used to compute the natural frequencies for delaminated and intact specimens of composite laminates. The FE models are validated using the analytical model. However, these FE models are computationally expensive and surrogate (approximation) models are introduced to curtail the computational expense. The simulator is employed to solve the inverse problem using different algorithms based on computational intelligence concepts. An artificial neural network model (ANN) is developed to solve the inverse problem for delamination detection directly and to provide surrogate models integrated with optimization algorithms (the gradient based local search and Real-coded Genetic Algorithm (RGA)). This approach is termed as surrogate assisted optimization (SAO) and it is seen that the engagement of surrogate models in lieu of the FE models in the optimization loop greatly enhances the accuracy of delamination detection results within an affordable computational cost. It also provides control when handling different variables. Meanwhile, to aid the building of
effective surrogate models using substantial number of training datasets, K-means clustering algorithm is harnessed and this effectively reduces the large training datasets usually required for ANN network training. Response surface methods (RSM) are also developed to directly solve the inverse problem. The principal advantage of the RSM is its ability to give physical mathematical models that are used to identify the size and location of delamination given any input changes in natural frequencies.
A delamination detection strategy that uses K-means clustering algorithm for database
selection and ANN, RSM and optimization algorithms integrated with surrogate models
based on ANN have been successfully developed. It is demonstrated that these algorithms show immense potentialities for use in delamination damage detection scenarios when applied to composite beams and plates. The algorithms successfully performed delamination detection given limited amount of training datasets. Prediction errors of the algorithms were quantified and they were shown to be robust in the presence of artificial errors and noise and even when applied to experimental and simulation data. Results clearly indicate remarkable accurate delamination damage detection capability of the algorithms. The algorithms in their inverse formulations are capable of predicting accurately delamination parameters.
These algorithms should hence be employed for application in the domain of SHM where their small computational requirements could be exploited for online damage detection.