COMPARISON OF GENETIC ALGORITHM AND ARTIFICIAL NEURAL NETWORK FOR DAMAGE DETECTION OF COMPOSITE STRUCTURES

摘要

Comparison of genetic algorithm (GA) and artificial neural network (ANN) for quantitative assessment of delamination in composite laminates was conducted. For this purpose, a theoretical model was established to relate the delamination parameters with the alteration in structural frequency response functions (FRFs). Based on the model, an objective function was designed, and minimising such a function using GA led to the determination of delamination parameters; meanwhile a BP-feedforward ANN was developed and trained using the eigenvalues extracted from FRFs. For validation, glass fibre-reinforced composite (GFRC) beams containing various delaminations were examined and eigenvalues were obtained from measurement by embedded fibre Bragg grating (FBG) sensors. Both algorithms were then collated in terms of identification precision and computational cost.