Structural damage detection in a helicopter rotor blade using radial basis function neural networks

A neural network approach is used for detection of structural damage in a helicopter rotor blade using rotating frequencies of the flap (transverse bending), lag (in-plane bending), elastic torsion and axial modes. A finite element method is used for modeling the helicopter blade. Radial basis function (RBF) neural networks are used and several combinations of modes are investigated for training and testing the neural network. Using the first 10 modes of the rotor blade for damage detection yields accurate results for the soft in-plane hingeless rotor considered in this study. Using a parametric study of the blade rotating frequency in conjunction with the neural network, it is found that a reduced measurement set consisting of five modes (the first two torsion modes, the second lag mode and the third and fourth flap modes) also gives good results for damage detection. Furthermore, taking only the first four flap modes also results in good damage detection accuracy. Three rotating frequency sets are therefore identified in this paper for structural damage detection in a helicopter rotor using RBF neural networks.