Support Vector Machine based Online Composite Helicopter Rotor Blade Damage Detection System

This work explores a feasibility of using vibratory hub loads and support vector machine (SVM) to predict damage and hence the life consumption of the composite helicopter rotor blade. Generally, the initial part of the composite's life is dominated by matrix cracking; the intermediate part by debonding/delamination and the final failure due to fiber breakage. The simulated hub loads under various damage levels are obtained using a comprehensive aeroelastic analysis of the composite rotor blade with physics based damage modes and are then linked with the life consumption of the blade using a phenomenological model. The SVM is used for classification of the useful life of the blade into three classes which are useful to decide the prognostic action. The performance of the blade damage detection system is demonstrated using simulated hub loads obtained using a two-cell airfoil section representing the stiff-inplane blade. The model based hub load variations are contaminated with noise to simulate the real data. It is observed that the SVM based damage detection system is more robust, reliable and easy to implement than the rotating frame measurement based methods.