Experimental verification of intelligent fault detection in rotor blades

Abstract

The common idea of examining damage in mechanical structures through vibrational methods is analysed from a practical point of view. The method presented is based on a straightforward time domain approach and utilizes response signals in their purest form. Error magnification related to computational manipulations, e.g. FFT, or similar procedures, is avoided. The fact that the proposed method is straightforward and involves a minimum of signal processing, renders the approach attractive for use in practice, as an intelligent fault detection system. The method was applied to a clamped section of a helicopter rotor blade. Damage was induced by a mass increase (locally added weight) and a mass reduction (holes drilled into the structure), respectively. The experiments show that minimal effects on fundamental eigenfrequencies are reliably detectable without any a priori knowledge of the structure. A series of measurements with damage locations changed successively demonstrates nicely the effects on natural frequencies. In obtaining this result it became apparent that every case study exhibited its own individual vibrational signature which is very desirable in a health-monitoring environment.