An algorithm based on wavelet analysis for automatically estimating the location and magnitude of impact forces exerted on a rectangular carbon fibre-epoxy honeycomb composite panel is developed. The technique employs a single piezoelectric sensor mounted distant from the impact zone and presumes that an impact is applied at one of several pre-established locations. Furthermore, it is presumed that the recorded vibration response is the superposition of the simultaneous 'assumed' impacts at these locations, with the aim of simultaneously identifying the actual impact location and force magnitude through an under-determined regularisation scheme. The algorithm aims to detect the most probable impact location amongst the spurious locations. Since a normal impact introduces a narrow-band time-localised event with high energy, the wavelet transform is an effective tool to locate this event, with the wavelet coefficient representing how closely correlated the wavelet is with the reconstructed forces. The larger the coefficient is in absolute value, the greater the similarity. Two case studies are undertaken to validate the proposed method, one under-determined problem with four potential impact locations and the other having twelve potential impact locations. Further, the effects of various 'mother wavelets' on the performance of the algorithm are studied. The results demonstrate successful localisation and reconstruction of the impact force using both orthogonal and non-orthogonal wavelets.
