Advanced feature extraction for the characterization of impact events on composites using multiresolution wavelet-based simulations

Abstract

The detection and characterization of impact events on composite structures from the induced wave responses, involves many challenges. Coexisting guided waves of high dispersive nature propagate after impact, entailing continuous time variations in their wavenumber, frequency and group velocity content, which complicate the exact estimation of the time-of-flight and the extraction of impact characteristics. A multiresolution finite wavelet domain computational method, combined with appropriate contact laws, is presented to provide enhanced simulation and characterization capabilities of impact events. An explicit multiresolution time integration scheme involves a coarse solution, followed by finer solutions that are sequentially added to the coarse one, until convergence is achieved. The hierarchical character of the approach makes the impact simulation very fast and accurate. Moreover, due to the filtering properties of Daubechies wavelet functions, each resolution component effectively models and isolates specific wavenumber spectra, providing the capability to separate coexisting wave modes, and to overcome difficulties imposed by the dispersive nature of the resultant guided waves. It is demonstrated that the multiresolution simulation can reveal wave characteristics and time-of-flight features that no other traditional single-resolution method can do, and provides the basis for the development of powerful inverse methods that can localize the impact and characterize the impactor parameters. The method is first applied on impacted composite strips and the structural responses of each resolution component are assessed in order to obtain the desired features for impact location estimation and characterization. Finally, the method is implemented on an impacted composite plate structure and the various wave characteristics simulated by the fine components are presented, evincing the advanced features that the proposed method provides.