Passive sensing of low velocity impact damage on composite panels in simulated environmental and operational conditions

During its lifetime, an aircraft structure is subjected to impacts from a wide range of sources (e.g. bird strike, tool drop, hail). For composite structures, low velocity impacts may generate Barely Visual Impact Damage (BVID) that is difficult to assess without complex inspection techniques and may significantly reduce the strength of the structure. To reduce the need for costly inspection techniques, there is interest in developing integrated Structural Health Monitoring (SHM) systems to monitor the occurrence of impacts and provide location and severity estimations to allow the operators to make maintenance decisions (e.g. repairs). Many methods have been developed to achieve this using information encoded in impact induced strain waves recorded by sensors attached to the structure. However, these so-called passive methods have mostly been developed in laboratory conditions and do not consider operational and environmental in-service conditions which may significantly influence the accuracy and feasibility of these estimates. The research reported in this thesis was conducted to address some of the main challenges associated of with passive SHM application for in-service conditions, specifically accuracy, robustness, reliability and feasibility. A data driven approach was taken as it allows flexibility in scaling up from simple coupons to complex structures. Novel signal processing and reference database methods were proposed for accurate and robust impact location and maximum impact force estimation which only requires data from a single reference impact case, making it feasible for the variable conditions in service. A novel kriging based extension to these methods was developed to allow uncertainty quantification that provides more reliable information to the operator. A multifidelity approach to constructing the reference database for the data driven approach was also developed to significantly reduce the requirement of experimental sampling with alternative data sources and increase feasibility. Lastly, ...