Hits: 474
On 31st October 2018, Dr Kyriaki Gkoktsi (AKTII) presented her work during an event organised by the SECED Young Members’ Group. The event, hosted at the Institution of Civil Engineers, was on Vibration, Offshore, and Blast Engineering. Kyriaki provided this article presenting aspects of her research work on operational modal analysis and signal processing.
This article presents a novel sub-Nyquist spectral estimation approach that can be embedded into smart wireless sensors for low-power structural health monitoring applications. The proposed method, termed multi-sensor power spectrum blind sampling approach, relies on arrays of wireless sensors that operate in a deterministic non-uniform in time-domain multi-coset sampling scheme, capable of acquiring compressed structural acceleration response signals at sampling rates below Nyquist. The acquired compressed measurements are transmitted to a central server and collectively processed using a power spectrum blind sampling (PSBS) technique to estimate the power spectrum density matrix of the acceleration responses. Acceleration measurements are treated as realisations of a stationary random process without making any assumption on the signal structure (e.g., sparsity). Structural modal properties are then extracted through standard frequency domain decomposition (FDD). The efficacy of the proposed approach is numerically assessed at a high signal compression level using field-recorded data obtained from a monitored bridge in Zürich, Switzerland. It is shown that accurate mode shapes can be obtained from 89% fewer measurements compared to conventional non-compressive FDD, leading to significant energy gains within battery-operated wireless sensors. It is envisaged that the proposed multi-sensor PSBS approach would support more sustainable long-term wireless structural health monitoring systems once sensors with multi-coset sampling capabilities become commercially available.