SPATIAL FILTERING TECHNIQUE-BASED ENHANCEMENT OF THE RECONSTRUCTION ALGORITHM FOR THE PROBABILISTIC INSPECTION OF DAMAGE (RAPID)

Abstract

A common active structural health monitoring (SHM) solution for thin-walled structures consists of processing ultrasonic guided waves signals, which are excited and sensed by means of a network of piezoelectric devices installed on the structure, with the purpose of providing a tomographic reconstruction-based damage probability map of the structure. A promising reconstruction algorithm typically employed within this framework is the Reconstruction Algorithm for the Probabilistic Inspection of Damage (RAPID) algorithm, which has been shown to provide satisfactory results in terms of damage detection and localisation. However, this algorithm comes with some disadvantages and minor issues, such as artefacts creation in case an unevenly distributed sensors layout is installed on the structure, which may significantly worsen the damage diagnosis performance of the monitoring framework. In this paper, an enhancement of the original RAPID algorithm is presented, which exploits spatial filtering techniques to reduce possible artificially created artefacts, thus allowing installing on structures any network of sensors without reducing the diagnostic performances. The improved damage localisation accuracy obtained using the proposed algorithm is proven by means of a case study involving a numerical model of a realistic composite panel with an unevenly distributed network of sensors.