EXPERIMENTAL VALIDATION OF A STRUCTURAL HEALTH MONITORING METHODOLOGY: PART I. NOVELTY DETECTION ON A LABORATORY STRUCTURE

Abstract

This paper is concerned with the experimental validation of a structural health monitoring methodology, previously only investigated using synthetic data. The structure considered here is a simplified model of a metallic aircraft wingbox i.e., a plate incorporating stiffening elements. Damage is simulated by a saw-cut to one of the panel stringers (stiffeners). The analysis approach uses novelty detection based on measured transmissibilities from the structure. Three different novelty detection algorithms are considered here: outlier analysis, density estimation and an auto-associative neural network technique. All three methods are shown to be successful to an extent, although a critical comparison indicates reservations about the density estimation approach when used on sparse data sets.