Open Access Open Access  Restricted Access Subscription or Fee Access

Damage Localization with SP2E Under Changing Conditions



The central concept of vibration-based structural health monitoring (SHM) is the computation of damage sensitive features like natural frequencies or model residues. These parameters are typically observed and evaluated from a statistical point of view following the statistical pattern recognition paradigm. It is of common knowledge that those features are not merely effected by local structural alterations, but also by environmental or operational conditions (EOC). To distinguish or eliminate these, many data normalization strategies exist for linear and even non-linear cases. Besides damage detection under environmental and operational variability, the determination of actual damage positions is desirable for large and remote engineering structures such as onshore and offshore wind turbines or bridges. The recently introduced state projection estimation error (SP2E) method appears to be a promising approach towards damage localization of mechanical systems. This procedure is purely data-driven (output-only) and therefore does not rely on updated numerical physical models (e.g. FE models). It is based on identified, parametric, linear time-invariant (LTI) systems, state estimation and advanced projection techniques. The method significantly differs from conventional modal approaches e.g. the comparison of mode shapes or curvatures. Although laboratory experiments proved a high sensitivity against local structural changes for some systems, damage localization with SP2E under varying conditions has so far not been investigated. This article deals with this important issue. Therefore, simulations of a linear parameter varying (LPV) system were conducted in the healthy and locally altered state. The principal component analysis (PCA) was then applied to distinguish between local and global changes and successfully locate the damage positions. As an important step for validation, these investigations were complemented by a laboratory test. This research helps to improve the understanding of the sensitivity of the SP2E method towards varying environmental and operational conditions. Additionally the applicability of linear data normalization strategies with respect to damage localization was tested. It therefore lays a foundation for future field applications.


Full Text: