Predicting the variability of natural frequencies and its causes by Second-Order Blind Identification

Abstract

Structural aging, degradation phenomena, and damage due to hazardous events are common causes of failure in civil structures and infrastructures. The increasing need of extending the structure lifespan for sustainability and economic reasons motivated the rapid development of remote, fully automated structural health monitoring systems. Different approaches have been developed for damage detection based on the incoming data. Modal-based damage detection is probably one of the most popular procedures for structural health monitoring of civil structures, also thanks to the development of robust automated operational modal analysis algorithms in the last decade. However, the sensitivity of modal parameter estimates and the associated damage features to environmental and operational factors represents a significant drawback to the extensive application of this technology. Thus, effective damage detection cannot skip the preliminary compensation of the effect of those variables on modal properties. Different approaches to compensate the environmental influence on modal property estimates are reported in the literature. In this article, the use of Second-Order Blind Identification is proposed. It is applied to a number of case studies in order to validate its effectiveness in the presence of one or more environmental or operational variables. Results demonstrate that it can model the variability of natural frequency estimates in operational conditions and, above all, it can give a fundamental insight in determining the causes of such variability.

Long-term monitoring for the condition-based structural maintenance of the Milan Cathedral

“…The paper focuses on the description of As a final remark on the lessons learned from the first months of continuous monitoring, it should be stated that the future vibration-based SHM of the monument will involve: (a) the use of both supervised (such as multiple linear regression [33]) and unsupervised (principal component analysis [34] or second order blind identification [35]) algorithms to remove/mitigate the masking effects induced by the temperature changes on the identified resonant frequencies and (b) the check of the invariance of mode shapes and mode complexity [21]. In addition, the availability of several quasi-static measurements should conceivably be exploited in a data fusion framework for an enhanced program of condition-based structural maintenance.…”

Frequency domain decomposition-based multisensor data fusion for assessment of progressive damage in structures

“…An extensive amount of research in SHM focuses on understanding the effect of temperature and operational changes on the structural responses, in particular on the dynamic behavior. For example, Rainieri et al proposed the use of Second‐Order Blind Identification to model the variability of modal features in presence of changes in environmental and operational conditions. This allows to extract the features that are sensitive to damage but rather insensitive or less sensitive to environmental and operational conditions.…”

Damage Identification in Structural Health Monitoring: A Brief Review from its Implementation to the Use of Data-Driven Applications

“…Structures in operation are subjected to the influence of operational and environmental conditions. These conditions affect the structures causing degradation, aging, and damage [42]; they are also the cause of possible false detections in SHM systems, due to the sensitivity of the methods to operational and environmental variables (EOVs) [43,44]. For that reason, the influence of these variables must be considered in the development of a reliable SHM system [45].…”

“…Rainieri et al [42] show the application of the second-order blind identification method (SOBI) to predict variations in natural frequencies, which makes it possible to compensate for the environmental influence on the use of OMA, this being a limitation of the PCA analysis. The implementation carried out requires a relatively low computational effort and obtaining a linear model between natural frequencies and unknown EOV sources.…”