Influence of the Autoregressive Model Order on Damage Detection

Abstract

An important step for using time-series autoregressive (AR) models for structural health monitoring is the estimation of the appropriate model order.To obtain an optimal AR model order for such processes, this article presents and discusses four techniques based on Akaike information criterion, partial autocorrelation function, root mean squared error, and singular value decomposition. A unique contribution of this work is to provide a comparative study with three different AR models that is carried out to understand the influence of the model order on the damage detection process in the presence of simulated operational and environmental variability. A three-story base-excited frame structure was used as a test bed in a laboratory setting, and data sets were measured for several structural state conditions. Damage was introduced by a bumper mechanism that induces a repetitive impact-type nonlinearity. The operational and environmental effects were simulated by adding mass and by changing the stiffness properties of the columns. It was found that these four techniques do not converge to a unique solution, rather all require somewhat qualitative interpretation to define the * To whom correspondence should be addressed. E-mail: farrar@ lanl.gov optimal model order. The comparative study carried out on these data sets shows that the AR model order range defined by the four techniques provides robust damage detection in the presence of simulated operational and environmental variability.

Structural health monitoring of bridges: a model-free ANN-based approach to damage detection

“…The former uses a system model to study a physical structure and predict the responses; examples of methods that fall into this category are the probabilistic, non-parametric and autoregressive models (AR) methods [11]. The physically based methods perform damage identification by comparing natural frequencies and mode shapes data between the healthy and damaged structural model [12]; examples of methods that fall into this category are the vibration-based damage identification methods (VBDIM), such as frequency response [13], mode shape and strain energy methods [14], modal flexibility and modal stiffness methods.…”

An improvement on feature extraction via time series modeling for structural health monitoring based on unsupervised learning methods

“…its orders and parameters) used in the normal condition will no longer provide a good fit and correctly predict the response of the damaged state. Therefore, the residual samples associated with this state will increase [16]. In this case, the increase in the model residuals is an indicator of damage occurrence.…”

“…From the engineering aspect, an improper order does not allow the time series model to capture the underlying dynamics of structure, which may lead to extract insensitive features to damage and weak detectability of damage [16]. Therefore, a reliable feature extraction via time series modeling depends strongly on obtaining an accurate and sufficient order.…”

Are Today’s SHM Procedures Suitable for Tomorrow’s BIGDATA?

“…Although, AR modeling is computationally less intensive [21], presents examples of application of SVR and ARX models that outperformed AR models in damage localization. This comparison was performed at a pre-specified model order and did not include the model order selection attempts [23], however it establishes an overall behavior of computation costs for these methods. The results imply these model-free methods are adoptable to analyze large SHM datasets; however, certain improvements are necessary to further reduce this computation time, especially in case of models with higher complexity.…”