Damage Detection from Continuous Long-Term Static Response Using Cointegration and Mewma Control Chart

The operational and environmental variability (OEV) will mask subtle changes caused by damage for in-situ civil infrastructures. Thus, it is one of the obstacles that significantly hinder the application of structure health monitoring. An approach based on cointegration and control chart technique is proposed for identifying damage from continuous, long-term, static response with the presence of varying OEV. Cointegration theory is used as a data normalization technique to remove the stochastic common trend, which is induced by OEV, in the measured static response observed from several locations on the structure. The multivariate, exponentially weighted, moving average control chart is employed as a statistic metric to detect the presence of damage and then the drop-one-out strategy is implemented to localize damage. The proposed method is validated by a numerical simulation and monitoring response from a large-span bridge. Results demonstrate that this method can successfully identify damage despite the operational and environmental variations.