Bridge structural health monitoring involves the collection and analysis of systematic measurements obtained from installed sensors on the bridge. While there are many quality certification systems available for civil engineering systems, e.g. American Society for Testing and Materials (ASTM), typical bridge monitoring specifications do not explicitly address performance requirements. As the applications of bridge structural health monitoring becomes more varied and advanced it will become necessary to ascertain the quality of measured data in order to check its suitability for new and varied applications. The quality of measured data has a direct impact on the results obtained from data analysis in structural health monitoring. The quality of measured data for a bridge monitoring system should meet three basic criteria: First, measured data should be devoid of anomalies such as signal clipping, intermittent noise spikes, temporary signal dropouts, and spurious trends. Second, the errors, including aliasing and quantization, and system noise should be within an acceptable range. Lastly, the measured data should satisfy required assumptions, such as stationarity and normality. Data qualification is defined as validating the quality of measured data and quantifying errors and noise to ensure the measured data sufficiently satisfies the above mentioned criteria. Despite the tremendous impact the quality of measured data has on bridge structural health monitoring, data qualification has not been formalized for bridge monitoring to provide a basis for bridge monitoring data specifications. This paper proposes measures of data qualification that can be used as a foundation for data specifications in bridge structural health monitoring. This paper will use data collected from monitored bridges in the Connecticut Long Term Bridge Monitoring program as examples of issues faced when using long term bridge monitoring data acquisition systems.
展开▼
