Sensor data analysis, reduction and fusion for assessing and monitoring civil infrastructures

Structural health monitoring data analysis is basically a logical inference problem, wherein we attempt to gain information on the structural state based on sensor responses. In this chapter, we first introduce Bayesian logic as the main instrument to formulate the inference problem in rigorous mathematical terms, properly accounting for data and model uncertainties. Next, an overview of the most popular data reduction techniques is provided, with a special focus on principal component analysis (PCA). The chapter then introduces the concept of data fusion and discusses techniques to handle multi-temporal and multi-sensor data based on Bayesian statistics. Alternative non-probabilistic logical models for handling uncertainties are outlined at the end.