Calibrating static measurement data from distributed fiber optics by the integration of limited FBG sensors based on the extended kernel regression method

Thanks to the rapid development of fiber optic sensors, the arrival of distributed sensors makes continuous and dense measurement possible for real structures. The distributed fiber optic sensors could provide a very high number of sensors along one single fiber, which is of great significance for structural health monitoring, especially for the detection of the damage position. However, the accuracy of data measured from fiber optic distributed sensors, such as strain and displacement etc., are sometimes not as accurate as those obtained from traditional fiber optic sensors such as fiber Bragg gratings (FBGs) and strain gauges. In this paper, to enhance the accuracy of static strain data measured from distributed fiber optic sensors, an extended kernel regression (EKR) method is applied to combine the distributed sensor measurement with those from four FBG sensors. These provide quite accurate static strain data; the strain values at locations where the FBGs are absent can therefore be predicted by the EKR method, which uses the data from distributed fiber optics as a biased model. The static experimental activities have been carried out in a laboratory, using a cantilever beam structure under different static loads. (paper)