Statistical Modeling of Time Series for Ice Accretion Detection on Bridge Cables

Abstract

In the northern regions, ice accretion on bridge cables poses serious risks to the structure as well as to vehicular traffic when ice falls from the cables onto the road. The detection and quantification of ice formation allows the anticipation of ice falls and increases the safety of the structures. In this paper, an ice accretion detector was developed on the basis of the statistical modeling of vibration response signals. Three different methods were tested on the acceleration signals obtained from a bridge cable. The methods included the Fourier transform analysis, the autoregressive model, and the continuous wavelet transform analysis. Damage-sensitive features (DSF) were extracted from these models and tested with the data collected from a laboratory experiment conducted in a climatic wind tunnel. It was found that all three DSFs were correlated to ice accretion. The wavelet-based DSF had the highest correlation, resulting in the largest change of the DSF value and in the smallest estimation error.