Research on temperature influences in long-span bridge eigenfrequencies identification

Aimed at the problem that consecutive time series of eigenfrequencies identified from ambient exciting bridge structure responses are always influenced by temperature, a problem being paid much attention by bridge monitoring and health estimation applications, a research work with modeling eigenfrequency and temperatures was performed upon long term acquisitions of vibration and temperatures from the Ting Kou Bridge in Hong Kong. The research results show that eigenfrequencies vary not only followed atmospheric temperature fluctuation but also associated remarkably with the temperature distribution patterns over the structure. To eliminate temperature complications from eigenfrequency time series, an approach based on SVM nonlinear modeling algorithm was proposed in the paper. The effect of its application of 600 hours test data from Ting Kou Bridge was also described.