Recent field measurements on long-span bridges during typhoon events have captured strong nonstationary features in the buffeting responses. In this study, the buffeting responses of Sutong Bridge during Typhoon Haikui in 2012 recorded by structural health monitoring system are analyzed to represent the nonstationary characteristics. As an accurately measured state variable, the acceleration response of the main girder is first selected to evaluate its own original stationarity in different time intervals using the run test method. The acceleration response of the main girder can be regarded as a zero-mean nonstationary random process which is in demand to extract its transient features in time–frequency domain. Hence, the evolutionary power spectral density (EPSD) of the acceleration responses, which can present the local turbulence energy distribution in both frequency and time domains, is estimated using the wavelet-based method. Also, an average wavelet spectrum is obtained by averaging the square values of wavelet coefficients along the time axis, and the comparison between the average wavelet spectrum and Fourier spectrum shows a great conformance which indirectly verifies the validity of the obtained evolutionary power spectral density. The results of this study exhibit that there are strong nonstationary characteristics existing in the buffeting responses of Sutong Bridge during Typhoon Haikui, and it is essential to incorporate the nonstationary features of winds in the analysis or design of long-span bridges from an aerodynamic viewpoint.
Power spectral density analysis of relative comb-line phase jitter in a twin-soliton molecule
