Skip to Main Content
Wavelet multi-resolution (WMR) and a bandpass digital Butterworth Alter were both proposed to extract structural vibration signal embedded in noisy GPS observables. The noisy observables were simulated by artificially adding random noises to a de-noised GPS time series with the sampling rate of 5 Hz. Modal parameters such as frequency, damping ratio, and modal shape coefficient were identified by using the random decrement technique and auto regressive moving averages (ARMA) model. The simulated results indicate that the filtered GPS time series can accurately capture the frequency, but may difficult in identifying the damping ration and modal shape. The false modal (or noise modal) may easily be produced due to the noise and low sampling rate of GPS observables, thus, the frequency should be determined under the consideration of modal shape.