Detection of wind-induced response by wavelet transformed GPS solutions

Global positioning system (GPS) solutions offer direct measurements for the dynamic monitoring of long-span and tall civil structures. Timely and correct interpretation of the solutions is essential to improve monitoring and reduce the number of false alarms. In this article, a new approach has been developed that employs the principal component analysis (PCA) of wavelet transformed GPS solutions to detect out-of-control signals. The methodology consists of prefiltering the original GPS solutions via finite infinite response median hybrid (FMH) filter and applying the PCA to Haar wavelet transform (HWT) of the FMH-filtered results. It has been tested off-line using observation data from a tall building monitoring experiment involving GPS and multisensors. Observation data were collected for about seven days in February 2000, including two strong wind periods. The PCA control chart of the wavelet transformed solutions was found to respond better to the wind periods as compared with that of the "raw" solutions. The proposed method can be applied to monitor wind/earthquake or thermally induced responses in the long-term deformation monitoring of large engineering structures by GPS.