Kinematic measurement and processing strategy for dynamic monitoring of engineering structures using global positioning system

Abstract

Demands in engineering structures protection against destructive stimulations have lead to a significant research in this area. In particular, the non-destructive evaluation sensor such as the Global Positioning System (GPS) is valuable to assess the serviceability, safety and integrity of these structures such as tall buildings and bridges. Nevertheless, the attainable accuracy of the GPS measurements is dependent upon the presence of errors or noises in the measurements. These include satellite and receiver clock errors, satellite geometry, satellite orbit, multipath and atmospheric errors. Some of the errors can be eliminated or minimised by applying differencing techniques, but others require particular attention if a high accuracy result is sought. This thesis explores the development of an integrated methodology and systematic processing for kinematic GPS method in continuous structural monitoring applications. The research presented here reinforced the theory of spectral representation of the signal, which was used to recover the signature of the disturbed signature from the priori signature. This method works when there were at least two sets of measurements from the so-called fixed and moving stations. It has been justified in this research that as these stations are closed together, they are strongly correlated with respect to GPS error sources and thus cancel some of the errors. A correlation coefficient between stations of up to 0.831 has been obtained in this study. By deriving their signatures using the Fast Fourier Transform, a method of minimising these spatial correlated errors by signatures differences and displacement detection with the aid of Kalman filter method has been developed. The developed technique is validated through a simulation test and real applications on a tower block and cable-stayed bridge. The test has demonstrated the potential of this technique for the improvement of observed values. Results have shown that an increase of almost 50% to 60% in position estimates was achieved by applying this technique. This can be interpreted by the Root Mean Square Error (RMSE) of simulated displacement in longitudinal direction with respect to true displacement has decreased from +0.004 m to +0.002 m by processing through the developed technique. Similarly, for vertical direction, the RMSE has decreased from +0.009 m to +0.006 m. The test conducted on the special designed simulation device shows that the responses with tip displacement of 1cm can be detected by kinematic GPS when compared with the actual displacement. A tower and cable-stayed bridge trials indicate the ability of the developed technique to detect displacement of more than 3 cm. The comparative results in the case of simulation study and real trials on structures proved that the proposed technique can enhance displacement measurement accuracy and capable of assessing the allowable safety tolerance of the engineering structures.