The knowledge of the external load acting on the structure under investigation is important for Structural Health Monitoring (SHM). The history of the external load can be used for updating the earlier lifetime prediction of the structure. An easy and direct way to get the external load information is to measure it directly using a certain type of force transducer. However, there are still many practical cases in which a direct measurement of the external load is physically or economically not feasible. For those cases, a possible solution is to reconstruct the external load using structural response measurements, e.g. displacement, strain, velocity, and acceleration. This process is defined as force reconstruction. It is a kind of inverse problem which often tends to be ill-posed, in the sense that the measurement noise and the modeling error can be amplified and can cause large deviations in the reconstructed force. Online force reconstruction is a research topic which studies how to realize force reconstruction in real-time. The thesis focuses on online force reconstruction.
The basic idea which is adopted in this thesis for online force reconstruction is to apply a real-time executable state and input estimation algorithm. The methodology for online force reconstruction consists of two stages, an offline stage and an online stage. In the offline stage, the modal parameters of the structure are identified, and a state-space model of the structure is constructed. In the online stage, a real-time executable state and input estimation algorithm is applied to provide an estimate of the force. The ill-posedness in the process of force reconstruction is relaxed by the convergence of the estimate error of the force. Together with the force, the structural responses at different positions can be reconstructed, too.
In this thesis, modifications to some of the available real-time executable state and input estimation algorithms are proposed, so that these algorithms are theoretically more suitable for online force reconstruction. In the earlier research studies, some algorithms have already been proposed for online force reconstruction. All the available algorithms, including the proposed modified algorithms, are potential candidates for online force reconstruction. In case there is a practical need for online force reconstruction, it is reasonable to raise the question which algorithm to choose. In this thesis, a study on application-oriented algorithm selection is performed. The assumptions and the mathematical conditions in the algorithms are translated into practical requirements. A benchmark study is performed in which a laboratory structure is taken as the benchmark structure. The modal parameters of this benchmark structure are identified using an Experimental Modal Analysis (EMA) technique. Two types of widely used sensors are installed on this benchmark structure, including strain gauge and accelerometer. Three different types of input forces are considered, including a quasi-static force, impact forces, and a wind load which is generated by an electric fan. In total eight different algorithms are tested. Based on the results from the benchmark study, an application oriented guidance for algorithm selection is extracted.
As an example of practical application, the presented methodology for online force reconstruction is applied for wind load reconstruction for the 600 meter tall Canton Tower. The field measurement data are recorded by the SHM system of the Canton Tower, including the data which were recorded during the Typhoon Nanmadol in 2011 and the Typhoon Kai-tak in 2012. The modal parameters of the Canton Tower during the Typhoon Nanmadol are identified by using an Operational Modal Analysis (OMA) technique. The available reduced-order Finite Element (FE) model of the Canton Tower is first modified to reflect the height adjustment of the Canton Tower, and then is updated according to the identified modal parameters. In consideration of the characteristics of the wind load and the sensor availability in the SHM system, a suitable algorithm is selected with the help of the application oriented guidance which is extracted from the benchmark study. The wind load on the Canton Tower during the Typhoon Kai-tak is reconstructed. The mean value of the wind load is calculated with the wind speed measurements and the aerodynamic force coefficients which are identified from a wind tunnel test. The dynamic part of the wind load is reconstructed by using the selected algorithm and the acceleration measurements. To validate the reconstruction results, two acceleration channels are selected. The acceleration measurements from these two channels are not used in wind load reconstruction. The strategy for the validation is to compare the reconstructed acceleration for these two channels with the real acceleration measurements. The validation results show that the reconstructed wind load is acceptable.
Finally, a summary of this thesis is provided, and some open topics are described. These open topics can be considered in future study on online force reconstruction.