Improvement of axial deformation prediction in high-rise buildings with field monitoring and adaptive unscented Kalman filter
Yun Zhou, Xianming Luo, Wenjie Zhang, Peng Ye, Jiahao Chen, Zong Du
Abstract
The deformation of vertical reinforced concrete members in high-rise buildings during their construction is constantly changing due to the complexity of the deformation mechanism , as well as the uncertainties that arise during the construction process ; in some cases, accurate estimation of the vertical displacement is crucial. Based on on-site strain monitoring data and an adaptive unscented Kalman filter , this paper establishes a state equation for the axial deformation process . The elastic, inelastic and temperature deformation of concrete during the entire construction process are reasonably considered. Deformation is predicted using both the traditional Kalman filter and unscented Kalman filter, both of which achieved satisfactory results. But the unscented Kalman filter can improve the prediction accuracy and robustness by utilizing unscented transformations. Anomaly detection is integrated into every inference step to detect abrupt changes in external conditions and dynamically adjust calculation parameters, thereby reducing the dependency of the prediction results on input parameters.