Improvement of GPS displacement measurement accuracy for high-rise buildings by machine learning

Abstract

This paper aims to enhance the accuracy of displacement responses of high-rise buildings measured by global positioning systems (GPS) under windstorm conditions with machine learning. To this end, a long short-term memory model (LSTM) and a particle swarm optimization-LSTM (PSO-LSTM) model are developed. The models are trained and validated by using field measurements, including strain and acceleration responses, original GPS-measured signals from a 600-m-high skyscraper during Typhoon Kompasu, and the performance of the models is assessed against several standard evaluation criteria. Then, the PSO-LSTM model with optimized hyperparameters is utilized to enhance the accuracy of displacement measurements by the GPS in the skyscraper during Tropical Storm Lionrock. The results of this study reveal that the developed PSO-LSTM model can effectively improve the accuracy of GPS-measured displacements of high-rise buildings under windstorms conditions.