Prediction of High-rise Building Main Settlement Based on LSTM Neural Network

In view of the nonlinear change characteristics of high-rise building settlement with time and other influencing factors during construction, this paper established a high-rise building main settlement prediction model based on long short-term memory neural network (LSTM) based on the monitoring data of the main settlement of a building in Baiyun District, Guangzhou, and compared it with the main settlement prediction model based on BP neural network. And the accuracy test is carried out. The results show that bias value and RMSE value of LSTM model are reduced by 88.89% and 20.55% compared with BP model respectively, which proves the accuracy and effectiveness of LSTM neural network algorithm in predicting the main settlement of high-rise buildings, and can provide scientific basis for the construction safety of the main structure of high-rise buildings.