基坑坡顶水平位移预测对基坑安全施工具有重要意义。以昆山市建滔广场花桥一期工程为例,结合BP神经网络模型和时间序列ARMA模型各自特性,建立了基于灰色关联度的线性组合模型预测其坡顶水平位移。实例分析表明组合模型可以在保证较好的预测精度的前提下,提高了预测的整体性,具有工程实用价值。 Prediction of horizontal displacement at foundation pit is of great significance to the safety construction of foundation pit. Taking the first phase project of Qiantao Square in Qiantao, Kunshan as an example, a linear combination model based on gray relational degree was proposed to predict the horizontal displacement at the top of rock slope, combining the characteristics of BP neural network model and time series ARMA model. The case study shows that the combination model can improve the prediction integrity under the premise of ensuring a good prediction accuracy and has engineering practical value.