融合改进GM(1,1)与神经网络的地铁基坑变形预测应用研究

An Application Study of the Improved GM(1,1) and Neural Network for the Deformation Prediction of a Subway Foundation Pit

城市地铁工程不断增长,基坑的开挖也越来越多,如何控制基坑对周边建筑物的影响已成为地铁施工关注的热点和难点。采用融合改进的GM(1,1)模型和神经网络,对基坑变形监测时间序列进行预测处理,同时将预测结果与改进的GM(1,1)、神经网络预测模型进行了定量的比较和分析。结果表明,提出的融合预测模型预测精度更高,为基坑变形监测预测提供了技术支持和参考。

As more and more urban subway projects are to be built, there are also more and more foundation pits to be dug.How to control the effect of the foundation pits on their surrounding buildings becomes the hot point and difficult point that attracts much attention of the public when a subway is to be built. In such a case,the integrally improved GM（1,1）model and the neural network are used to predict and process the time series of the deformation monitoring of the foundation pit,and meanwhile the predicted results are quantitatively compared with those obtained from the improved GM（1,1） and the neural network prediction models in the paper.The results of the comparison and analysis show that the prediction of the integrated prediction model proposed in this paper is more accurate, which provides technical support and useful reference for monitoring and predicting the deformation of foundation pits.