基于BP神经网络的建筑物沉降预测——以泉州市东海湾某建筑项目为例

Settlement Prediction of Buildings Based on BP Neural Network——Taking a construction project at East Bay in Quanzhou City as an example

利用人工神经网络强大的学习能力,提出了基于BP人工神经网络的建筑物沉降预测方法.以泉州市东海湾某实例工程1~12期的沉降观测数据为基础,建立网络模型.将13~16期建筑物沉降的实测数据和模型的预测数据进行对比,发现两者间的误差相对较小,证明BP神经网络预测模型具有较高的精确性和稳定性,且具有一定的工程应用价值.

Here we proposed a prediction method for the settlement of buildings based on BP artificial neural network with its strong nonlinear mapping and learning ability. In this paper we built a net- work model with the settlement observation data from phase 1 to 12 of a construction project in Quanzhou City as a foundation, and compared the values actually observed with those we predicted from phase 13 to 18,and the error between the two was relatively small.The results proved the high accuracy and stability of BP neural network prediction model and is reasonably reliable.