论文部分内容阅读
为探索人工神经网络用于建筑物变形预报的可能性，采用BP网络代替传统的回归分析方法，处理变形观测数据．结果表明，当影响建筑物变形的因素较多，且各因素间存在非线性关系或关系不确定时，回归分析预报的结果存在较大偏差，BP网络有助于提高建筑物变形预报的精度．
In order to explore the possibility that artificial neural network can be used to forecast building deformation, BP network is used instead of traditional regression analysis method to deal with deformation observation data. The results show that when there are many factors that affect the deformation of buildings, and there is a non-linear relationship between the factors or the relationship is uncertain, the results of regression analysis have big deviations. BP network can help to improve the accuracy of building deformation forecasting.