Research on application of wavelet analysis and RBF neural network to prediction of foundation settlement
Changdong Li · Rock and Soil Mechanics · 2008
Foundation settlement is a kind of severe environmental hazard. The monitoring data of foundation settlement are usually disturbed by rainfall and engineering construction. As a result,there are a lot of data breakpoints in the settlement curve. Therefore,based on wavelet analysis and RBF neural network theory,a new method for foundation settlement is proposed. Firstly,based on the de-noising process of monitoring data by wavelet analysis,the foundation settlement curve that is close to the practical situation can be obtained. Afterwards,the prediction is carried on by radial basic function (RBF) neural network method. The wavelet analysis and RBF neural network method can provide engineering design with scientific basis. Finally,based on the engineering instance analysis and the contrast study between different kinds of wavelets,the results show that the de-noising and prediction effect of triple B-spline wavelet is the best among the chosen wavelets,and it has a good future in the field of engineering application.