APPLICATION OF A REFINED BP ALGORITHM BASED ELMAN NETWORK TO SETTLEMENT PREDICTION OF SOFT SOIL GROUND

The Elman network has characteristics of dynamic adaptation,quick convergence and high accuracy.With the consideration of the basic feature of the settlement of soft ground,an Elman network model for predicting the settlement of soft ground is established.Considering the limitation of classical BP algorithm,modified BP algorithm is used to train the network.Through application in real case,the model of the Elman network presented in this paper is proved to be reliable and applicable.