Abstract: The settlement prediction through the observed settlement with time data have a wide application in the high speed rail embankment construction. The settlementtime relationship of an embankment is a reverse "S" shape. The conventional settlement prediction method using the observed data have the limitations from the model and from the sampling data. The predicted settlement can often not reflect the practical situation. An optimized model which combines the contribution of different reverse "S" shape models is proposed in this paper. Different weights are assigned to the different sample models through the optimization process. The settlement obtained through this method has a better prediction result and can be applied in the settlement prediction of high speed rail embankments.

Key words: Settlement Prediction, Reverse "S", Shapes, Optimization Model, WuhanGuangzhou HighSpeed Rail, Red Clay Embankment