Application Of Time Series And Bayesian Model In Settlement Prediction

At present, the most settlement prediction models have low precisions, and can't give a good estimation scope of the settlement.Firstly, this paper describes the significance and the engineering background for settlement prediction, and summarizes the art of state of settlement predication. Secondly, the new predication model to oercome the shortcoming of the existing models is put forword.The main research results are as follows:(1) The Multi-layer Recursive Time Series Settlement Prediciton Model, which is based on time series model and multi-layer recursive theory, is put forward to enhance the prediction precision of the settlement predication.(2) The time-dependant parameters in the new model are predicated in the second layer, and the parameters identification is studied.(3) At present, the results predicted from most predictions models just give one special value, and the settlement uncertainty scope is unknown. In order to solve this problem, based on Bayesian Predication Model, the prediction mean value, uncertainty scope and probability of the settlement are studied.(4) The case shows that the prediction model developed in this paper has higher precision, and probability analyses for the prediction value will guides actual project well.