沉降预测的多层递阶时间序列模型研究  

Study on multi-layer recursive time series model for settlement prediction

作　　者：张仪萍[1] 王士金[1] 张土乔[1] 

机构地区：[1]浙江大学土木工程学系,浙江杭州310027

出　　处：《浙江大学学报（工学版）》2005年第7期983-986,共4页Journal of Zhejiang University：Engineering Science

基　　金：国家自然科学基金资助项目(50278089);浙江省自然科学基金资助项目(501163).

摘　　要：为了研究沉降模型中参数的时变特性,提出了多层递阶时间序列模型,该模型将沉降预测分为预测模型参数的预测和在此基础上的沉降预测两部分.采用参数递推公式计算沉降时间序列模型的参数序列,并根据计算出的参数序列构建其时间序列预测模型.依此类推,可建立多层参数的预测公式,直至认为参数不随时间变化为止,然后根据参数预测结果来预测后期沉降.实例分析和计算表明,多层递阶时间序列模型能较好地反映参数的时变特性,取得较精确的沉降预测结果.In order to study the time-dependent characteristics of the parameters in the settlement prediction model, A multi-layer recursive time series model was presented. Prediction of settlement was divided into two parts： the prediction of time-dependent parameters, and the prediction of settlement based on the parameters＇ prediction. The parameters of the settlement time series prediction model were calculated through a recursive formula. Based on which the parameters yielded, the parameters time series prediction models were presented. By analogy, the multi-layer parameter prediction models could be developed until the parameters were time-independent. Lastly, based on the parameter predicting results, the post settlement was calculated through the time series model. The case result shows that the multi-layer recursive time series model is effective in analyzing the time-dependent characteristics of parameters, and has considerable prediction precision.

关 键 词：时变参数 多层递阶 时间序列 (Aufo—Regressive)AR模型 

分 类 号：TU451[建筑科学—岩土工程]