摘要
在构建并联组合模型进行变形预测时,单项模型权值的确定是个关键问题。为了提高变形预测的精度,以基坑监测数据为例,采用GM(1,1)模型与ARMA模型进行组合,在拟合误差平方和最小的准则下,使用粒子群算法求解两单项模型的最优权值,进而构建并联组合模型进行变形预测。结果表明,该方法融合各单项模型的优势,可以提高预测精度,避免求解线性规划问题,具有较好的实用性。
In a parallel combined model being built to predict deformation, one of a key issue is to determine the weights of individual model. This paper takes the deformation data of foundation pit as an example, by using GM (1,1) model and ARMA model to integrate a combination model and adopting the particle swarm optimization to search the optimal weights of two single models under the principle of minimum fitting error sum of squares. The results show that the method can integrate the advantages of each individual model to improve prediction accuracy, without solving linear programming problems, and can be practical.
出处
《测绘工程》
CSCD
2017年第1期73-76,共4页
Engineering of Surveying and Mapping
基金
国家自然科学基金资助项目(41274020)