摘要
针对城市轨道交通工程基坑沉降监测数据长期、非等时距的特点,构建了一种基于滑动窗口的非等时距改进GM(1,1)模型。基于滑动窗口更新GM(1,1)模型变量,充分利用GM(1,1)模型适用于短期预测的优势;利用时距信息对累加生成序列赋权,削弱非等时距观测产生的影响,从而提高模型预测精度。通过对某城市轨道交通工程基坑沉降监测数据进行预测分析,发现改进GM(1,1)模型预测结果的均方根误差与平均相对误差分别为1.12 mm和9.25%,相较于传统GM(1,1)模型分别提高约32%和21%,验证了改进GM(1,1)模型能够更好地满足基坑工程沉降预测的应用需求。
According to the characteristics of long-term and unequal time interval of foundation pit settlement monitoring data in urban rail transit engineering,an improved unequal time interval GM(1,1)model based on sliding window was constructed.By using the sliding window to update variables of the GM(1,1)model,and weighting the accumulated generation sequence according to the time interval information,the characteristics of GM(1,1)model suitable for short-term prediction are fully utilized,the influence of unequal time interval is weakened,and the prediction accuracy of the model is improved.Through the prediction and analysis of the monitoring data of the foundation pit settlement of an urban rail transit project,it is found that the root mean square error and the average relative error of the prediction results of the improved GM(1,1)model are 1.12mm and 9.25%,respectively,which are about 32%and 21%higher than that of the traditional GM(1,1)model.The experimental results show that the improved GM(1,1)model can meet the application requirements of the foundation pit settlement prediction.
作者
肖玉泰
范士杰
臧建飞
牟春霖
刁瑾
XIAO Yutai;FAN Shijie;ZANG Jianfei;MU Chunlin;DIAO Jin(College of Oceanography and Space Informatics,China University of Petroleum(East China),Qingdao 266000,China;China Railway Design Group Co.,Ltd.,Tianjin 300251,China)
出处
《测绘工程》
2024年第6期62-67,共6页
Engineering of Surveying and Mapping
基金
国家自然科学基金资助项目(42304024)
中国石油大学(华东)自主创新科研计划项目(22CX06034A)。
关键词
滑动窗口
非等时距
沉降监测
灰色模型
sliding window
unequal time interval
settlement monitoring
grey model
