摘要
为有效监测高层建筑物沉降,提出了二阶齐次序列的直接离散模型DDGM(2,1),结合蚌埠市某小区高层建筑物沉降观测数据,建立高层建筑物沉降预测模型。为减小累计误差,推导出了与DDGM(2,1)模型等价的通项形式预测模型。同时将DDGM(2,1)、Verhulst和DNDGM(2,1)模型的预测值与实际值进行对比,计算各个模型的RMSE,MAE,MSE等评价指标。研究结果表明:在建模拟合区域,直接离散通项形式模型的预测精度优于直接离散模型,而在模型预测区域,其预测精度劣于直接离散模型,两者之间MAE的绝对误差为10%。DDGM(2,1)模型拟合及预测的沉降量均小于0.211mm,平均绝对误差(MAE)较DNDGM(2,1)模型减小78%,较Verhulst模型减小9%,有较好的预测能力。
To effectively monitor the settlement of high-rise buildings,a direct discrete model of second-order chi-square sequence,DDGM(2,1),is proposed to establish a prediction model for the settlement of high-rise buildings by combining these data with the observation data of high-rise building settlements in the neighborhood of Bengbu city.To reduce the cumulative error,a generalized prediction model equivalent to the DDGM(2,1)model is derived.Moreover,the predicted values of the DDGM(2,1),Verhulst and DNDGM(2,1)models are compared with the actual values,and evaluation indices such as the RMSE,MAE and MSE of each model are calculated.The results show that the prediction accuracy of the direct discrete generalized form model is better than that of the direct discrete model in the simulation region,while its prediction accuracy is inferior to that of the direct discrete model in the model prediction region;moreover,the absolute error in the MAE between the two models is 10%.The DDGM(2,1)model-fitted and predicted settlements were less than 0.211 mm,and the mean absolute error(MAE)was reduced by 78%compared with that of the DNDGM(2,1)model and by 9%compared with that of the Verhulst model,which indicated a better prediction ability.
作者
黄丹
陈凯
檀秋芬
彭佩
HUANG Dan;CHEN Kai;TAN Qiufen;PENG Pei(School of Architecture and Engineering,Wuhu Institute of Technology,Wuhu Anhui 241000,China;Wuhu Assembly Engineering Technology Research Center,Wuhu Anhui 241000,China)
出处
《佳木斯大学学报（自然科学版）》
CAS
2024年第11期102-105,138,共5页
Journal of Jiamusi University：Natural Science Edition
基金
安徽省高校自然科学研究重大项目(KJ2020ZD73)
安徽省教育厅自然科学重点项目(KJ2021A1321)。
作者简介
黄丹(1990-),女,河南商丘人,讲师,硕士,研究方向:智能建造工程技术。
