摘要
单输出高斯过程模型针对单点建模和预测,未顾及各监测点间的相互影响,是一种局部建模和变形规律分析。多输出高斯过程模型把监测点间的互扰性作为预测模型的影响因素,对多个监测点建模预测时能充分利用监测点间的相关信息。通过建筑物和滑坡的2个工程实例验证分析,与单输出高斯过程模型相比,多输出高斯过程模型的预测值与实测值更为吻合,预测均方根误差分别提高了44.5%和77.8%,表明多输出高斯过程模型具有更高的预测精度,同时验证了该模型预测建筑物和滑坡变形的有效性。
Single output Gaussian process model is aimed at single-point modeling and prediction and does not take into account the interaction between monitoring points. It is a kind of local modeling and deformation law analysis. Multiple output Gaussian process model takes the mutual interference between monitoring points as the influencing factor of the prediction model, and can make full use of the relevant information between the monitoring point when modeling and predicting multiple monitoring points. Validation analysis through two engineering examples of buildings and landslides, the predicted and measured values of the multiple output Gaussian process model are more consistent that compared with the single output Gaussian process model, and the predicted root mean square error is improved by 44.5% and 77.8%, respectively. The results show that the multiple output Gaussian process model has higher prediction accuracy and validates the validity of the model for predicting building and landslide deformation.
作者
邱小梦
欧阳亮酉
周世健
QIU Xiaomeng;OUYANG Liangyou;ZHOU Shijian(Yangtze River College,East China University of Technology,334000,Fuzhou,Jiangxi,PRC;Jiangxi East Road Bridge Construction Group Co.Ltd.,334000,Fuzhou,Jiangxi,PRC;Nanchang Hangkong University,330063,Nanchang,PRC)
出处
《江西科学》
2019年第6期832-835,845,共5页
Jiangxi Science
基金
江西省教育厅科技项目(181523)
东华理工大学长江学院院长基金项目
关键词
多输出高斯过程模型
变形监测
预测
单输出高斯过程模型
multiple output Gaussian process model
deformation monitoring
prediction
single output Gaussian process model
