摘要
受多种因素影响,滑坡变形具有趋势性和随机性的特点。从滑坡变形监测数据着手,将监测数据分离成趋势值和随机值,建立了滑坡变形的回归-神经网络预测模型。该模型采用逐步回归方法对滑坡变形的趋势值进行预测,用BP神经网络预测方法对滑坡变形的随机值进行预测。利用金沙江乌东德坝址区金坪子滑坡TP06点高程位移变化实测数据,对该模型进行了验证。结果表明:预测误差不超过11%,具有较高的预测精度。
Because affected by many factors,it has trend and randomness for the monitoring data of landslides deformation.This paper started from the monitoring data for landslide deformation,the data was separated into trend values and random values,established a landslide deformation prediction model of regression-Neural network.The model used stepwise regression to predict trend values of landslide deformation and used BP neural network to predict to predict the random value of landslide deformation.The actual monitoring data for elevation displacement variation of Jinpingzi landslide TP06 point in Wudongde dam site area of the Jinsha River was used for model verification.The results show that the error of 6 issue data to predict is less than 11%,which is of high precision.
出处
《人民黄河》
CAS
北大核心
2012年第7期90-92,共3页
Yellow River
基金
国家自然科学基金资助项目(41061041)
江西省教育厅科技研究项目(GJJ11472)
关键词
预测模型
滑坡变形
逐步回归
BP神经网络
prediction model
landslide deformation
stepwise regression
BP neural network
