摘要
金矿开展矿坝的变形监测工作,引入多层递阶回归分析模型,有较高的预测精度,但方法较繁琐,计算较复杂。由于变形数据可分离成趋势项与随机项,趋势项可用多元线性回归良好地拟合;随机项的预测,文中采用Elman网络建模计算,最后利用矿坝的实测高程位移数据进行验证,并与多层递阶回归进行比较。结果表明:回归-Elman网络模型比多层递阶回归的预测精度更高,效果更好,且方法简洁实用性强。
A deformation monitoring for a gold mine is conducted in Shangdong using multi-layered hierarchical regression analysis.A good prediction accuracy is obtained,but the method is complicated in calculation.The deformation monitoring data can be seperated into trend values and random values.Regression method can predict trend values well.To predict the random value,this paper has introduced Elman neural network.The actual monitoring data for elevation displacement variation of mine dam is used for model verification.The prediction result is compared with multi-level recursive regression.It is concluded that better accuracy of prediction can be obtained through the regression-Elman network model and this method is concise and practical.
出处
《测绘工程》
CSCD
2016年第1期39-42,共4页
Engineering of Surveying and Mapping
关键词
ELMAN神经网络
回归分析
变形预测
多层递阶回归
组合模型
Elman neural network
regression
deformation prediction
multi-level recursive regression
combined model
