摘要
针对尾矿坝安全监测需要对尾矿坝地下水位进行预测及时分析尾矿坝稳定性的问题,提出利用主成分分析和神经网络建立尾矿坝地下水位的预测模型。首先采用自相关分析选择预测模型的输入变量,然后利用主成分分析法对原始多维输入变量进行降维和去相关处理,最后利用提取的主成分作为神经网络的输入,对某尾矿坝地下水位进行预测实验。实验结果表明,经过主成分分析处理后的神经网络预测模型不仅简化了网络的结构,具有更高的预测精度,而且提高了网络泛化性能和稳定性。
In order to forecast groundwater levels for stability analysis of a tailing dam in the tailing dam monitoring system, the mixed method based on principal component analysis （PCA） and artificial neural networks was put forward to realize the function of the neural network prediction model of groundwater levels near tailing dam. The input variables of the prediction model were selected by autocorrelation analysis. The PCA was used to preprocess these original input variables, thus both the dimensions and correlativity of the input variables can be reduced. The principal components of input variables were chosen as the input of neural network, and the proposed prediction model was used to forecast groundwater level in a tailing dam. The simulation results show that the mixed method not only simplifies the network structure and has a high prediction precision, but also improves the generalization performance and robustness of the neural network prediction model.
出处
《水文地质工程地质》
CAS
CSCD
北大核心
2014年第2期13-17,共5页
Hydrogeology & Engineering Geology
基金
山东省高等学校科技计划项目(J10LG22)
关键词
主成分分析
神经网络
尾矿坝
地下水位
预测模型
principal component analysis
artificial neural networks
tailings dam
groundwater level
prediction model
