基于ELM的尾矿坝浸润线预测

为了进行尾矿坝浸润线预测,提出一种极限学习机(ELM)方法。ELM网络能够很好地描述浸润线与其影响因素的非线性关系,将最小干滩长度、库水位、渗流量、竖直位移、水平位移5个主要因素作为ELM网络的输入,浸润线埋深作为网络的输出。为了提高ELM的预测准确性,利用均方误差指标选取归一化方法、激活函数、隐含层节点个数,最终确定最大值归一化方法预处理数据,输入5-9-1ELM网络,选取激活函数为sin型函数进行浸润线预测。同时选取BP神经网络,采用相同的归一化方法和网络形式进行对比。结果表明ELM模型在浸润线短期预测中可行性更高,预测精度佳。

Prediction of Phreatic Line of Tailings Dam Based on Elm

In order to predict the phreatic line,a method based on extreme learning machine(ELM)is proposed.The ELM network could well describe the nonlinear relationship between the seepage line and its influencing factors.The five main factors of seepage line,including the minimum dry beach length,reservoir water level,seepage flow,vertical diplacement and horizontal displacement,were used as the input of the ELM network,and the height of the phreatic line was used as the output of the network.In order to improve the prediction accuracy of ELM,the normalization method,activation function,and number of hidden layers nodes are selected using the mean square error,and the maximum normalization method is finally determined to preprocess the data,which were entered the 5-9-1 ELM network and select the sin-type activation function.At the same time,the selected BP neural network that has the same normalization method and network form are used for comparison.The results show that ELM model has higher feasibility and better prediction accuracy in the short-term prediction of seepage line.