基于RBF神经网络的水质参数预测模型研究

Research on Water Quality Parameter Prediction Model Based on RBF Neural Network

本文利用Matlab对非线性函数逼近精度较高的RBF神经网络建立预测模型,以1998-2017年乌梁素海TN、TP、COD浓度的年际变化为数据,划分数据集和训练集,训练得到TN、TP和COD的RBF神经网络预测模型,通过对TN、TP和COD真实值和训练值及真实值和预测值的对比分析可得出均方根误差相对较小,证明RBF神经网络模型对水质参数的预测精度相对较高。

In this paper,Matlab is used to establish a prediction model for RBF neural network with a high nonlinear function approximation accuracy,and the interannual variation of TN,TP and COD concentrations in Wuliangsuhai from 1998 to 2017 is used as the data,and the dataset and training set are divided to obtain the RBF neural network prediction model of TN,TP and COD through training.The comparison and analysis between the true value and the trained value and between the true value and the predicted value of TP and COD shows that the root mean square error is relatively small,which proves that the RBF neural network model has a relatively high prediction accuracy for water quality parameters.