BP神经网络在水质参数预测中的应用

The Application of BP Neural Network to Prediction of Water Quality Parameters

本文利用BP神经网络良好的非线性映射特性,建立了水质预测模型,分别建立三种不同训练算法的氨氮预测神经网络,利用MATLAB神经网络工具箱编程实现。选用大量数据,反复加以训练,通过比较,挑选拟合程度好、精度误差小、泛化性能优秀的预测神经网络,并根据测试样本加以检验分析,最终选定L-MBP网络作为氨氮含量神经网络预测方案。

In this paper, the well nonlinear mapping characteristics of BP neural network was used in the establishment of water quality prediction model. Different neural network with different training algorithms were established to predict ammonia, and Neural Network Toolbox in MATLAB was used to program. Largeamm ounts of data was choosed to be trained repeatedly to select a prediction neural network with a good fit of the model, small precision error and a good generalization performance,as a result of compari- son. To be based on the test samples, tests and analysis could be found,and the L-M BP Neural Network was selected as the ammonia content of neural network prediction program.