基于多层感知人工神经网络方法的水温预测
摘要:为了确定一个小流域的水温,本文对三个人工神经网络的效率进行了对比分析。本文采用了从1997-2006年的数据进行试验,利用从2007-2008年的数据对其准确性进行了校核。试验期的相关系数均在0.82以上,校核期的相关系数达到了0.9。神经网络为预测水温以及河流水温的变化趋势提供了一个很好的工具。
ABSTRACT:The efficiency of three Artificial Neural Networks has been analysed, for determination of water temperature of a micro-basin river. Data from 1997 to 2006 were used for training. Its accuracy was tested with data from 2007 and 2008. A high correlation coefficient (R) always above 0.82 for training set and 0.90 for validation set was obtained. Neural Networks provided us with a good tool to forecast water temperature and are a valuable tool for predictions of river’s water temperature.