摘要
水资源已经极大地制约了义乌市的经济发展,对义乌市降水量的预测,意义十分重大。BP神经网络具有自学习、自组织和容错性等一系列优点,用其来进行降水量预测是可行的。经过多次试预测,选出12月NINO-3区海表平均温度作为预测因子,义乌市5—6月的降水总量为预测对象。将1981—1999年12月NINO-3区海表平均温度的数据作为网络训练样本,2000—2002年义乌市5—6月的降水总量作为测试样本。网络测报结果平均误差为15.90%,预报情况良好。
Water resources have greatly restricted the economic growth of Yiwu. It is significant to predict precipitation in this city. With advantages of self- learning, self-organization and fault tolerance, BP neural networks can be used to predict precipitation. Through multiple trial tests, we selected the average sea surface temperatures of NINO - 3 zone in December as prediction factors, and total precipitation in May and June as prediction objects. The average sea surface temperatures of NINO - 3 zone in December from 1981 to 1999 were adopted as network training samples and total precipitation in May and June of 2000 - 2002 as test samples. The results of network prediction have a mean error of 15.90%. This study can offer reference to Yiwu and other areas in using BP neural networks to predict precipitation.
出处
《浙江水利科技》
2014年第2期52-54,共3页
Zhejiang Hydrotechnics
基金
国家自然科学基金资助项目(41171430)
关键词
BP神经网络
义乌市
降水量
预测
BP neural networks
Yiwu
precipitation
prediction