摘要
文章采用传统预报方法——多元线性回归和新方法——BP网络和投影寻踪技术,分别建立预报模型。利用长江宜昌站8—12月径流资料来研究预报模型的可行性和实用性。研究表明,3个预报模型的精度均在20%以内,尤其是BP模型预报精度均在10%以内,效果较好,具有一定的实用价值。
The forecasting models are established respectively by adopt the traditional forecasting method,the multiple linear regressions,and the new methods,BP neural network and projection pursuit.The feasibility and practicality of forecasting models are studied by using the runoff data from August to December of Yichang station in Yangtze River.The study results show that the accuracy of the three forecasting models is within 20%,especially within 10% for the accuracy of the BP model,which has the better effect and certain practical value.
出处
《东北水利水电》
2012年第4期43-44,52,共3页
Water Resources & Hydropower of Northeast China
关键词
径流预报
多元线性回归
BP网络
投影寻踪
runoff forecasting
multiple linear regressions
BP network
projection pursuit
