神经网络算法在南水北调冰期输水中的应用

通过分析南水北调沿线气候和气温特点，用LevenbergMarquart算法改进传统BP神经网络算法进行气温稳定转负日期的预报。把该模型应用到南水北调沿线新乡、安阳、邢台和石家庄2003—2006年气温稳定转负日期预报中，取得良好效果。立冬日作为气温转负日期统计的基准点，提高了预报精度，说明中国传统二十四节气对于冰情预报及其它同气温和气候相关的预报有重要的参考价值。

Ice forecast by artificial neural networks in the Middle Route of the South to North Water Diversion Project

By anlyzing the meteorological and hydrological character, an ANN model based on feed-forward back-propagation (FFBP) and improved by Levenberg-Marquardt algorithm is applied to forecast the dates which the air temperature goes below zero degree in the South-to-North Water Diversion the Middle Route of Project. The forecast results in the winter of 2003-2006 are mostly in good agreement with the measured data. The accuracy has been improved obviously when the Begining of Winter is taken as the datum point of the static data, which shows that the traditional Solar Terms of Chinese Calender is very important for the forecast of ice condition and correlated air temperature and climate factors.