人工神经元网络对单站降水预报的试验分析

应用Ｂ－Ｐ网络原理，选取西北地区２０多个站点的气象资料为样本资料，建立兰州７～８月降水预报专家系统。经过１９９６～１９９８年试验、修正、改进，系统性能稳定，具有一般预报员的预报能力，强于常规专家系统。

Experiment and Analysis of Forecast Rainfall of One Station Though Neural Network System

Based on the B-P neural network and selected meteorological data of more than 20 stations in Northwest of China, the rainfall forecasting system which is able to forecast rainfall of Lanzhou area in July and August is established. After experimented, modified and improved from 1996 to 1998, it is proved that the system is stable and has the forecast ability of forecaster. Furthermore, the system is better than normal expert system.