摘要
为了探索ＢＰ网络的参数调整特性，进行了参数α、β的选取对ＢＰ算法的收敛速度和模型的稳定性的影响研究。通过ＢＰ网络用于气象预测建模的参数调整个例分析表明：参数α、β的取值对ＢＰ模型的稳定性无显著影响，但参数值的调整尤其是β值的调整对建模的收敛速度有明显的影响。
The effect of the convergence speed and stability of B P algorithm based on the selection of parameters are studied to investigate the adjustable feature of parameters of B P network.The results of analysing some cases of B P network applied to the parameters adjustment of meteorological forecast modelling show that the valuation of parameters affects the B P network model not so markedly,but the adjustment of parameters affects the convergence speed of modelling markedly.
关键词
参数调整
模型特性
气象预测
B-P网络
Parameter adjustment, Artificial neural network, B P algorithm, Model feature, Meteorological forecast
