提出一种改进的BP算法 ,即自适应BP算法。该方法采用两种策略 :一是在权重修改公式中加动量项 ;二是学习率随总误差的变化作自适应调整 ,亦即总误差增加时 ,学习率将减小 ,反之学习率增大。以上两种策略能有效的抑制网络陷于局部极小并缩短了学习时间。实例研究表明 ,该算法用于河道洪水的预报 ,能取得令人满意的结果 An improved BP algorithm is proposed, namely adaptive BP algorithm. The method adopts two strategies: one is to add the momentum term to the weight modification formula; the other is that the learning rate is adjusted adaptively with the change of the total error, that is, when the total error increases, the learning rate will decrease; on the contrary, the learning rate will increase . The above two strategies can effectively restrain the network from being locally minified and shorten the learning time. The case study shows that this algorithm can be used to predict the flood of the river and achieve satisfactory results