中文摘要：

利用BP神经网络技术对焉耆盆地农田排水量进行预测。利用灰色关联度分析确定了排水量与各影响因素的关系，选取了对排水量影响最大的5个因素作为BP网络的输入，利用均匀设计方法，确定了最优的神经网络结构。估算结果表明利用BP神经网络可以准确地估算农田排水量，最大相对误差仅为-2．45％。

英文摘要：

BP neural network is used to estimate the farmland drainage in Yanqi Basin, Xinjiang Autonomous Region, China. The correlation between farm drainage and the influencing factors was estimated by gray correlation degree method. The five most important influencing factors were chosen to be the input for the BP neural network to estimate the farmland drainage. In order to obtain the optimal structure of BP neural network, the uniform design was employed. The results of estimation show that BP neural network can estimate farmland drainage accurately with the largest relative error only of 2.45G.