中文摘要：

针对GM模型要求的样本点少、不必有较好的分布规律,且计算量少、操作简便,而BP神经网络可以反馈校正输出的误差,具有并行计算、分布式信息存储、强容错力、自适应学习功能等特点,将GM（1,1）模型与BP神经网络模型相结合,建立了混合神经网络预测模型,并结合实例进行了检验性预测。结果表明：混合神经网络模型在预测精度方面优于传统灰色模型。该模型的算法概念明确、计算简便,有较高的拟合和预测精度,具有良好的应用前景。

英文摘要：

The GM model has many advantages, with less calculation and easy operation. It needs neither larger sample points nor better regulate distribution. While the BP neural network can feedback the corrected output errors, it has the characteristics, such as parallel computation, distributed information storage, strong fault tolerance capability and learning adaptivity , et al. Thus a hybrid neural network prediction model is established, with the advantages from both GM （1 ,1 ） model and BP neural network model. It has been applied in test predictions with examples. The results showed that hybrid neural network model in forecast accuracy is better than the traditional gray model. The model algorithm with advantages of clear concept, simple calculation, a higher fitting and prediction accuracy, has good prospect of application.