中文摘要：

针对传统的BP神经网络存在的不足及局限性,文章提出了利用小波分析和神经网络相结合的方法应用于GPS可降水量预测中。小波神经网络是将小波基函数来替代传统神经网络中的激活函数,它将小波分析和神经网络有机融合在了一起,同时具备小波分析和神经网络的良好特性。通过相同样本数据训练和学习以及对预测结果的对比分析,表明小波神经网络在可降水量预测中比BP神经网络具有更好的容错能力和逼近能力,且其收敛速度快,预测精度高。

英文摘要：

Aiming at the deficiencies and limitations of traditional BP neural network, this paper pro- posed a GPS precipitation prediction method by combining wavelet analysis and neural network method： a wavelet basis function was used to replace the activation function of traditional neural network, and the wavelet analysis and the neural network were organically fused, which would have good characteristics of both wavelet analysis and neural network. Sample data training, learning and prediction results analysis showed that the wavelet neural network could have better fault tolerance ability and approximation capability than BP neural network in precipitation prediction with good convergence speed and high prediction ac curacy.