经验模态分解与神经网络的GPS可降水量预测

Prediction of GPS water vapor based on EMD and ANN

本文将经验模态分解与神经网络相结合的方法应用于GPS可降水量预测。对GPS可降水量进行经验模态分解,在分解过程中采用径向基神经网络处理端点问题,针对经验模态分解得到的每个分量分别运用径向基神经网络进行预测,并重构出最后的预测结果。重构结果与直接运用神经网络进行预测的GPS可降水量、实测GPS可降水量进行比较,结果表明,相对于直接预测,经验模态分解与神经网络结合的方法具有更高的预测精度。

The method of Empirical mode Decomposition （EMD） and Artificial Neural Network （ANN） were applied to predict GPS precipitable water vapor （PWV） in the paper. Firstly it decomposed GPS PWV by the method of EMD, and solved the endpoint problerm Secondly it separately predicted each component of GPS PWV EMD results with RBF neural network, and reconstructed the final prediction results. Compared with the PWV predicted by ANN and GPS PWV, it was concluded that the accuracy of the PWV predicted by the method of EMD and ANN is higher than the accuracy of the PWV predicted by ANN.