%0 Journal Article
%T 基于逻辑回归–径向基神经网络的广西前汛期降水预测<br>Prediction of Precipitation during the Pre-Rainy Season in Guangxi Province Based on Logistic Logical Regression-Radial Basis Function Neural Network
%A 蒙芳秀
%A 苏健昌
%A 覃碧莉
%A 吴俊皇
%A 蒋宜蓉
%J Advances in Applied Mathematics
%P 8540-8549
%@ 2324-8009
%D 2022
%I Hans Publishing
%R 10.12677/AAM.2022.1112901
%X 基于2000~2020年4~6月广西5个代表站地面气象观测站降水量，建立基于逻辑回归–径向基神经网络(logical regression-radial basis function neural network，简称LR-RBF)的广西前汛期降水预测模型。结果表明，基于LR-RBF预测广西前汛期降水效果较好，实测降水与预测降水的R值均高于0.91，呈现高度相关。预测精度较径向基神经网络(Radial Basis Function, RBF) MAE值最大减少为45.9%，而RMSE值最大减少为35.71%，特别是贺州站；与逐步回归相比，R值最高增大13.09%，MAE最大减少26.91%，RMSE最多减少23.10%。结果表明，LR-RBF预测能力有显著的提升，对广西前汛期防洪防控工作具有一定的指导价值。<br />
Based on the daily precipitation data of 5 representative stations in Guangxi Province from April to June 2000 to 2020, a logistic regression-radial basis function neural network method (LR-RBF) is established to prediction precipitation during the pre-rainy season. From the results observed, the LR-RBF method has a good effect on modeling precipitation during the pre-rainy season in Guangxi Province. For instance, the R value between observed precipitation and predicted precipitation is higher than 0.91, showing a high correlation. Compared with the radial basis function neural net-work (RBF), from the prediction accuracy, MAE decreased by 45.9%, and the reduction of RMSE is 35.71%, especially at Hezhou station. Compared with the stepwise regression, the increase of R value is 13.09%. Meanwhile, the decrease of MAE is 26.91%, and the decrease of RMSE is 23.10%. The results show that the prediction ability of the LR-RBF method has been significantly improved, which provides a reference for flood control in Guangxi Province during the pre- rainy season.
%K 逻辑回归，径向基神经网络，降水预测，前汛期<br>Logistic Regression
%K Radial Basis Function Neural Network
%K Precipitation Prediction
%K Pre-Rainy Season
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=58886