基于ARIMA-RBF算法的月度降雨量预测研究

Study on Prediction of Monthly Rainfall Based on ARIMA-RBF Algorithm

为了对月度降雨量进行科学预测,将ARIMA模型与RBF神经网络相结合,提出一种基于ARIMA-RBF耦合算法的月度降雨量预测模型。首先,利用ARIMA模型对月度降雨量线性部分进行拟合预测,计算ARIMA模型预测的残差;然后,利用RBF神经网络对ARIMA模型残差进行拟合预测;最后,利用RBF神经网络预测结果对ARIMA模型进行补偿修正,得到最终降雨量预测结果。将该方法用于重庆市沙坪坝月度降雨量实际预测中,预测结果精度高于单一ARIMA模型以及RBF神经网络,能够满足实际预测需求。结果表明:将线性拟合算法和非线性拟合算法结合起来用于月度降雨量预测是一种较为优越的算法。

In order to forecast the monthly rainfall,the ARIMA model and the RBF neural network are combined,and a monthly rainfall forecast model based on ARIMA-RBF coupling algorithm is proposed. At first,the ARIMA model is used to predict the monthly rainfall and the error of the ARIMA model is calculated. Then,the RBF model is used to predict the error of the ARIMA model. Finally,the RBF model is used to compensate the ARIMA model. The method is applied to the actual forecast of monthly rainfall in Shapingba City,Chongqing. The accuracy of forecasting results is higher than that of single ARIMA model and RBF neural network,which can meet the needs of actual forecast. The results show that the linear fitting algorithm and nonlinear fitting algorithm for monthly rainfall forecast is a relatively superior algorithm.