基于ARMA模型的水文数据预测

The Prediction of Hydrological Data Based on ARMA Model

针对小样本水文数据序列难以准确预测的特点,将时间序列分析运用于水文数据的预测分析,研究基于AIC定阶准则和遗传算法定阶的ARMA模型,并将其运用于周期性水文数据的预测。根据模型建立的需要及数据周期性的特点,对原始数据进行季节差分等优化处理,并将建立的模型运用于某水文站流量数据的预测。结果表明,基于ARMA模型对流量数据的预测精度远远高于传统的神经网路模型,其中基于AIC准则定阶的模型比遗传算法定阶的模型精度高2.96%,从而为小样本水文数据的预测分析提供一种新的思路。

Because of the difficulty in the prediction of hydrological data for small samples, the article put forward a ARMA model which is used to analyze the periodic hydrological data based on AIC order selection criteria and genetic algorithm. According to the need of the model and the characteristics of data periodicity, the original data were optimized by the seasonal difference,and the model was applied to forecast the discharge data of a hydrological station. Results show that the accuracy of ARMA model is more higher than that of the traditional BP neural networks model, Furthermore, the accuracy of model based on AIC order selection criteria is 2.96% higher than that of the model based on genetic algorithm. Therefore, the article provides a new idea for the analysis of the hydrological data for small samples.