论文部分内容阅读
鉴于降水量数据的高维非线性性和周期性,建立了支持向量回归(SVR)预测模型用于降水量预测,由于对该模型输入特征的选取极为重要,因此提出了一种基于季节自回归(SARI)的输入特征选取方法.利用已有的降水量数据建立SARI模型,通过观察模型表达式提取建立SVR模型所需的输入特征用于训练支持向量机,并通过网格参数寻优法确定SVR模型的参数,进行降水量预测.实例分析中,应用此模型对黄土丘陵半干旱区域的降水量进行预测,将预测结果与季节时间序列(SARIMA)模型的预测结果进行对比,结果表明,模型具有更高的预测精度和拟合优度,可以用于降水量的预测.
In view of the high dimensional nonlinearity and periodicity of precipitation data, a support vector regression (SVR) prediction model is established for predicting precipitation. Because of the importance of the input features of the model, a seasonal autoregressive (SARI) ), The SARI model is established by using the existing precipitation data, the input features needed to establish the SVR model are extracted by observing the model expressions and used to train the SVM, and the SVR model is determined by the grid parameter optimization method , The model is used to predict the precipitation.The case analysis is used to predict the precipitation in the loess hilly semi-arid region, and the prediction results are compared with the predicted results of the seasonal time series (SARIMA) model. The results show that the model has more High prediction accuracy and goodness of fit can be used to predict precipitation.