温特线性与季节指数平滑法在气温中的应用

The Application of Winter Linear and Seasonal Exponential Smoothing Method in Temperature Prediction

由于月平均气温是以一年为周期呈现周期性波动,因此用温特线性与季节性指数平滑法对气温进行预测。以福州市2000—2012年月平均气温数据为样本数据,利用MATLAB软件通过建立温特线性与季节性指数平滑预测模型对福州市2013年1—12月份的气温进行预测,并通过计算机的迭代运算,得到一组最佳的平滑常数(α,β,γ),使得预测模型的均方误差MSE最小。研究结果显示,福州市月平均气温预测模型的平滑常数为(0.5,0.05,0.05),均方误差MSE为1.909 7,预测精度较高。

Due to the monthly average temperature periodic fluctuations is based on one year period, so this article uses the Winter l.inear and Sea- sonal Exponential Smoothing to forecast the temperature. Taking Fuzhou 2000-2012 monthly mean temperature data as sample data, using the MATLAB software by establishing Winter Linear and Seasonal Exponential Smoothing Forecasting Model forecast the temperature of Fuzhou Jan uary 2013 - December, and iterative computation by computer, get a set of optimal smoothing constant （α,β,γ） , makes the prediction model of minimum Mean Square Error （MSE）. Fuzhou monthly average temperature prediction model for the smoothing constant （0.5, 0.05, 0.5）, Mean Square Error is 1. 9097, higher prediction precision.