中文摘要：

小波分析属于时频分析的一种，它在时域和频域同时具有良好的局部化性质。本文利用小波分解与重构方法将时间序列分解成趋势项、周期项和随机项。根据Morlet小波的类周期特性，分析并挑选出合适的周期，结合逐步最小二乘估计方法，计算出该时间序列周期项的近似表达式。最后用此小波混合方法研究建国后中国玉米产量变化情况，结果表明该方法比直接二次多项式拟合预测的平均相对误差小2．336％，反映小波混合预测方法的有效性。

英文摘要：

The wavelet analysis belongs to the time-frequency approach and has the good localization nature in the time domain and the frequency range simultaneously. This article uses the wavelet decomposition and restructuring method to decompose the time series into the tendency part, periodic part and the stochastic part. According to the likely periodic property of Morlet wavelet, we analyze and choose the appropriate periods. Using Ordinary Least Square method step by step, we work out the approximate expression of the periodic part of this time series with sine and cosine functions. As/in application, we use this wavelet mixed method to discuss the corn output case in China from 1949 to 2005. The fact that the mean relative error is 2. 336% less than the one by directly polynomial fitting up to two degree shows that this wavelet mixed method is more effective.