经济波动谱分析方法的比较研究

Comparative study of spectral analysis method of business fluctuation

为解决我国经济序列长度较短及各主周期分量难以同时准确分辨的问题 ,提出加窗周期图谱估计与最大熵谱估计相结合的谱估计方法 ,并将各主要周期分量单独分辨 ,且通过对序列进行三角函数拟合来确定周期长度准确值。从谱估计曲线、主周期分量的分辨和周期长度的准确性 3个方面与一般的谱分析方法进行了比较 。

A spectral estimation of combining periodogram windowing spectral estimation and maximum entropy spectral estimation is suggested to solve the problem that the length of economic time series in China is short. A method of determining the accurate value of period length of each main hiding period is put forward to deal with the problem that each main hiding period is difficult to be resolved accurately at the same time, The method resolves hiding period one by one and fitting data with triangular function. The method is shown to be effective and reasonable by comparing with ordinary spectral analysis method in three aspects of spectral estimation curve, resolving of main hiding periods and accuracy of period length to show that it is effective and reasonable.