云南东川市人均粮食的时序分形特征

The time-series fractal characteristicsof grain per capita in Dongchuan city,Yunnan

应用1964～1995年东川市人均粮食资料,重构其嵌入相空间序列,探讨了其时序的分形特征,并计算了其平均可预报年限.结果表明,东川市人均粮食的时序分布具有分形特征,且该时序是一混沌序列,其饱和关联维数约为3.1631,系统的有效自由度数为8个,平均可预报尺度不超过27～28年.研究结果为人均粮食的时序预测预报模型提供了依据.

Based on the statistical data of grain per capita in Dongchuan city from 1964 to 1995,the time-series imbedding space is rebuilt, the time-series fractal characteristics are investigated, and the average predictable years is calculated. The results show that the temporal distribution of grain per capita in Dongchuan city has fractal characteristics; this time-series is a chaotic series; its saturated correlative dimension is 3.1631; the effective degree of freedom (DOF) is 8;and the average predictable years is no more than 27 to 28 years. The obtained conclusions provide a scientific basis for establishing the forecast model of grain per capita. It is a probing study on the application of fractal theory to time-series of grain per capita and its prospect needs more cases to test.