摘要
考虑到因确定性系统的外在随机因素和内在随机性所造成的长期预报的不准确性,我们将有关的浑沌理论和数理统计理论结合起来,提出了d维相空间模回归预报法。经调试后,该模式的距平符号报准率一般不低于66％,而相对误差一般不大于10％。
Considering the error of long-term forecast, which is caused by the external ran-dom factors and the internal randomness of a deterministic system, the theory about chaos is combined with statistics theory in this paper, then a modular regression model of d-dimensions phase space is set up. After debugging the model,the forecast accuracy will reach more than 66 per cent.
出处
《北京大学学报（自然科学版）》
CAS
CSCD
北大核心
1992年第5期596-601,共6页
Acta Scientiarum Naturalium Universitatis Pekinensis
关键词
相空间
关联维
模回归
长期预报
Phase space
Fractal dimension
Modular regression
Forecast
