混沌时间序列的优化预测模型及算法研究

针对混沌时间序列难以预测和控制问题,提出了基于趋势的混沌预测模型,利用混沌系统的初值、参数敏感性来微调和控制系统扰动,并用改进的最优化方法估计模型的参数,在其相空间中对时序未来值进行预测.算例表明,选取最佳的模型阶数能增加预测的准确程度,它不仅克服了仅用延迟嵌入技术的弊端,也降低了直接使用预测误差决定输入模式的盲目性.预测效果比其他时序方法要好.

Research on optimal prediction model and algorithm about chaotic time series

Facing the of difficulty to estimate and control chaotic time series, a chaotic estimating model was suggested, where the initial value of the chaotic system and the sensitivity of the parameter are used to inching and control the disturbance of the system, and estimated the parameter of the model by using the best update option. The intending series value was forecasted is in its mutually space. The example shows the increase of the precision in the estimated process by selecting the best model steps. It not only conquer the deficiency of using detention inlay technology alone, but also decrease blindness of using forecast error to decide the input model directly. The prediction result is better than the method of statistics and other series means.