基于前后向线性预测的时变参数估计方法研究

在基时间函数展开理论基础上,通过同时引入非平稳信号的前向线性预测和后向线性预测来估计模型的时变系数,提出了一种新的时变参数建模方法,有效地克服了现有方法对瞬时频率估计的滞后问题和无法准确估计初始时刻值的困难;仿真试验证明,在相同阶次和基函数维数条件下,该方法对非平稳信号瞬时频率的估计效果要优于现有的方法。

一种改进稳定性的时域电场积分方程法

研究了时域电场积分方程隐式MOT算法中时间基函数对其稳定性的影响,由理论分析和数值结果得出了选用高频分量较少的时间基函数不利于隐式MOT算法稳定性的结论。在此基础上,将一种多项式时间基函数与新的时间平均方法相结合,提出了一种改进稳定性的隐式MOT算法,计算实例表明比先五步后三步平均法更加精确,比以前的隐式MOT算法更加稳定。

Wavelets as Basis Functions in the Adaptation＇s Methods： Author＇s Model for Forecasting Short-Term

The aim of this article is to present author＇s application of wavelets to predict short-term macroeconomic indicators Proposed to predict short-term time series （in particular for predicting macroeconomic indicators）, proprietary model is based on wavelet analysis with Haar wavelets, Daubechies wavelets, and adaptive models; they are the trend crawling model and alignment exponential model. Adaptive models have been modified through the introduction of wavelet function and combined into a single forecast model. Obtained from conducted research results, it shows the model an effective instrument to predict the short-term.