论文部分内容阅读
研究了一种新的AR SαS过程的谱估计算法。该算法将整个数据作为一个整体,利用分数低阶p阶矩从前向、后向两个方向对数据进行处理,获得了一种高分辨率的参数估计算法——双向最小p范数法(Bidirectional Least p Norm,BLPN)。利用得到的参数,结合共变谱的定义,构建了AR SαS过程下的共变谱估计表达式,并分别对AR SαS过程参数估计、α稳定分布噪声中的正弦信号的谱估计进行仿真。仿真结果表明,基于BLPN的ARSαS模型的共变谱估计方法对于不同的α值均具有良好的韧性,特别是在α值较小或者短时数据时,本文方法的性能明显优于基于FLOM的AR SαS模型共变谱估计方法。
A new spectral estimation algorithm for AR SαS process is studied. The algorithm takes the whole data as a whole and processes the data from the forward and backward directions by using fractional lower order p moments to obtain a high resolution parameter estimation algorithm - bidirectional minimum p-norm Least p Norm, BLPN). With the help of the parameters and the definition of covariation spectrum, the expression of covariable spectrum estimation under AR SαS process is constructed and the spectral estimation of sinusoidal signal in AR SαS process parameter estimation and α stable distribution noise are simulated respectively. The simulation results show that the covariably spectrum estimation method of ARSαS model based on BLPN has good toughness for different α values, especially when the α value is small or the short-time data, the performance of the proposed method is better than the FLOM-based AR Covariogram estimation method for SαS model.