基于噪声辅助非均匀采样复数据经验模态分解的混沌信号降噪

鉴于非均匀采样复数据经验模态分解(NSBEMD)相对传统分解方法的优势和噪声的NSBEMD特性,提出了一种基于噪声辅助NSBEMD的混沌信号自适应降噪方法.该方法首先以含噪混沌信号和高斯白噪声分别为实、虚部来构造复数据并进行NSBEMD,然后根据虚部各IMF的能量来估算实部各IMF中包含的噪声能量,最后根据噪声能量的估计值对实部IMF进行奇异值分解(SVD)降噪.噪声估计实验验证了噪声能量估计方法的可行性,而Lorenz信号和太阳黑子月平均数的降噪实验则表明,相对于现有EMD降噪方法,本文方法能够进一步消除噪声,更清晰地恢复出混沌吸引子的拓扑结构.     

An Improved Local Weighted Linear Prediction Model for Chaotic Time Series

Previous research working on local prediction state with some unsuitable neighbor points （such as false points and pseudo-false neighbor points） are the main source of errors of local prediction and these unsuitable neighbors cannot be eliminated entirely. Therefore, an improved local weighted linear prediction model based on local integrated correlation, which can reduce the influence of the residual unsuitable neighbors, is proposed to predict chaotic time series in our study. Simulation results show that the performance of the improved model is superior to the other local prediction models in the prediction of chaotic time series without and with additive white Gaussian noise.     

基于复数据经验模态分解的噪声辅助信号分解方法

集成经验模态分解(EEMD)在一定程度上减轻了经验模态分解(EMD)中的模态混叠,但集成平均会带来新的模态混叠、频谱丢失和运算量增大等问题,影响到对信号物理特征的分析与提取.因此,本文提出一种基于复数据经验模态分解(CEMD)的噪声辅助信号分解方法,在CEMD中以白噪声分解的内禀模态函数(IMF)在指定方向上的投影为基函数来辅助观测信号分解过程中的极值选取,从而减小模态混叠,同时利用噪声投影的影响在求包络质心时被消除的特性,减小EEMD因集成平均带来的相关问题.仿真结果表明,本文方法在进一步降低模态混叠效应的同时,明显提高了运算速度,并且在一定程度上减轻了频谱丢失问题.