动态出现和/或消失时频信号的模型和分析

针对动态出现和/或消失的时频信号，本文提出了一种时频模型和分析方法.该模型以时频信号各分量的幅度和相位为状态变量，并利用多项式预测模型为它们建立起状态方程，再视多分量混合时频信号的观测为测量方程，这就为多分量时频信号建立了状态空间模型.为了应对时频信号分量动态出现和/或消失的情况，本文利用非参数迭代自适应谱估计方法对时间加窗的信号进行分析，得到该时间窗内的短时谱，以该短时谱中噪声谱的3倍标准差准则来确定信号分量的数目.由此，基于提出的模型，就可利用无迹卡尔曼滤波算法来分析动态出现和/或消失的时频信号.分析和仿真均表明：提出方法无论在精确度、分辨率以及交叉时频谱分离等方面均优于文献报道的方法.

A Time-Frequency Model and Analytical Method for Multiple Modulated Components with Dynamic Births and Deaths

In this paper,a new time-frequency model and analytical method are proposed for multiple modulated components with dynamic births and deaths.The proposed model takes the amplitude and phase of components as the state variables whose evolutions are described as polynomial prediction models.The observations of the signal are then viewed as the measurements of the states.In this way,a new state space model is built to describe the signal with multiple modulated components.In order to cope with the dynamic births and deaths of the components,a non-parametric iterative adaptive approach with a time window is employed to obtain the short time spectrums.The number of signal components is found under the rule of three times standard deviation of the noise in the spectrums.Now,the unscented Kalman filter can be used to analyze the dynamic births and deaths signals based on the proposed model.Simulation results verify the effectiveness of the proposed method while its performances shown in terms of accuracy,resolution and crossing spectrum separation are much better than the ones reported in literature.