混合高斯自回归模型参数估计方法之ML-DC  

ML-DC Algorithm of Parameters Estimation for Gaussian Mixture Autoregressive Model

作　　者：王平波[1] 蔡志明[1] 刘旺锁[1] 

机构地区：[1]海军工程大学电子工程学院,武汉430033

出　　处：《信号处理》2007年第6期864-868,共5页Journal of Signal Processing

基　　金：973基金项目;编号为5132102ZZT32.

摘　　要：混合高斯自回归模型可以对有色非高斯数据的概率密度和功率谱密度进行有效的拟合.而ML-DC算法则可解决这一模型的参数估计问题。描述了混合高斯自回归模型及其参数估计问题之后,分别导出了功率谱密度参数的最大似然估计和概率密度参数估计的动态簇算法,并由此组成了参数耦合估计的ML-DC算法。最后结合一组仿真实例对其估计性能进行了详细探讨,指出并解释了算法的适用范围。With Gaussian mixture autoregressive model,the probability density and power spectrum density of non-Gaussian colored processes can be fit. Its parameters can be estimated through the ML-DC algorithm. After descriptions of the model and the estimation problem, maximum likelihood estimation of autoregressive parameters and the dynamic clutter algorithm for Gaussian mixture parameters are dedueed, respectively. Based on them, ML-DC algorithm for coupling estimation between power spectrum density parameters and probability density parameters is built up. Finally, a numerical instance in simulation is illustrated where performance of estimation is discussed in detail.

关 键 词：混合高斯自回归模型 最大似然估计 动态簇算法 

分 类 号：TN911.7[电子电信—通信与信息系统] F830.9[电子电信—信息与通信工程]