基于状态空间模型的最小熵反褶积

Minimum Entropy Deconvolution Based on State Space Model

中文关键词  反褶积  反褶积辨识  信号处理；非最小相位系统

英文关键词  deconvolution  system identification  signal processing  non-minimun phase system

中文摘要

      本文提出了基于状态空间模型的最小熵反褶积方法用于辨识非最小相位系统。该方法用状态空间模型对系统建模，首先用最小二乘反褶积估计对应真实系统的最小相位系统，然后用由最优平滑器估计的输入序列构造熵函数搜索具有真实相位的系统，利用主导极点的概念提出了启发式搜索算法，大大减少了搜索时间。仿真及应用于实际数据均表明本文方法是行之有效的。

英文摘要

      In this paper, the method of minimum entropy deconvolution is proposed to identify the non-minimum phase system. The system is modelled as the state space model. Least square deconvolution is used to identify the minimum phase system corresponding to the system of the true phase, that is to estimate the amplitude spectrum of the system. The phase of the system is searched according to the entropy which is the function of the input sequence estimated by the optimal smoother. The heuristic search algorithm is proposed from the idea that a system has the leading poles which dominate the transient response of the system. The searching time is greatly reduced by this algorithm.