一种基于过采样技术的非最小相位系统盲辨识方法

提出了一种在时域中盲辨识非最小相位系统传递函数的方法,该方法通过对系统输出过采样来获得系统结构参数的信息.对于单输入单输出的线性离散系统,经过输出过采样后可以等价为单个输入、多个输出的传递函数模型,这多个输出函数具有相同的分母多项式和不同的分子多项式.采用子空间分解法可以确定分子多项式的参数,通过对系统输出信号自相关函数的处理可以得到分母多项式,最终可以得到原系统的结构参数.与传统高阶矩方法相比,该方法对噪声的敏感度更低,辨识的精度和速度也有很大提高.仿真结果表明,当信噪比大于15dB时,该方法可以有效地辨识出系统参数.与高阶矩方法相比,辨识门限信噪比降低了10dB,估计精度提高了20%,辨识速度加快了3倍.

Blind Identification Method for Nonminimum Phase System Based on Over Sampling

Based on over sampling approach, a new time domain blind identification method for non-minimum phase system was proposed. For a linear time invariant single-input single-output system, when the system output was over sampled, the system would be equivalent to a transfer function model of single input and multi-output. This transfer function model has common denominator polynomial and different numerator polynomials. The parameters of numerator polynomials can be determined by the subspace decomposition and the parameters of denominator polynomial can be given through the relation analysis between input and output. Finally, the structure of the original system can be obtained. Compared with high order statistics approaches this method is more insensitive to noise and also has higher estimate accuracy and estimate speed. Simulation result shows that the method can identify system parameters efficiently when the signal-to-noise ratio (SNR) is bigger than 15 dB. The threshold of SNR is 10 dB lower than that of the high order statistics method. The estimate accuracy and estimate speed are increased by 20% and 300% respectively.