摘要
针对功放线性化算法中常用的Hammerstein模型,本文提出了一种基于信号高阶统计量的两步辨识方法,分别提取其线性记忆模块及无记忆非线性模块的参数。首先利用输出信号的高阶累积量的特殊切片,构建Hankel矩阵并将模型记忆深度的确定转换为矩阵的求秩问题,并同时提取线性记忆模块的参数,理论推导表明,模型的非线性效应并不影响线性效应的辨识;提出了一种迭代算法以提取无记忆非线性模型的参数,结果表明,若无记忆非线性模块传函为奇函数时,利用具有对称分布的独立同分布信号(I.I.D)作为激励,仅需一次迭代即可以求得非线性系数的全局最优值。仿真结果验证了该方法的可行性和高效性。
This paper proposes a two-step identification method of Hammerstein model based on higher-order-cumulant. The memory depth determination is converted into finding the rank of Hankel matrix constructed by cumulants of system output,and linear block coefficients extraction method is given, the theoretical derivation and simulation result indicates that the extraction process is not affected by nonlinearity of Hammerstein model. In step two, an iterative method is proposed for identifying the nonlinear block. Assume the input signal is i.i.d. with symmetric distribution, conclusion shows that if the transfer function of nonlinear block is odd, there is only one iteration step to converge global true parameters of nonlinear block. Numerical results that demonstrate the validity and efficiency of proposed scheme are presented in this paper.
出处
《微波学报》
CSCD
北大核心
2014年第S1期25-28,共4页
Journal of Microwaves