基于多音激励的Volterra频域核非参数辨识方法

摘要

为解决Volterra频域核辨识困难的问题,探讨了多音信号激励下Volterra频域核的输出性质,提出了一种基于多音激励的Volterra频域核的非参数辨识方法.该方法选择具有合适频率基的多音信号作为测试激励信号,使各阶核的输出频率成分具有唯一性,进而利用Vandermode法使各阶核的输出分离.推导出了Volterra频域核的辨识公式,并进行了理论和仿真分析.理论分析和仿真结果表明:该方法可准确地辨识出任一阶Volterra频域核,绝对误差达到10-4V数量级,克服了传统辨识方法无法准确辨识及辨识结果仅限于前3阶Volterra频域核的缺点,具有精度高、可操作性强的特点.%In order to solve the problem of difficultly identifying Volterra frequency-domain kernels,the output properties of Volterra frequency-domain kernels stimulated by multi-tone signal were investigated,and a novel non-parametric method to identify Volterra frequency-domain kernels was proposed based on multi-tone stimulus.With this method,a multi-tone signal with a proper frequency basis is selected as the test stimulus signal to make each kernel put out unique frequency components,further the output of each kernel is separated by the Vandermode method.The formula to identify Volterra frequency-domain kernels was derived,and theoretical and simulation analyses were carried out.The results indicate that any order Volterra frequency-domain kernel can be precisely identified by the proposed method,and the order of magnitude of the absolute error is 10-4 V.The proposed method overcomes the defects of the conventional methods,i.e.,they are unable to correctly identify Volterra frequency-domain kernels and the identification results are limited to the first three orders,and it has a high precision and is easy to operate.