一种应对窄带非线性主动噪声控制的混合结构  

A Novel Hybrid Structure for Narrowband Nonlinear Active Noise Control

作　　者：罗磊[1] 孙金玮[1] 黄博妍[1] 白宗龙 

机构地区：[1]哈尔滨工业大学电气工程及自动化学院,黑龙江哈尔滨150001

出　　处：《电子学报》2017年第8期1801-1809,共9页Acta Electronica Sinica

基　　金：国家自然科学基金(No.61171183;No.61471140;No.61601140)

摘　　要：分析参考信号为窄带信号、初级路径为非线性路径、次级路径为非最小相位系统的主动噪声控制(ANC)的初级噪声特性,探究有限脉冲响应(FIR)反馈预测结构和前馈函数链接人工神经网络(FLANN)结构的消噪原理,并探讨其各自的优势及局限性.通过分析总结,本文提出FLANN-FIR混合结构ANC系统,该系统对FIR反馈预测结构和前馈FLANN结构进行优势互补,不仅可以消除可预测部分噪声,对不可预测部分噪声一样具有抑制作用.仿真表明,在不同信噪比参考信号和不同非线性初级路径下,提出的混合结构都具有良好的稳态误差和收敛速度.Analyzing the primary noise of an ANC system which contains narrowband reference signal,nonlinear primary path and non-minimum phase secondary path,and exploring the feedback structure based on FIR filter and the feedforward structure based on FLANN filter. By discussing the advantage and the disadvantage of these two structures,we proposed a newhybrid structure in this paper. Proposed FLANN-FIR structure composites the advantages of feedforward and feedback structures,not only the predictable noise can be reduced,but also the unpredictable noise can still be suppressed.Simulations under the cases of different signal noise ratio and different nonlinear primary path demonstrate that proposed hybrid structure performs best on the convergence rate and steady-state property.

关 键 词：窄带主动噪声控制 非线性路径 非最小相位 函数链接人工神经网络 反馈预测 

分 类 号：TP273.2[自动化与计算机技术—检测技术与自动化装置]