基于对角结构的有源噪声控制算法研究

发布时间：2018-04-11 05:07

本文选题：有源噪声控制 + 非线性自适应控制　；参考：《南京大学》2017年硕士论文

【摘要】：有源噪声控制(ANC)随着人们对生活质量要求的提高越来越受关注,在实际应用中也越来越广泛。通过自适应控制器和电声设备行程的有源噪声控制系统来实现噪声控制的研究也在不断的深入,而对其研究的核心就是对有源控制算法进行优化。在本文中,我们深入研究了非线性自适应有源噪声控制方向的基本理论及算法,在该领域的基本算法的基础上,提出了一种基于对角结构的双线性滤波器用于非线性有源噪声控制。使用此种结构,我们推导了多通道的基于对角结构的双线性滤波最小均方算法(DBFXLMS)。在论文中给出的模拟仿真实验中,相对于之前的沃尔泰拉滤波最小均方算法(VFXLMS)和S滤波最小均方算法(FSLMS)两种基础算法,本算法被证明可以有效的实现控制效果的提升。为了改进算法,降低DBFXLMS的算法复杂度,我们给出了基于减少部分输入信号数量的简化对角结构双线性滤波最小均方算法(SDBFXLMS)以及减通道对角结构双线性滤波最小均方算法(CRDBFXLMS)。根据仿真显示,在保证不降低控制效果的基础上,改进后的算法大幅度降低了其算法复杂度。为了算法复杂度的进一步优化,我们还将误差滤波器的知识运用到之前提出的算法中,推导出双线性误差滤波最小均方算法(BFELMS),对角结构双线性误差滤波最小均方算法(DBFELMS)和减通道对角结构双线性误差滤波最小均方算法(CRDBFELMS)。通过仿真,我们给出了上述优化算法的控制效果,达到了之前提出的DBFXLMS算法的优异控制效果,并且在计算分析了算法的复杂度后,新提出的改进算法均相对于DBFXLMS有着极大的优化,最优化结果可以相比DBFXLMS降低近80%的加法计算量和81%的乘法计算量。本文所提出的基于对角结构的双线性滤波算法及双线性误差滤波最小均方算法等一系列的优化算法相比传统用于非线性有源噪声控制的基于自适应沃尔泰拉滤波器的VFXLMS算法和函数链接型人工神经网络FSLMS算法具有更好的表现。
[Abstract]:Active noise control (ANC) has attracted more and more attention with the improvement of quality of life (QOL), and has been widely used in practice.The research of active noise control through adaptive controller and active noise control system of electro-acoustic equipment travel is also deepening, and the core of the research is to optimize the active control algorithm.In this paper, we deeply study the basic theory and algorithm of nonlinear adaptive active noise control.A bilinear filter based on diagonal structure is proposed for nonlinear active noise control.Using this structure, we derive a multichannel bilinear filtering minimum mean square algorithm based on diagonal structure (DBF XLMS).In the simulation experiments presented in this paper, compared with the previous VFXLMSalgorithm and the S-filtered LMS algorithm, this algorithm has been proved to be effective in improving the control effect.In order to improve the algorithm and reduce the complexity of DBFXLMS, we present a simplified diagonal structure bilinear filter minimum mean square (SDBF XLMS) algorithm based on reducing the number of partial input signals and a subtractive channel diagonal structure bilinear filter minimum mean square algorithm (CRDBF XLMSM).The simulation results show that the improved algorithm greatly reduces the complexity of the algorithm on the basis of not reducing the control effect.In order to further optimize the complexity of the algorithm, we also apply the knowledge of error filter to the proposed algorithm.The bilinear error filtering minimum mean square algorithm (BFELMS), diagonal structure bilinear error filtering minimum mean square (DBFELMS) algorithm and subtractive channel diagonal structure bilinear error filtering minimum mean square algorithm (CRDBFELMS) are derived.Through simulation, the control effect of the above optimization algorithm is given, which achieves the excellent control effect of the previous DBFXLMS algorithm, and after calculating and analyzing the complexity of the algorithm,Compared with DBFXLMS, the new improved algorithm has a great optimization. Compared with DBFXLMS, the optimization results can reduce the addition computation by nearly 80% and the multiplication computation by 81%.In this paper, a series of optimization algorithms based on diagonal structure bilinear filter and bilinear error filter minimum mean square algorithm are proposed. Compared with the traditional adaptive Voltelian filter for nonlinear active noise controlThe VFXLMS algorithm and the FSLMS algorithm of the function link type artificial neural network have better performance.
【学位授予单位】：南京大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TB535;TN713

【相似文献】