基于自适应滤波的噪声抵消算法研究与应用

发布时间：2018-06-13 16:27

本文选题：自适应滤波 + 噪声抵消　；参考：《河北科技大学》2015年硕士论文

【摘要】：本文首先研究了自适应滤波器的基本原理、系统组成结构以及实际应用,深入分析了以最小均方误差(MMSE)为准则的LMS算法和基于最小二乘准则的RLS算法;分析了两种算法的优点以及缺点,并进行了对比;对于传统的LMS算法的收敛速度以及稳态误差之间的无法兼顾,在误差接近零时出现较大幅度的波动以及在低信噪比情况下性能降低等问题,提出了一种变步长LMS算法,通过仿真验证本文提出的算法优越于传统的滤波算法以及改进的算法。其次在自适应滤波算法的基础上,对于消噪系统如何使用自适应算法消除噪声进行系统的分析,自适应噪声抵消系统的实质就是运用自适应滤波理论有效的消除噪声,对于传统消噪算法,重点分析了结构、性能以及衡量指标。对于传统的自适应噪声抵消算法难以解决脉冲噪声以及参考通道信号的难以获取的问题,提出了一种改进的自适应噪声抵消算法,并与传统的算法进行了分析以及对比。仿真结果证明对于周期性信号提取问题本文提出的算法具有一定的优越性并且克服了传统算法处于脉冲噪声干扰的环境下性能恶化的问题。最后,基于对自适应噪声技术理论分析以及研究的基础上,运用TI公司的高速处理器件TMS320VC5509A、性能优越的音频处理芯片TLV320AIC23及其它相关模块共同构成一个自适应噪声抵消系统,利用DSP集成开发环境(CCS4.2)的软件仿真模式下,实现了传统的LMS和RLS及本文提出的自适应滤波算法。在硬件仿真器模式下,分别选取noise92库里的白噪声和工厂背景噪声作为环境噪声进行语音信号消噪,采用简单而且容易实现的LMS算法作为核心滤波算法,选取硬件仿真器进行下载链接,在此系统平台上进行语音信号实时消噪处理。本文详细介绍了噪声抵消系统结构组成部分,并给出了DSP以及音频芯片AIC23的链接框图和程序流程图,最后给出了测试结果,具有一定的参考价值。
[Abstract]:In this paper, the basic principle, system structure and practical application of adaptive filter are studied. The LMS algorithm based on minimum mean square error (MMSE) and the RLS algorithm based on least square criterion are deeply analyzed. The advantages and disadvantages of the two algorithms are analyzed and compared, and the convergence speed and steady-state error of the traditional LMS algorithm can not be taken into account. In this paper, a variable step size LMS algorithm is proposed, which is superior to the traditional filtering algorithm and the improved algorithm by simulation, due to the large fluctuation of the error near 00:00 and the degradation of the performance in the case of low signal-to-noise ratio (SNR). Secondly, on the basis of adaptive filtering algorithm, how to use adaptive algorithm to eliminate noise is analyzed systematically. The essence of adaptive noise cancellation system is to use adaptive filtering theory to eliminate noise effectively. For the traditional denoising algorithm, the structure, performance and measurement index are analyzed. For the traditional adaptive noise cancellation algorithm is difficult to solve the problem of impulse noise and reference channel signal difficult to obtain, an improved adaptive noise cancellation algorithm is proposed, and compared with the traditional algorithm. The simulation results show that the proposed algorithm is superior to the periodic signal extraction problem and overcomes the problem that the traditional algorithm is in the environment of impulse noise interference. Finally, based on the theoretical analysis and research of adaptive noise technology, an adaptive noise cancellation system is composed of TMS320VC5509A, TLV320AIC23, an audio processing chip with superior performance, and other related modules. Under the software simulation mode of DSP integrated development environment (CCS 4.2), the traditional LMS and RLS and the adaptive filtering algorithm proposed in this paper are realized. In the hardware simulator mode, the white noise and factory background noise in noise92 library are selected as the ambient noise to remove the noise of speech signal, and the simple and easy to realize the algorithm is used as the core filtering algorithm. The hardware simulator is selected to download the link, and the real-time speech signal denoising processing is carried out on this system platform. In this paper, the structure of noise cancellation system is introduced in detail, and the link block diagram and program flow chart of DSP and AIC23 are given. Finally, the test results are given, which have certain reference value.
【学位授予单位】：河北科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN911.4;TN713

【参考文献】