基于相关信号分离技术的有源噪声控制

发布时间：2018-05-27 04:00

本文选题：有源噪声控制 + 盲信号分离　；参考：《内蒙古工业大学》2013年硕士论文

【摘要】：随着现代工业的发展，噪声污染越来越严重，传统的噪声控制方法对于中高频噪声控制效果较好，对低频控制效果不明显，而有源噪声控制方法的提出为解决这一难题带来了希望，并且由于其具有体积小、重量轻、易于控制等优点，，使得很多专家和学者投入大量精力和时间致力于有源噪声控制的研究。现在有源噪声控制技术中基本上都是根据线性自适应滤波理论通过不同的自适应控制算法来实现降噪的。但这类算法在收敛速度和稳态失调上具有矛盾性，增大步长可以加快收敛速度，同时也会增大稳态失调。并且通常这类自适应算法对于单通道可以取得较好的降噪量，但因多通道时计算量大、收敛速度慢，加上布放次级声源的数目和位置难以准确给出，使得多通道降噪难以实现取得明显效果。针对现有控制方法在多通道降噪方面的不足，本文提出了将盲信号分离技术引入到有源噪声控制中，并以两通道降噪模型为例，着重介绍了两噪声源信号相关情况下盲信号的解析分离方法。通过仿真实验证明了该降噪模型下实现相关信号分离的可行性和准确性。盲分离完成后得到的两个信号就是两噪声信号，针对该信号做傅里叶变换，然后根据频谱图所示得到噪声信号的主要频率成分，利用这些主要频率成分信息来完成次级声源信号的构造。这种基于盲信号分离的有源噪声控制方法与传统降噪方法的不同之处在于摈弃了传统的自适应控制方法，采用盲信号分离技术实现混合声源信号的分离。对于多通道降噪而言正是因为采用了盲分离技术，可以获得各个噪声信号的确切信息，从而使我们可以将多通道中的每个通道看做一个单通道去控制，能够有针对性的来安排次级声源的数目和位置，与传统降噪方法相比这也是本文的创新点之一。文章最后通过实验证明了在两通道情况下本文所讨论的降噪方法能够在长287.5cm，宽175cm的区域内取得平均3.5dB(A)的降噪量，相比传统有源降噪方法获得了更广的消声区域。
[Abstract]:With the development of modern industry, the noise pollution is becoming more and more serious. The traditional noise control method has better effect on middle and high frequency noise control, but not on low frequency noise control. The method of active noise control brings hope to solve this problem, and it has the advantages of small volume, light weight, easy to control and so on. Many experts and scholars devote a lot of energy and time to the study of active noise control. At present, the active noise control technology is based on the linear adaptive filtering theory through different adaptive control algorithms to achieve noise reduction. However, this kind of algorithm is contradictory in terms of convergence rate and steady-state misalignment. Increasing the step size can accelerate the convergence rate and increase the steady-state misalignment at the same time. In general, this kind of adaptive algorithm can achieve better noise reduction for a single channel, but due to the large amount of computation and the slow convergence speed of multi-channel, it is difficult to give the number and position of the secondary sound source. It makes the multi-channel noise reduction difficult to achieve obvious results. In view of the shortcomings of existing control methods in multi-channel noise reduction, a blind signal separation technique is proposed for active noise control, and a two-channel de-noising model is used as an example. An analytical separation method for blind signals with correlation between two noise sources is introduced. Simulation results show the feasibility and accuracy of the proposed model. The two signals obtained after blind separation are two noise signals. Fourier transform is made for this signal, and then the main frequency components of the noise signal are obtained according to the spectrum diagram. The main frequency component information is used to construct the secondary sound source signal. The difference between the active noise control method based on blind signal separation and the traditional noise reduction method lies in the rejection of the traditional adaptive control method and the separation of mixed sound source signals by blind signal separation technology. For multi-channel noise reduction, it is precisely because of the blind separation technology that the exact information of each noise signal can be obtained, so that we can treat each channel in the multi-channel as a single channel to control. The number and location of secondary sound sources can be arranged pertinently, which is one of the innovations in this paper compared with traditional noise reduction methods. In the end of the paper, it is proved by experiments that the noise reduction method discussed in this paper can achieve an average noise reduction of 3.5 dB / A in a region with a length of 287.5 cm and a wide 175cm. Compared with the traditional active noise reduction method, the noise reduction method can obtain a wider range of noise suppression.
【学位授予单位】：内蒙古工业大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TB535

【参考文献】