基于自适应滤波及模态分析的有源噪声控制方法研究

发布时间：2018-01-14 02:14

本文关键词：基于自适应滤波及模态分析的有源噪声控制方法研究　出处：《西安电子科技大学》2014年博士论文　论文类型：学位论文

【摘要】：随着经济的发展和社会的进步，噪声污染日益严重，，危害人们的身心健康，引起机械设备结构的声疲劳，并降低武器装备的作战性能和声隐身性能。因此噪声控制在民用和军事领域中都具有十分重要的意义。有源噪声控制技术是控制低频噪声的有效方法，根据噪声控制机理可分为声源控制、传播途径控制和接受者保护三类，研究内容和手段具有多样性。论文从接受者保护和声源控制的角度出发，针对基于自适应滤波技术的前馈有源噪声控制方法和基于模态分析的结构声辐射控制方法中存在的一些主要问题开展研究工作。论文所做主要工作概况如下： 1.在自适应前馈有源控制技术中，非因果成分会降低噪声控制性能，为了使系统尽量满足因果约束，提出一种参考信号的拾取方法。该方法将参考传声器放置在噪声源来波方向上，并通过距离调整来增大参考传声器和误差传声器的时间延迟，用于抵消串扰和反射等因素引起的非因果成分，从而提高噪声控制性能。在此方法中，为了检测宽带噪声源的来波方向，提出了一种基于KR积的内插聚焦估计算法。该算法将各频率点的协方差矩阵进行KR积扩展后，借助最小二乘内插聚焦矩阵，构造无噪声聚焦后的各频率数据组合矩阵，进而采用MUSIC方法实现过定和欠定两种情况下的来波方向估计。最后，将噪声源的定位结果用于指导参考传声器的布放，通过比较传声器布放位置及数目调整前后的噪声控制效果，验证参考信号拾取方法的有效性。 2.在宽带噪声的自适应前馈有源控制技术中，无延迟子带自适应滤波方法由于能避免信号通道的延迟且收敛速度快，成为宽带噪声控制中的主要方法。但该方法的控制效果会受到分析滤波器组的群延迟和带内混叠误差的影响，因此分析滤波组的设计是关键。现有的分析滤波器组是调制线性相位原型低通滤波器得到的，它不能保证滤波器组的低群延迟和小旁瓣效应，致使噪声控制效果不明显。为此，提出迭代二阶锥规划方法设计分析滤波器组。将带内混叠误差和群延迟构造为关于原型滤波器权系数的二次函数，进而通过迭代二阶锥优化得到原型滤波器权系数，最后利用均匀DFT调制为分析滤波器组。与同阶线性相位分析滤波器组相比，该滤波器组具有更低的群延迟和更小的带内混叠误差。在此基础上，采用部分子带权系数更新的归一化最小均方误差算法，构建了频域无延迟噪声控制系统。实验结果表明，该系统的噪声控制效果明显，不仅能有效降低残余噪声功率谱，提高噪声衰减量，而且还具有收敛速度快和计算复杂度低等优点。 3．在窄带噪声的自适应前馈有源控制技术中，从理论上分析了频率偏差造成自适应滤波器权系数的均值偏移不能收敛到零矢量，是导致控制性能恶化的根本原因。因此为了提高噪声控制效果，必须对干扰信号的频率进行精确估计。为此，提出了一种用于实值信号的快速子空间跟踪-ESPRIT类频率估计方法。该方法的信号子空间维数与信号频率的个数相等，从而避免了传统方法用于实值频率估计时，信号子空间的维数为频率个数2倍的问题。仿真结果表明算法的精度高，满足实际应用的要求。最后，结合不需要次级通道估计的FXLMS算法，在DSP平台上实现了弹性薄板的基频和谐波分量的振动控制，从而降低了封闭声腔中的结构声辐射的声压级。 4．在基于模态分析的弹性结构声辐射的控制技术中，根据输入输出测量数据建立精确受控系统模型是实施该类控制方法的前提。目前，主要采用子空间辨识方法来建立系统状态方程，但该方法计算量大，难以得到有效应用。为此，利用广义能观矩阵的列空间与观测矢量相关矩阵的信号子空间一致的特征，提出了一种在线子空间辨识方法用于系统状态方程的建立。该方法包含了两个递推过程：一是利用矩阵求逆变换计算观测矢量；二是基于多级分解幂迭代子空间跟踪算法估计广义能观矩阵，以保证按指数全局收敛到主子空间。仿真结果表明该方法能提高辨识精度，不仅减小广义能观矩阵估计值与真实值的夹角，还能降低辨识得到的输出数据与真实值之间的均方根误差。最后，利用有限元和声学分析软件，构建了一种基于振动控制的封闭空间降噪系统模拟环境。采用所提的在线子空间辨识法建立了压电结构系统模型，结合线性二次最优控制方法产生压电陶瓷作动器次级电压。通过控制与声模态耦合大的结构模态，有效降低了舱室结构声辐射的声压级，验证了在线子空间辨识方法应用在结构声辐射控制中的有效性。
[Abstract]:With the development of economy and the progress of the society, noise pollution and harm people's physical and mental health, caused by fatigue structure of mechanical equipment, and operational performance and reduce weaponry stealth performance. So the noise control is very important in civil and military field. The active noise control technology is an effective method to control the low frequency according to the noise, the noise control mechanism can be divided into control source, transmission and receiver protection three, research contents and methods of diversity. The protection from the recipient and source control angle, aiming at some main problems of adaptive filtering feedforward active noise control method and radiation control of structural acoustic modal analysis based on the method based on the research work done in this paper. The main work are summarized as follows:
1. in the feedforward adaptive active control technology, non causal components will reduce the noise control performance, in order to enable the system to meet the causality constraint, proposes a method for picking up the reference signal. This method will be placed in the reference microphone noise source to the wave direction, and through adjusting the distance to increase the reference microphone and microphone error time delay. For the non causal components offset caused by crosstalk and reflection and other factors, so as to improve the noise control performance. In this method, in order to detect the broadband noise source to wave direction, we propose a KR based product focusing interpolation estimation algorithm. The algorithm makes the covariance matrix of each frequency of KR product extension, by using the least squares interpolation focusing matrix, the frequency of data combination matrix without noise after focusing, and then using the MUSIC method to achieve a fixed and underdetermined two cases to estimate the wave direction. After that, the location result of noise source is used to guide the placement of reference microphone. The effectiveness of the reference signal pickup method is verified by comparing the noise control effect before and after adjusting the location and number of microphone placement.
2. control feedforward adaptive active noise in broadband technology, delayless subband adaptive filtering method can avoid the delay due to the signal channel and the convergence speed, and become the main method of broadband noise control. But the control effect will be affected by the analysis filters group delay and band aliasing error, therefore analysis and design of filter group is the key. The analysis filters the existing modulation is a linear phase prototype low-pass filter is obtained, it can not guarantee that the low group filter group delay and small sidelobe effects, resulting in the noise control effect is not obvious. Therefore, the analysis filter bank design iteration two order cone programming method. The band aliasing the group delay error and constructed as a quadratic function of the prototype filter coefficients, and then through the iteration of two order cone optimization prototype filter coefficients, finally using uniform DFT modulation Analysis of filter group. With the same order linear phase filter analysis compared with internal aliasing error of the filter has lower group delay and smaller. On this basis, the part of the sub coefficient updating weighted normalized least mean square error algorithm, constructed the frequency domain without delay noise control system. Experimental results show that the noise control effect of the system is obvious, not only can effectively reduce the residual noise power spectrum, improve the noise attenuation, but also has the advantages of fast convergence and low computational complexity.
3. control feedforward adaptive active noise in narrowband technology, the mean shift cannot converge from the theoretical analysis of the frequency deviation caused by the adaptive filter coefficients to zero vectors is to control root cause of performance deterioration. So in order to improve the noise control effect, to accurately estimate the frequency of interference signal. Therefore, put forward a for real valued signal fast subspace tracking -ESPRIT frequency estimation method. The number of signal subspace dimension and signal frequency of this method is the same, so as to avoid the traditional method for real valued frequency estimation, the dimension of the signal subspace is 2 times the number of frequency. The simulation results show that the algorithm has high precision, meet the requirement of practical application. Finally, combined with the FXLMS algorithm without the secondary path estimation, in the DSP platform to achieve the fundamental frequency and harmonic components of elastic thin plate Control, thereby reducing the sound pressure level in the enclosure structure acoustic radiation.
4. based on the control technology of elastic structural acoustic radiation modal analysis, according to the input and output measurement data to establish a precisely controlled system model is a prerequisite for the implementation of the control method. At present, mainly to establish the state equation of the system using subspace identification methods, but the computation amount is large, it is difficult to get the effective application. Therefore, the use of features the signals associated with the column space observation vector matrix generalized observability matrix space consistent, a method of online subspace identification is used to establish the state equation of the system is presented. This method includes two processes: one is to use recursive matrix inverse calculation of observation vector; two is the power iteration subspace decomposition the tracking algorithm to estimate the generalized observability matrix based on exponential to ensure global convergence to the principal subspace. The simulation results show that this method can improve the identification accuracy, not only can reduce the generalized The angle between the estimated value and the true value of the view matrix, but also can reduce the root mean square error between the output data and the actual identification value obtained. Finally, using finite element and acoustic analysis software to build a simulation environment for closed space noise reduction system of vibration control. Based on the established model of piezoelectric structure system using online subspace identification method the combination of two linear optimal control method to produce piezoelectric ceramic actuator. Through the coupling of secondary voltage control and modal acoustic modes in large, effectively reducing the sound pressure level of cabin structure acoustic radiation, verify the effectiveness of the application of online subspace identification method in the control of structural acoustic radiation.

【学位授予单位】：西安电子科技大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TN911.4;TN713

【参考文献】