基于麦克风阵列误差背景下的DOA估计算法研究
发布时间：2019-05-17 15:34
【摘要】：作为信号处理领域的一个重要分支,麦克风阵列由于其特有的优越性被逐渐应用于视频会议、声纹识别、人工智能等方面。作为超分辨算法之一的多重信号分类(MUSIC)算法由于突破了瑞利限而具有很好的精度。但是为了解决目前应用环境中存在的“多径衰落”、“多址干扰”等问题,要求DOA(波达方向估计)算法精度更高,系统响应更快,适应性更强,因此必须对MUSIC算法进行改进。然而,麦克风阵列的幅度相位和位置误差又影响着DOA估计精度。故本文就是在充分考虑麦克风阵列幅度相位和位置误差的情况下,研究了 DOA估计算法。本文的主要工作包括:1.对MUSIC算法进行改进。MUSIC算法精度虽高,但是计算量过大。针对这一缺点,引入MSCS(MUSIC对称压缩谱)算法在其中做均衡处理。当MSCS算法搜索到谱峰附近时,以该极值点为中心重新划定一个小区域的搜索范围,在此范围内替换成MUSIC算法进行谱峰搜索,从而在降低运算量的同时,有效地提高了算法精度。仿真实验表明,在较高信噪比(大概25dB以上)时,该算法估计精度与MUSIC算法基本保持一致,系统响应时间却降低为原来的2/3,由此可见,算法的性能得到较大改善。2.麦克风阵列存在幅度相位误差时的DOA估计。研究了存在幅度相位误差时的麦克风自校正方法,该方法不像常规方法一样需要获取声源的精确位置,仅需在关于阵列镜像对称的位置各放置一个校正声源即可,然后对获取的数据进行矩阵分解等操作即可估计出误差。通过消声室实验获取实测数据进行实验验证,结果表明,即使是在0-4000Hz这种幅度相位误差不大的语音频段,该自校正方法的估计偏差依然很小。3.麦克风阵列存在位置误差时的DOA估计。在阵列位置误差校正的研究中引入遗传算法的思想,先通过理想阵元情况下的MUSIC算法获取估计角的初始值,然后对这一初始群体中的个体进行适应度评价获取最优个体,再通过选择、变异、交叉等操作产生下一代搜索群体,重复评优过程并减小偏差因子直至找出最优个体,代入MUSIC谱公式,获取最终的估计角度。实验和仿真结果表明,最大的估计偏差不超过0.1cm,绝大多数位置的估计偏差都在0.02cm以内。
[Abstract]:As an important branch of signal processing, microphone array has been gradually applied to video conferencing, voiceprint recognition, artificial intelligence and so on because of its unique advantages. As one of the super-resolution algorithms, the multi-signal classification (MUSIC) algorithm has good accuracy because it breaks through the Rayleigh limit. However, in order to solve the problems of "multi-path fading" and "multiple access interference" in the current application environment, DOA (Direction of arrival estimation) algorithm is required to have higher accuracy, faster system response and stronger adaptability, so it is necessary to improve the MUSIC algorithm. However, the amplitude phase and position error of microphone array affect the accuracy of DOA estimation. Therefore, this paper studies the DOA estimation algorithm under the condition of fully considering the amplitude phase and position error of microphone array. The main work of this paper includes: 1. The MUSIC algorithm is improved. Music algorithm has high accuracy, but the amount of computation is too large. In order to solve this problem, MSCS (MUSIC symmetric compression spectrum algorithm is introduced to equalize it. When the MSCS algorithm searches near the spectral peak, the search range of a small region is redefined with the extreme point as the center, and the spectral peak search is replaced by the MUSIC algorithm in this range, which not only reduces the computational complexity, but also effectively improves the accuracy of the algorithm. The simulation results show that the estimation accuracy of the algorithm is basically consistent with that of the MUSIC algorithm, but the response time of the system is reduced to 2 鈮，
本文编号：2479206
[Abstract]:As an important branch of signal processing, microphone array has been gradually applied to video conferencing, voiceprint recognition, artificial intelligence and so on because of its unique advantages. As one of the super-resolution algorithms, the multi-signal classification (MUSIC) algorithm has good accuracy because it breaks through the Rayleigh limit. However, in order to solve the problems of "multi-path fading" and "multiple access interference" in the current application environment, DOA (Direction of arrival estimation) algorithm is required to have higher accuracy, faster system response and stronger adaptability, so it is necessary to improve the MUSIC algorithm. However, the amplitude phase and position error of microphone array affect the accuracy of DOA estimation. Therefore, this paper studies the DOA estimation algorithm under the condition of fully considering the amplitude phase and position error of microphone array. The main work of this paper includes: 1. The MUSIC algorithm is improved. Music algorithm has high accuracy, but the amount of computation is too large. In order to solve this problem, MSCS (MUSIC symmetric compression spectrum algorithm is introduced to equalize it. When the MSCS algorithm searches near the spectral peak, the search range of a small region is redefined with the extreme point as the center, and the spectral peak search is replaced by the MUSIC algorithm in this range, which not only reduces the computational complexity, but also effectively improves the accuracy of the algorithm. The simulation results show that the estimation accuracy of the algorithm is basically consistent with that of the MUSIC algorithm, but the response time of the system is reduced to 2 鈮，
本文编号：2479206
本文链接：https://www.wllwen.com/kejilunwen/xinxigongchenglunwen/2479206.html
