近场声源定位的改进最大似然方法研究

Improved Maximum Likelihood Localization of Near-Field Sources

研究近场声源定位性能优化问题,针对常规最大似然方法在空间非均匀高斯噪声背景下定位准确率下降的问题,基于平面阵建立了近场声源信号模型,推导了空间非均匀噪声条件下求解声源方位和距离信息的最大似然定位方法,提出使用引力搜索算法解决了以上最大似然方法在多维参数空间搜索的高运算复杂度问题,通过仿真验证了改进方法的可行性和有效性,并说明估计精度较高,在低信噪比下方位和距离的均方误差都小于常规最大似然方法,并且在高信噪比条件下方位和距离的均方误差都逼近克拉美-罗界。

This paper studied the optimization of localization performance of near-field sound sources.Aimed at solving the problem of degraded localization accuracy of conventional maximum likelihood method in locating multiple near-field sources in the context of non-uniform spatial noise,the near-field signal model based on planar sensor array was first constructed and then the maximum likelihood localization method was derived in detail to obtain the values of the azimuth and distance of sound sources.Moreover,a gravitational search algorithm（GSA） was employed to further reduce the high computational complexity incurred in the multi - dimensional search process of the deduced maximum likelihood localization method.The simulations were conducted to validate the feasibility and efficiency of the proposed method.The simulation results show that the proposed method has a better estimation accuracy,with lower mean squared error of both azimuth estimation and distance estimation than that of conventional maximum likelihood method at low signal-to-noise ratio（SNR）,and can approach the corresponding Cramer-Rao Bound（CRB） of mean squared error of both azimuth estimation and distance estimation at high signal-to-noise ratio（SNR）.