北京邮电大学学报

  • EI核心期刊

北京邮电大学学报 ›› 2011, Vol. 34 ›› Issue (3): 21-25.doi: 10.13190/jbupt.201103.21.xiangw

• 论文 • 上一篇    下一篇

针对鲁棒性三维声源定位问题的全局性算法

向文,艾文宝,张鑫   

  1. 北京邮电大学 理学院, 北京 100876
  • 收稿日期:2010-04-06 修回日期:2011-01-06 出版日期:2011-06-28 发布日期:2011-03-29
  • 通讯作者: 艾文宝 E-mail:wenbaoai@gmail.com
  • 基金资助:

    国家自然科学基金项目(10971017); 中央高校基本科研业务费资金项目(BUPT2009RC0701)

A Global Algorithm for Robust 3D Acoustic Source  Localization Problem

  • Received:2010-04-06 Revised:2011-01-06 Online:2011-06-28 Published:2011-03-29
  • Contact: Wen-Bao AI E-mail:wenbaoai@gmail.com
  • Supported by:

    The National Natural Science Funds

摘要:

针对到达时间差度量误差和声源具有鲁棒性的三维声源定位问题,提出了一种用半正定规划方法求解的线性校正总体最小二乘全局性算法(LCTLSSDP). 将带二次等式约束的分式二次规划声源定位模型转化为带二次不等式约束的非凸齐次二次优化问题;利用对偶理论设计算法求出该模型的最优解. 理论证明LCTLSSDP算法能找到问题的最优解. 实验结果显示,LCTLSSDP算法有稳健的定位结果.

关键词: 三维声源定位, 最小二乘, 半正定, 二次分式优化

Abstract:

A linear correct total least squares method by semidefinite programming global algorithm (LCTLSSDP) is proposed for robust 3D acoustic source localization model considered time difference of arrival. The new algorithm finds the optimal solution of this ratio of quadratic functions model with a equality constraint by solving a nonconvex homogenous quadratic optimization problem with inequality constraints. The dual theory is also used when designing.

Key words: 3D sound localization, least squares, semidefinite positive, ratio of quadratic functions optimization