摘要
针对无源时差定位(TDOA)领域的非线性方程求解问题,提出了一种基于最大似然估计的改进型拉格朗日规划神经网络迭代求解算法。该算法利用最大似然估计构建代价函数,结合时空约束条件,建立TDOA方程的一般约束优化问题,并通过迭代求解算法对网络的收敛性和渐近稳定性进行了证明。针对两种常见的阵列排布方式进行了仿真验证与性能分析。仿真实验结果表明,该算法能够提供精确的坐标估计,误差小于1.414×10^(-3)。与传统算法相比,该方法在各类噪声环境下表现出更优的性能,尤其在0 dB噪声环境下,其均方误差为0.7866。
A modified Lagrange programming neural network(LPNN)iterative algorithm based on maximum likelihood estimation was proposed for solving nonlinear equations in the field of passive time difference of arrival(TDOA)localization.The cost function based on maximum likelihood estimation was established,and the general constrained optimization problem of TDOA equations in combination with space-time constraints was constructed.In addition,the convergence and asymptotic stability of the network were proven through the iterative algorithm.Simulation verification and performance analysis of two commonly used array element placement methods in the field of TDOA localization were conducted.The results of simulation experiments show that the algorithm can provide accurate coordinate estimation with an error less than 1.414×10^(-3).Compared to conventional algorithms,this method has better performance in various noise environments,with a mean square error of 0.7866 in 0 dB noise environments.
作者
史红伟
左越
SHI Hongwei;ZUO Yue(College of Electronic and Information Engineering,Changchun University of Science and Technology,Changchun 130013,Jilin,China)
出处
《沈阳工业大学学报》
CAS
北大核心
2024年第6期832-839,共8页
Journal of Shenyang University of Technology
基金
吉林省自然科学基金项目(YDZJ202301ZYTS412)。
关键词
无源定位
时差定位
到达时间差
最大似然估计
拉格朗日规划神经网络
模拟神经网络
一般约束优化问题
代价函数
passive localization
TDOA localization
time difference of arrival
maximum likelihood estimation
Lagrange programming neural network
analog neural network
general constrained optimization problem
cost function
