中文摘要：

在多传感器多目标跟踪领域中,当传感器为被动式的,传统的多维分配算法利用拉格朗日松弛算法求解。拉格朗日乘子更新一般用次梯度方法,但每次迭代都要进行多次极小化运算来求对偶解,导致实时性差。针对这个问题,提出了一种改进的基于拉格朗日松弛的数据关联算法,通过代理修正次梯度方法更新拉格朗日乘子,并在允许时间内获得近似解。仿真实验表明,与现有的次梯度算法相比,此算法具有更少的运算时间和更高的关联正确率。

英文摘要：

In the field of multisensor-multitarget tracking,lagrangian relaxation algorithm is used to solve the classic multidimensional assignment problem when all the sensors are passive sensors which obtained the angle only.The sub gradient or the accelerated sub gradient is applied to update the lagrangian multipliers,but it needs to minimize all the sub problems at every iterative time to solve the dual solution in the classic algorithm.This leads to long compute time and bad real-time performance.Aimed at the problem,an improved data association algorithm based on the lagrangian relaxation is introduced in this paper.It uses the surrogate modified sub-gradient to update the lagrangian multipliers.A Monte Carlo simulation is used to analyze the performance of the algorithm.Compared with the classical algorithm,new algorithm has less compute time and higher association accuracy via simulation.