多目标跟踪中联合概率数据关联优化算法

Joint Probability Algorithm in Data Association Based on DHN Artificial Nerve Network

研究多目标实时精确跟踪问题,针对在战场目标密集和电磁环境复杂的情况下,经典多目标跟踪联合概率数据关联算法的计算量剧增,在目前的硬件条件下,算法的实时性和有效性已不能满足要求。为提高算法的计算速度和效率,通过量化可行联合事件中相匹配的量测与目标的统计距离,来定义优化计算的约束条件,从而构建能量函数,把搜索最优可行矩阵问题转化为整数规划问题,对算法进行改进。在分析优化条件完备性的基础上,构建DHN人工神经网络,利用非线性计算能力对整数规划问题进行求解,并采用温控扰动的办法提高收敛速度,降低计算量。仿真结果表明,算法能在复杂条件下实时有效地量测数据,准确分配给目标航迹。

In the complex electromagnetic environment with dense target concentration, the classic joint probability data association algorithm in multi-target tracking is associated with the enormous amount of computing. The realtime ability and associating effectiveness of algorithm cannot meet the requirements base on present hardware. The conditions of constraints for optimizing are defined by quantizing the statistical distance of targets with measurement in the feasible joint event, thus constructing an energy function and translating the problem of searching the best feasible matrix into an integer planning optimization problem. On the basis of analyzing the completeness of optimality conditions, the preferred joint probability data association algorithm is invented by constructing DHN artificial neural network and using the nonlinear computing ability. And the method of disturbance controlled by system temperature is adopted to improve the convergence rate and reduce the amount of computing. The simulation results indicate that the algorithm can distribute the measured data to target-path effectively in real-time.