多传感器任务分派的快速启发式规划新算法

A New Fast Heuristic Programming Algorithm for Multi-sensor Task Assignment

在多目标多传感器管理中经常采用的线性规划算法中,随着传感器个数和目标个数的增加,计算量会爆炸式增长,使得跟踪系统不能实时计算,为此,根据传感器管理中线性规划的特点,提出了一种快速启发式算法,考虑组合中的传感器个数将组合的分配效用转化为权重,递推分配权重最大的组合,逐步减小组合和目标的个数。证明了权重最大的组合分派能实现组合中的传感器的最大效用。仿真结果表明该算法在与采用线性规划方法的跟踪精度相当的情况下,能有效地减小计算量。

In multi-sensor management for multi-target, the standard linear programming method may not be suitable for real-time implementation because the computation burden increases exponentially with the sensor number and target number. A classical heuristic programming method is shown deficient via our example. Instead of optimizing the efficiency of sensor sets, directly, we design a new performance index, as a combination of the standard efficiency and the assigned sensor number so that the sensor resource can also be considered. Based on this new index, a new heuristic programming method is proposed in recursive form. This scheme can decrease the sensor number to be assigned step by step. Hence the optimization burden is simplified significantly. In computer simulation of multi-sensor multi-target tracking, our new method remains the same tracking accuracy, while cut about half of the computation burden, compared with the classical heuristic programming method.