摘要
针对多传感器调度中辐射控制及频繁切换的问题,提出传感器工作时长约束及辐射控制的调度方法。该方法将目标跟踪及辐射控制过程建立为部分可观马尔可夫决策过程;采用不敏卡尔曼滤波更新目标信念状态,预测目标跟踪精度;同时,结合隐马尔可夫滤波器推导多步辐射代价,动态确定传感器工作时长;考虑跟踪任务需求,得到工作时长约束及辐射控制的调度策略。仿真结果表明,在满足跟踪任务需求下,该方法有效降低了系统辐射代价,避免了频繁切换问题,增强了系统稳定性和实用性。
To deal with the excessive switching problem in the sensor scheduling,a scheduling algorithm of emission control with the sensor timestep constraint was proposed.Firstly the process of target tracking and emission control was formulated as a partially observable markov decision process(POMDP).Then the unscented kalman filter(UKF)was used to update the target belief state and predict the target tracking accuracy.Meanwhile,the multi-step emission cost derived by using hidden Markov model(HMM)filter was given to dynamic determine the sensor timestep.Considering the tracking task requirement,a scheduling policy of emission control with the sensor timestep constraint was given.Simulation results demonstrated that the system radiation cost was reduced while meeting the tracking task requirement.Furthermore,the system was more stability and practicability because the excessive switching avoided.
作者
乔成林
段修生
单甘霖
王俭臣
QIAO Chenglin;DUAN Xiusheng;SHAN Ganlin;WANG Jianchen(Army Engineering University,Shijiazhuang 050003,China;Northwestern Polytechnical University Military Representative Bureau,Xi'an 710065,China)
出处
《探测与控制学报》
CSCD
北大核心
2018年第3期91-96,102,共7页
Journal of Detection & Control
基金
国防预研基金项目资助(012015012600A2203)
关键词
传感器调度
频繁切换
部分可观马尔可夫决策过程
时长约束
辐射代价
sensor scheduling
excessive switching
partially observable Markov decision process (POMDP)
timestep constraint
emission cost