改进的SMC-CBMeMBer前向后向平滑检测前跟踪算法

Improved SMC cardinality-balanced multi-Bernoulli forward-backward smoothing track-before-detect algorithm

针对在雷达观测下机动弱小目标的检测前跟踪(TBD)问题中,基于序贯蒙特卡洛的势均衡多伯努利检测前跟踪(SMC-CBMeMBer-TBD)算法存在目标的数目估计不准确及状态估计精度随时间下降的问题,提出了一种基于SMC-CBMeMBer前向后向平滑检测前跟踪的改进算法。该算法在预测和更新过程之间加入多目标粒子群优化算法(MOPSO),基于观测值设置适应度目标函数,使粒子集群向后验概率密度较为集中的位置分布,缓解了粒子贫乏的问题;在更新步骤之后加入平滑递归方法,利用观测值平滑滤波值,算法运算时间虽有一定延长,但获得了数目和状态估计精度的提升。仿真实验表明,与CBMeMBer-TBD方法相比,所提算法在对机动目标数目估计和目标状态估计的准确度等性能上都有所改进。

For the tracking problem of multiple maneuvering targets in radar observation,the sequential Monte-Carlo cardinality-balanced multi-Bernoulli track-before-detect(SMC-CBMeMBer-TBD)algorithm is inaccurate in the estimation of the number of targets and the precision of state estimation.An improved algorithm based on SMC-CBMeMBer forward backward smoothing track-before-detect algorithm was proposed.In the algorithm,the multi target particle swarm optimization(MOPSO)was added between the process of prediction and update,and the fitness function was set up based on the observation value to make the particle set move to the position of the larger posterior probability density distribution,and solve the particle poverty in the heavy sampling process.In the update step,the algorithm was used.Then the smoothing recursive method was added,and the arithmetic operation time was prolonged,but the number and the state estimation precision were improved.The simulation results show that compared with the CBMeMBer-TBD method,the proposed algorithm improves the accuracy of the estimation of the number of maneuvering targets and the accuracy of the target state estimation.