摘要
为了减轻粒子滤波计算复杂性,提出了一种基于交互式多模型(IMM)架构的自适应Unscented粒子滤波算法(AUPF)。IMM-AUPF算法在粒子滤波重采样步骤中设计了一个重采样控制器,根据滤波性能在线调节重采样粒子的数量。并将自适应粒子滤波算法应用于交互式多模型估计方法中,有效地解决了地面机动目标跟踪问题。实验结果表明:基于粒子滤波的多模型滤波器在估计精度方面优于标准的交互式多模型滤波器,且IMM-AUPF算法在计算复杂性方面优于交互式多模型Unscented粒子滤波算法。
An adaptive unscented particle filter （AUPF） based on the interacting multiple model （IMM） frame was developed to curtail the computational complexity of particle filters. According to some filtering performance,the algorithm designs a re-sampling controller in the re-sampling step to tune online the number of re-sampling particles in each model. The IMM-AUPF,combining the adaptive unscented particle filter with the interacting multiple model estimator,efficiently addresses the ground maneuvering target tracking problem. Simulations show that particle filter-based filters outperform IMM-based filters in terms of the estimation accuracy and that the IMM-AUPF behaves better than the interacting multiple model unscented particle filter in terms of the computational complexity.
出处
《清华大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2010年第4期539-542,共4页
Journal of Tsinghua University(Science and Technology)
基金
国防"十一五"预研计划项目(102060310)
总装预研基金项目(402040502)
