摘要
针对目标跟踪中非线性滤波精度下降甚至发散的问题,提出了一种时变噪声统计估计的自适应无迹卡尔曼滤波(Unscented Kalman Filtering,UKF)算法。首先将系统模型和滤波算法修正为适于噪声非零均值时的情况,然后根据极大后验估计原理,推导出一种次优的时变噪声统计估计器,其系数通过指数加权的衰减因子计算得到,最后与传统UKF算法结合形成自适应的滤波算法。仿真结果表明,该算法保证了滤波收敛性,能够对目标进行有效跟踪,而且滤波精度显著提高。
To solve the problem that the filtering accuracy is decline even divergence during target tracking in nonlinear system, a kind of adaptive unscented Kalman filtering （UKF） algorithm based on time-varying noise statistics estimation was proposed. Firstly, system models and filtering algorithm were modified to be suitable for the nonzero mean situation. Afterwards, according to maximum a posterior estimation theory, a sub-optimal time-varying noise statistics estimator was designed, whose coefficient was calculated by use of exponential weighted fading factor. Finally, the adaptive filtering algorithm was developed with the aid of combination with traditional UKF algorithm. Simulation results showed that this algorithm can converge and achieve the effective target tracking, and the filtering accuracy can be remarkably increased.
出处
《火炮发射与控制学报》
北大核心
2013年第1期51-55,共5页
Journal of Gun Launch & Control
基金
航空科学基金(20105196016)
关键词
无迹卡尔曼滤波
自适应滤波
目标跟踪
时变噪声统计
unscented Kalman filtering
adaptive filtering
target tracking~ time-varying noise statistics
