论文部分内容阅读
针对容积积分卡尔曼滤波(Cubature Quadrature Kalman Filter,CQKF)受模型不确定性影响较大及需要精确已知噪声统计特性的缺点,提出了一种自适应强跟踪CQKF算法。算法根据强跟踪滤波原理,引入渐消因子调整状态预测协方差矩阵,强迫残差序列正交,有效抑制了模型不确定性引起的滤波发散。在滤波过程中,利用Sage-Husa时变噪声统计估值器对过程噪声及量测噪声实时估计,提高了算法在未知时变噪声环境下的滤波精度。目标跟踪仿真实验验证了算法的有效性和鲁棒性。
To overcome the shortcomings that Cubic Quadrature Kalman Filter (CQKF) is greatly affected by the uncertainty of the model and needs accurate statistics of known noise, an adaptive strong tracking CQKF algorithm is proposed. According to the principle of strong tracking filter, the algorithm introduces a fading factor adjustment state prediction covariance matrix, forces the residuals to be orthogonal, and effectively suppresses the filter divergence caused by the model uncertainty. In the filtering process, the Sage-Husa time-varying statistical estimator is used to estimate the process noise and measurement noise in real time, which improves the filtering accuracy of the algorithm in the unknown time-varying noise environment. Target tracking simulation experiments verify the effectiveness and robustness of the proposed algorithm.