由于传统分布式跟踪方法在先验噪声协方差与其实际值不相匹配时跟踪误差较大, 提出了一种采用自适应一致性无迹卡尔曼滤波的分布式目标跟踪方法, 该方法首先执行分布式UKF算法得到对当前移动目标状态的估计值, 然后通过一个系统错误检测机制, 确定是否需要对噪声协方差值进行更新。如需要, 则根据当前获得的测量信息去估计当前噪声协方差, 并联合该估计值和先前的噪声协方差值获得一个新的先验噪声协方差值。最后根据新获得的噪声协方差值对获得的目标状态估计值进行修正。实验结果表明该方法具有较好的准确性和鲁棒性: 在噪声未知环境下, 基于ACUKF的分布式跟踪方法相比于基于容积信息滤波和基于分布式无迹卡尔曼滤波的跟踪方法, 最大跟踪误差值分别减少了49.93%和 51.46%; 在目标过程噪声发生动态变化的情况下, 提出的方法相比于上述两种传统跟踪方法, 跟踪误差值分别减少了40.67%和40.06%。

As the traditional distributed methods of target tracking may suffer from performance degradation owing to mismatch between the noise distributions assumed as a priori and the actual ones, a distributed target tracking method was proposed based on adaptive consensus unscented Kalman filter to improve the accuracy and robustness of the tracking results. More specifically, at each time step, a distributed UKF (DUK) would be implemented to obtain the estimations of the moving target. Next, an online fault-detection mechanism was adopted to judge if it was necessary to update current noise covariance. If it was necessary, the estimations of the current noise covariance would be calculated according to the measurement information. By utilizing a weighting factor, the filter would combine the last noise covariance matrices with the estimations to obtain the new noise covariance matrices. Finally, the state estimations would be corrected according to the new noise covariance matrices and previous state estimations. The experiment results demonstrate that: in unknown noise environments the tracking errors of the proposed method are reduced by as much as 49.93% and 51.46% when compared with those of the distributed tracking methods based on the cubature information filter and DUK, respectively; in dynamic noise environments the tracking errors of the proposed method are reduced by as much as 40.67% and 40.06% when compared with those of the above two traditional methods, respectively. These results demonstrate that the proposed method performs well in terms of accuracy and robustness on distributed tracking with uncertain noise.