一种抗野值的UPF算法

An Algorithm of Unscented Particle Filter for Outliers Rejection

为解决野值对滤波效果产生的不利影响,从改进粒子滤波器提议分布(proposal distri-bution)的容错能力入手,提出一种抗野值的UPF(unscented particle filter)算法。提出的抗野值容错算法将抗"飞点"的UKF(unscented Kalman filter)算法作为野值环境中的粒子取样策略,通过抗"飞点"UKF算法中引入的门限函数向量对新息的控制,进而实现UPF算法对野值容错的目的。仿真实验验证了所构造的容错滤波算法的有效性和实用性。

To eliminate the unfavorable influence of outliers on the filter, a new unscented particle filter （UPF） algorithm for outlier rejection was proposed in terms of improving the fault tolerance of the proposal distribution. The unscented Kalman filter （UKF） algorithm for outliers＇ rejection was chosen as the proposal method in the particle filter. The vector of control over abnormal innovations, which can efficiently r and the unscented particle filter remarkably improves the threshold function brought into the UKF has a estrain the unfavorable influence out of outliers, fault tolerance to outliers at the same time. The simulation results proved that the proposed fault tolerant UPF algorithm is available for outliers＇ rejection.