UKF和PF融合算法在动力定位船舶状态估计中的应用研究

Application research of state estimation algorithm for dynamic positioning ship on fusion of unscented kalman filtering and particle filtering

针对船舶动力定位状态估计时使用扩展卡尔曼滤波导致模型失配而产生滤波精度不高甚至滤波发散的问题,设计一种融合无迹卡尔曼滤波和粒子滤波的动力定位船舶状态估计算法。该算法以粒子滤波作为整体框架,运用无迹卡尔曼滤波对粒子状态的每次更新进行最优化估计,从而最优化了每个粒子的状态,再根据每个粒子的重要性分布,得出船舶复合运动中的低频状态。Matlab仿真结果表明,该方法能够从含有高频和噪声干扰的测量信息中估计出的船舶低频运动状态,相比于直接使用UKF,该方法的滤波精度更高,滤波性能也比较稳定。

Considering the problem of low accuracy and instability when using extended kalman filter （EKF） in state estimation of dynamic positioning ship,a new fusion algorithm based on unscented kalman filter （UKF） and particle filter （PF） is designed.The algorithm takes PF as its overall frame,it uses UKF to find the optimal estimation of each particle state while updating the particle distribution.The state of ship in low frequency can be divided from its compound motion accord- ing to the importance factors of each particle. Simulation results show that the new algorithm can track the true state of ship in low frequency quickly from metrical information which contain high frequency signal and noise. Compared with the algorithrn using UKF, this fusion algorithm has high precision and stable filtering effect.