基于人工鱼群优化的扩展卡尔曼粒子滤波算法

Extended Kalman Particle Filter Algorithm Based on Artificialfish School Algorithm Optimization

针对扩展卡尔曼粒子滤波算法中由于粒子退化和贫化而导致的滤波精度降低问题,提出了一种人工鱼群优化的扩展卡尔曼粒子滤波算法。通过人工鱼群优化算法中的觅食和聚群行为,对采样过程进行优化,使得粒子不断地朝高似然域移动来寻找最优位置,从而改善样本分布,加速样本集的收敛,缓解了退化现象;然后对重采样过程进行优化,以提升样本的多样性,从而克服了粒子样本贫化问题。实验结果表明,改进后算法提高了对系统状态的预估精度,更适合在对精度要求高的系统中进行滤波计算。

Considering the problem of declining in accuracy caused by particles degeneracy and impoverishment in the extended Kalman particle filter algorithm, an artificial fish school optimized extended Kalman particle filter al- gorithm was presented in this paper. It uses the alternation of behaviors of preying and swarming in the artificial fish school algorithm to optimize the sampling process firstly, which makes prior particles move towards the high likelihood region and finds the optimal position, so particles samples was improved, the convergence of particles sample was ac- celerated and the phenomenon of particles degeneracy was relieved. Then the re - sampling process was optimized, as a result, the diversity of samples was enhanced. The problem of sample impoverishment was also resolved. Experi- mental results show that the improved algorithm improves the estimation accuracy of the system state and the algorithm is more suitable for filtering calculation based on high precision requirement system.