【摘要】 针对非线性、非高斯系统中的粒子滤波算法存在粒子权值退化和重采样后引起样本枯竭问题,提出一种自适应差分演化粒子滤波算法。用一种自适应参数控制策略对差分演化算法的参数进行控制,并以此代替粒子滤波中的重采样算法。通过对状态更新后的粒子做自适应差分变异、自适应杂交和选择等优化操作,利用权值大小选出下一时刻的粒子集合。实验表明,该算法能有效缓解粒子权值退化和样本枯竭问题,缩短算法运行时间,提高估计精度,同一般的差分演化粒子滤波算法相比,状态估计的精度更高。更多还原

【Abstract】 In this paper,a new adaptive differential evolution particle filter algorithm is proposed for the phenomenon of sample degeneracy and sample impoverishment which is caused by resampling in particle filter of the non-linear and non-Gaussian systems.It has insteaded the re-sampling algorithm by the differential evolution algorithm which is controlled by a kind of adaptive parameter Control strategy.Then operate these samples which have been updated by state with adaptively mutating,adaptively crossover and adaptively selecting,then select the set of particle by particle weights which will been used in the next time.Simulation results show that it can effectively relieve weights degeneracy and increase the diversity of particles,making the running time more shorten and improved the estimation precision,the precision of state estimation is higher than that of the general differential evolution of particle filter algorithm.更多还原