论文部分内容阅读
粒子滤波算法在非线性滤波领域受到广泛关注,但是该算法存在样本退化问题.为了改进粒子滤波算法的性能,这里结合自适应优化机制对粒子滤波算法的建议分布选择机制及重采样技术进行改进.对于粒子滤波的建议分布选择,提出一种基于自适应退火参数优化的混合建议分布方法.通过混合建议分布不足的分析,利用退火参数来优化控制状态转移先验分布函数和观测似然函数之间的比例,同时,基于自适应参数优化机制来动态调整退火参数的值.对于粒子滤波的重采样,提出了基于部分分层重采样优化算法的自适应重采样技术.通过有效样本大小的评估来执行自适应重采样策略,此外,基于部分分层重采样算法,利用权重优化的思想对其重采样前后权重计算的方法进行优化.通过相关算法的性能比较,所提改进粒子滤波算法的有效性得以验证.
Particle filter algorithm is widely concerned in the field of nonlinear filtering, but the algorithm has the problem of sample degeneration.In order to improve the performance of particle filter algorithm, this paper combines the adaptive optimization mechanism to improve the proposed distribution selection mechanism and resampling algorithm of particle filter algorithm. For the proposed selection of particle filter, a hybrid recommendation distribution method based on adaptive annealing parameter optimization is proposed.Analyzing the distribution of hybrid recommendation, the annealing parameters are used to optimize the control transfer between state transition prior distribution function and observation likelihood function , And the parameters of annealing parameters are dynamically adjusted based on the adaptive parameter optimization mechanism.For the resampling of particle filter, an adaptive resampling technique based on partial layer resampling optimization algorithm is proposed.By evaluating the effective sample size The adaptive resampling strategy is implemented.In addition, based on the partial layer resampling algorithm, the method of weight optimization is used to optimize the method of weight calculation before and after resampling.According to the performance comparison of the relevant algorithms, the effectiveness of the improved particle filtering algorithm Be verified.