一种新的改进辅助变量粒子滤波算法

发布时间：2018-03-05 01:09

本文选题：辅助变量粒子滤波　切入点：正则化　出处：《微电子学与计算机》2017年02期 　论文类型：期刊论文

【摘要】：针对辅助变量粒子滤波(AVPF)对状态估计无法获得较好的滤波精度的问题,本文提出了一种改进的辅助变量粒子滤波算法.将正则化的思想引入到辅助变量粒子滤波的重采样中,在重采样中将离散的概率分布函数近似为连续的分布函数,该方法不仅保留了AVPF在重采样之前依据似然值的大小对原粒子集中的各个权值进行修正,从经过平滑后的后验密度中重采样的特点,在重采样中引入正则化思想后还能够保持粒子的多样性,增加有效样本数目,能够有效抑制样本退化.针对一个被广泛采用的双峰,高度非线性的系统模型,在选取不同的过程噪声下,进行Monte Carlo仿真实验.仿真实验表明,改进的辅助变量粒子滤波具有更好滤波精度.
[Abstract]:In order to solve the problem that AVPFs can not obtain better filtering accuracy for state estimation, an improved particle filter algorithm for auxiliary variables is proposed in this paper. The regularization idea is introduced into the resampling of particle filter with auxiliary variables. The discrete probability distribution function is approximated as a continuous distribution function in resampling. This method not only preserves the AVPF to modify the weights of the original particle set according to the magnitude of the likelihood value before the resampling. From the characteristics of resampling in the smoothing posteriori density, the regularization idea can maintain the diversity of particles, increase the number of effective samples, and effectively suppress the degradation of samples. The highly nonlinear system model is simulated with Monte Carlo under different process noise. The simulation results show that the improved particle filter with auxiliary variables has better filtering accuracy.
【作者单位】：陕西师范大学计算机科学学院;
【分类号】：TN713

【相似文献】