模糊自适应混合退火粒子滤波算法

THE ALGORITHM OF FUZZY ADAPTIVE HYBRID ANNEALED PARTICLE FILTER

针对非线性、非高斯系统状态的在线估计问题,及粒子滤波本身固有的退化问题,在已提出的混合退火粒子滤波算法的基础上提出一种新的粒子滤波算法。在滤波算法中,根据系统的状态噪声统计特性和量测噪声统计特性的关系引入调整因子,再由模糊推理系统产生退火系数。用状态参数分解和退火系数来产生重要性概率密度函数。在保留原算法优点的基础上取得了更佳的退火系数。仿真实验表明该粒子滤波器的性能优于混合退火粒子滤波算法。

A new particle filter algorithm is proposed based on the hybrid annealed particle filter（HAPF） for on-line estimation of non-Gaussian nonlinear systems and inherent degeneracy problem of the particle filter.In the filtering algorithm,according to the relation between the statistical properties of state noise and measurement noise of the system,we introduce an adjustment factor,then an annealed coefficient is produced by fuzzy inference system.The state parameters separation and the annealed coefficient are used to produce important probability density function.Using the algorithm,we get better annealed coefficient on the basis of keeping the advantages of HAPF.Simulation experiments show that the performance of the proposed filtering algorithm outperforms the HAPF.