论文部分内容阅读
针对非线性、非高斯环境下具有不确定动态模型参数的系统状态估计问题,提出了一种新颖的区间衍生粒子滤波算法.该算法利用区间滤波生成的重要性密度函数,在系统状态转移概率密度的基础上,融入最新的系统观测数据,从而提高了对系统状态后验概率的逼近程度.为了进一步提高算法的实时性,提出一种类似光子衍射的粒子衍生过程,进而缓解了滤波精度与运算量之间的矛盾.通过陀螺/星敏感器组合定姿问题验证了该算法的有效性和鲁棒性.
Aiming at the problem of system state estimation with uncertain dynamic model parameters under non-linear and non-Gaussian environment, a novel interval-derived particle filter algorithm is proposed. The algorithm uses the importance density function generated by interval filter to calculate the state transition probability density , The new system observation data are integrated into the system to improve the approximation of the posterior probability of the system state.In order to further improve the real-time performance of the algorithm, a particle-like derivation process similar to photon diffraction is proposed, and the filtering accuracy and operation The contradiction between the two methods is verified by the combined attitude determination by gyro / star sensor.