摘要
为了减小建模误差和未知量测噪声特性对非线性状态估计的影响,该文提出了1种新的容积平滑变结构滤波算法。融合了非线性容积变换规则,可避免线性化误差。利用滑模变结构思想计算最优平滑边界层,约束建模误差的影响。利用变分贝叶斯实时估计动态系统的量测噪声特性,有助于优化平滑边界层的阈值。仿真结果表明,相比传统非线性滤波算法,该文算法精度可提高28.5%,具有更好的滤波性能。
A new cubature smooth variable structure filter algorithm is proposed to decrease the effect of modelling error and unknown measurement noise characteristic on nonlinear state estimation. The nonlinear cubature transform rule is combined to avoid the linearization error. Through the sliding mode variable structure idea,the optimal smoothing boundary layer is calculated to restrict the modelling error. The variational Bayesian is used to estimate the real-time measurement noise characteristic in the dynamic system,and the threshold value of smoothing boundary layer is optimized. The simulation results demonstrate that compared with the traditional nonlinear filter algorithm,the proposed algorithm improves the accuracy by 28.5%,and shows a better filter performance.
作者
张磊
郭健
钱晨
陈庆伟
Zhang Lei;Guo Jian;Qian Chen;Chen Qingwei(School of Automation,Nanjing University of Science and Technology,Nanjing 210094,China)
出处
《南京理工大学学报》
EI
CAS
CSCD
北大核心
2019年第3期255-260,共6页
Journal of Nanjing University of Science and Technology
基金
国家自然科学基金(61673214
61673217
61673219)
江苏省研究生创新基金(KYLX16_0450)
关键词
变分贝叶斯
平滑变结构滤波
非线性状态估计
容积变换
variational Bayesian
smooth variable structure filter
nonlinear state estimation
cubature transform