摘要
在最小距离设计准则下,提出了一种基于Monte-Carlo采样和K均值聚类的模型集合设计实现方法。通过讨论由于真实模式与模型之间的差异所可能引发的问题,提出基于UKF与模型误差的交互式多模型改进算法。新算法中每个模型采用Unscented Kalman Filter处理非线性估计问题。Monte-Carlo仿真实验表明当真实模式远离模型集合中的各模型时,新算法比IMM更具鲁棒性。并且当真实模式保持不变时,从全局角度考虑新算法比IMM优越。
Under the principle of Minimum Distance Design, Monte Carlo sampling technique and K-means cluster algorithm were utilized here to design the model set. A problem which may arise from the difference between true mode and model was taken into account, and an improved Interacting Multiple Model algorithm based on Unsecnetd Kalman Filter and model error as proposed. Unscented Kalman Filter was used here for each model to handle non-linear estimation problem. The results of Monte-Carlo simulations show that the new algorithm is more robust than IMM when the true mode is different from any models, and it achieves global superiority in comparison with IMM when the true mode is constant.
出处
《系统仿真学报》
CAS
CSCD
北大核心
2009年第12期3565-3569,共5页
Journal of System Simulation
基金
国家自然科学基金项目(60632050
60472060)
