基于预测校正思想的气动数据融合方法  

An aerodynamic data fusion method based on predictive correction

作　　者：邓晨 陈功 敖厚军 任斯远 DENG Chen;CHEN Gong;AO Houjun;REN Siyuan(Chengdu Innovation Center for Intelligent Unmanned Aerial System,Chengdu 610072,China;Chengdu Fluid Innovation Center,Chengdu 610072,China)

机构地区：[1]成都智能空中无人系统创新中心,四川成都610072 [2]成都流体动力创新中心,四川成都610072

出　　处：《飞行力学》2024年第6期9-14,共6页Flight Dynamics

摘　　要：针对气动数据融合研究中不确定度加权融合算法和基于建模的融合算法难以提高数据准度和依赖模型精准度的局限性,提出了一种基于预测校正思想的气动数据融合方法。该方法通过输入变量离散全排列的方式来类比时间项,有效地将基于时间序列的预测校正算法应用于气动数据融合。在此基础上提出了基于集合卡尔曼滤波的数据融合算法,并与传统的Co-Kriging建模融合算法进行对比。结果表明,相比于单源数据建模结果,两种融合算法的预测准度都有较大提升,证明了此类算法的适用性;相比于Co-Kriging建模融合算法,基于集合卡尔曼滤波融合算法的全局性更好,适用性更强,有效解决了现有气动数据融合算法鲁棒性较差的问题。Aiming at the limitations of both uncertainty-weighted fusion and modeling-based fusion algorithms,which are difficult to improve the data accuracy and depend on the model accuracy in the aerodynamic data fusion research,a new method of aerodynamic data fusion based on the idea of predictive correction fusion is proposed.The method effectively applies time series-based predictive correction algorithms to aerodynamic data fusion by analogizing time terms with discrete full permutations of input variables.The data fusion algorithm based on ensemble kalman filter(EnKF)is proposed,and compared to the Co-Kriging algorithm.The results show that compared with the single source data modeling results,the prediction accuracy of the two fusion algorithms is greatly improved,which proves the applicability of these algorithms.Compared with Co-Kriging fusion algorithm,EnKF has higher accuracy and stronger applicability,and effectively solves the problem of poor robustness of downstream aerodynamic data fusion algorithm.

关 键 词：预测校正 数据融合 集合卡尔曼滤波 时间序列 

分 类 号：V211.3[航空宇航科学与技术—航空宇航推进理论与工程]