【摘要】：針對數(shù)據(jù)同化過程中集合數(shù)目有限情形下的虛假相關問題,通過模糊控制算法判斷觀測點與狀態(tài)更新點之間的距離,構造觀測位置等價權重,與集合轉換卡爾曼濾波方法相結合,提出一種新的數(shù)據(jù)同化方法。利用經典的Lorenz-96混沌模型,比較分析集合轉換卡爾曼濾波(ETKF),局地化集合轉換卡爾曼濾波(LETKF)和模糊控制數(shù)據(jù)同化算法(FETKF)在不同參數(shù)變化時的性能,由此探討3種方法的優(yōu)劣。研究結果表明:新方法能夠使每一步狀態(tài)更新獲得更有效的觀測信息,減小因觀測數(shù)據(jù)難以得到有效利用而帶來的誤差,同時避免了同化過程中的虛假相關問題,從而提高濾波精度。
[Abstract]:In order to solve the problem of false correlation in the case of finite number of sets in the process of data assimilation, the distance between observation point and state update point is judged by fuzzy control algorithm, and the equivalent weight of observation position is constructed. A new data assimilation method is proposed by combining with the set transform Kalman filtering method. The classical Lorenz-96 chaotic model is used to compare and analyze the performance of set converted Kalman filter (ETKF),) local set converted Kalman filter (LETKF) and fuzzy control data assimilation algorithm (FETKF) under different parameters. The advantages and disadvantages of the three methods are discussed. The results show that the new method can obtain more effective observation information for each step state update, reduce the error caused by the difficulty of using the observed data effectively, and avoid the false correlation problem in the assimilation process. Thus, the filtering accuracy is improved.
【作者單位】： 西北師范大學物理與電子工程學院;
【基金】：國家自然科學基金項目(41461078、41061038) 蘭州市科技計劃項目(2015-3-34)資助
【分類號】：O415.5;TP273.4
，

本文編號：2413485