基于有偏卡尔曼的多传感器数据融合研究  

Research on Multi-sensor Data Fusion Based on Biased Kalman

作　　者：岳元龙[1] 陈亚南 孙钦[2] 左信[1] YUE Yuan-long;CHEN Ya-nan;SUN Qin;ZUO Xin(Department of Automation,China University of Petroleum Beijing,Beijing 102249,China;CNOOC(China)Beijing Research Center,Beijing 100028,China)

机构地区：[1]中国石油大学(北京)自动化系,北京102249 [2]中海石油(中国)有限公司北京研究中心,北京100028

出　　处：《仪表技术与传感器》2022年第1期82-86,共5页Instrument Technique and Sensor

基　　金：中海石油(中国)有限公司北京研究中心研发计划项目(CCL2021RCPS0063RSN)。

摘　　要：推导出偏参数为矩阵形式的有偏卡尔曼滤波(BKF)的完整迭代过程,该算法在均方误差条件下优于卡尔曼滤波(KF),可以进一步提高估计的精度。将BKF与多传感器融合算法中的扩维融合和序贯式融合相结合,推导出多传感器扩维有偏卡尔曼滤波和多传感器序贯有偏卡尔曼滤波算法,并从理论上证明了多传感器序贯BKF融合在均方误差条件下优于扩维BKF融合。仿真实验结果表明:在均方误差条件下,多传感器扩维BKF融合和序贯BKF融合优于扩维KF融合和序贯KF融合,序贯BKF融合优于扩维BKF融合。The complete iterative procedure of the biased Kalman filter(BKF)with bias parameters in matrix form was derived,which outperforms the Kalman filter(KF)under the mean square error condition and can further improve the accuracy of estimation.By combining the BKF with the extended dimensional fusion and sequential fusion in the multi-sensor fusion algorithm,the multi-sensor extended dimensional biased Kalman filter and multi-sensor sequential biased Kalman filter algorithms can be derived.And it is theoretically demonstrated that multi-sensor sequential BKF fusion outperforms expanded dimensional BKF fusion under the mean square error condition.The simulation experimental results show that both multi-sensor extended dimensional BKF fusion and sequential BKF fusion are better than extended dimensional KF fusion as well as sequential KF fusion,and sequential BKF fusion is better than extended dimensional BKF fusion under the mean square error condition.

关 键 词：卡尔曼滤波 数据融合 有偏估计 多传感器融合 序贯融合 扩维融合 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]