Multiple-model self-calibration unscented Kalman filter method
-
摘要:
基于无迹Kalman滤波方法(UKF)、自校准无迹Kalman滤波方法(SUKF)和多模型估计理论(MME),针对工程实际中强非线性系统状态方程受未知输入(如医用机械臂惯导单元的零漂误差、列车行驶中遭遇突风和机载元器件故障等)影响的问题,提出了一种多模型自校准无迹Kalman滤波方法(MSUKF),将多模型自校准Kalman滤波方法(MSKF)的适用范围扩展到了强非线性领域。该方法同时采用UKF与SUKF进行计算,根据贝叶斯定理实时分配两者先验估计值的权重,通过加权融合进而得到最终的状态估计。大量数值仿真结果表明:本文方法精度比滤波发散的UKF提高了50%,与无偏的SUKF相比也提升了4%以上,具有更强的适应性和鲁棒性。
-
关键词:
- 自校准滤波 /
- 多模型估计 /
- 无迹Kalman滤波 /
- 未知输入 /
- 故障诊断
Abstract:Based on the unscented Kalman filter (UKF), the self-calibration unscented Kalman filter (SUKF) and the multiple-model estimation (MME), considering the influences of unknown inputs (such as drift error of the IMU in medical manipulator, gust encountered by the running train and failure of onboard components) on the strongly nonlinear system state equation in engineering, the multiple-model self-calibration unscented Kalman filter (MSUKF) was proposed to expand the application scope of the multiple-model self-calibration Kalman filter (MSKF). According to the Bayes' theorem, this filtering method used the UKF and the SUKF whose weights were assigned automatically to obtain the final filtering result through weight-average way. A large number of simulation results showed that the accuracy of MSUKF was 50% higher than that of divergent UKF, and 4% higher than that of unbiased SUKF, presenting stronger adaptability and robustness.
-
表 1 UKF、SEKF和MSEKF方法的方均根误差均值
Table 1. Mean of root mean square error of UKF,SUKF and MSUKF methods
方法 方均根误差均值 MSUKF 0.086 SUKF 0.090 UKF 0.173 -
[1] KALMAN R E. A new approach to linear filtering and prediction problems[J]. Journal of Basic Engineering,1960,82(1): 35-45. doi: 10.1115/1.3662552 [2] SUNAHARA Y. An approximate method of state estimation for nonlinear dynamical systems[J]. Journal of Basic Engineering,1970,92(2): 385-393. doi: 10.1115/1.3425006 [3] FUJIMOTO O,OKITA Y,OZAKI S. Nonlinearity compensation extended Kalman filter and its application to target motion[J]. OKI Technical Review,1997,63(159): 1-12. [4] JULIER S J,UHLMANN J K. A new extension of Kalman filter to nonlinear systems[C]//Proceedings of 11th International Symposium Aerospace/Defense Sensing,Simulation and Controls. Orlando,US: SPIE,1997: 182-193. [5] JULIER S,UHLMANN J,DURRANT-WHYTE H F. A new method for the nonlinear transformation of means and covariances in filters and estimators[J]. IEEE Transactions on Automatic Control,2000,45(3): 477-482. doi: 10.1109/9.847726 [6] JULIER S J,UHLMANN J K. Unscented filtering and nonlinear estimation[J]. Proceedings of the IEEE,2004,92(3): 401-422. doi: 10.1109/JPROC.2003.823141 [7] DAN S. Optimal state estimation: Kalman,Kalman H∞ and nonlinear approaches[M]. Hoboken,US: Wiley-Interscience,2006. [8] ARASARATNAM I,HAYKIN S. Cubature Kalman filters[J]. IEEE Transactions on Automatic Control,2009,54(6): 1254-1269. doi: 10.1109/TAC.2009.2019800 [9] PITT M K,SHEPHARD N. Filtering via simulation: auxiliary particle filters[J]. Journal of the American Statistical Association,1999,94(446): 590-599. doi: 10.1080/01621459.1999.10474153 [10] 傅惠民,肖强,吴云章,等. 秩滤波方法[J]. 机械强度,2014,36(4): 521-526. FU Huimin,XIAO Qiang,WU Yunzhang,et al. Rank filter method[J]. Journal of Mechanical Strength,2014,36(4): 521-526. (in Chinese FU Huimin, XIAO Qiang, WU Yunzhang, et al . Rank filter method[J]. Journal of Mechanical Strength,2014 ,36 (4 ):521 -526 . (in Chinese)[11] 傅惠民,肖强,娄泰山,等. 非线性非高斯秩滤波方法[J]. 航空动力学报,2015,30(10): 2318-2322. FU Huimin,XIAO Qiang,LOU Taishan,et al. Nonlinear and non-Guassian rank filter method[J]. Journal of Aerospace Power,2015,30(10): 2318-2322. (in Chinese FU Huimin, XIAO Qiang, LOU Taishan, et al . Nonlinear and non-Guassian rank filter method[J]. Journal of Aerospace Power,2015 ,30 (10 ):2318 -2322 . (in Chinese)[12] BLANKE M,SCHRÖDER J. Diagnosis and fault-tolerant control[M]. 2nd ed. Berlin,German: Springer,2006. [13] CHEN J,PATTON R. Robust model-based fault diagnosis for dynamic systems[M]. Boston,US: Kluwer Academic Publishers,1999. [14] GILLIJNS S,DE MOOR B. Unbiased minimum-variance input and state estimation for linear discrete-time systems[J]. Automatica,2007,43(1): 111-116. doi: 10.1016/j.automatica.2006.08.002 [15] 傅惠民,吴云章,娄泰山,等. 自校准Kalman滤波方法[J]. 航空动力学报,2014,29(6): 1363-1368. FU Huimin,WU Yunzhang,LOU Taishan,et al. Self-calibration Kalman filter method[J]. Journal of Aerospace Power,2014,29(6): 1363-1368. (in Chinese FU Huimin, WU Yunzhang, LOU Taishan, et al . Self-calibration Kalman filter method[J]. Journal of Aerospace Power,2014 ,29 (6 ):1363 -1368 . (in Chinese)[16] 傅惠民,娄泰山,肖强,等. 自校准扩展Kalman滤波方法[J]. 航空动力学报,2014,29(11): 2710-2715. FU Huimin,LOU Taishan,XIAO Qiang,et al. Self-calibration extended Kalman filter method[J]. Journal of Aerospace Power,2014,29(11): 2710-2715. (in Chinese FU Huimin, LOU Taishan, XIAO Qiang, et al . Self-calibration extended Kalman filter method[J]. Journal of Aerospace Power,2014 ,29 (11 ):2710 -2715 . (in Chinese)[17] 傅惠民,杨海峰,肖梦丽,等. 非线性状态方程自校准滤波方法[J]. 航空动力学报,2019,34(2): 267-273. FU Huimin,YANG Haifeng,XIAO Mengli,et al. Nonlinear state equation self-calibration filtering method[J]. Journal of Aerospace Power,2019,34(2): 267-273. (in Chinese FU Huimin, YANG Haifeng, XIAO Mengli, et al . Nonlinear state equation self-calibration filtering method[J]. Journal of Aerospace Power,2019 ,34 (2 ):267 -273 . (in Chinese)[18] MAGILL D. Optimal adaptive estimation of sampled stochastic processes[J]. IEEE Transactions on Automatic Control,1965,10(4): 434-439. doi: 10.1109/TAC.1965.1098191 [19] LI Xiaorong. Hybrid estimation techniques[J]. Control & Dynamic Systems,1996,76: 213-287. [20] MAZOR E,AVERBUCH A,BAR-SHALOM Y,et al. Interacting multiple model methods in target tracking: a survey[J]. IEEE Transactions on Aerospace and Electronic Systems,1998,34(1): 103-123. doi: 10.1109/7.640267 [21] LI X R,BAR-SHALOM Y. Performance prediction of the interacting multiple model algorithm[J]. IEEE Transactions on Aerospace and Electronic Systems,1993,29(3): 755-771. doi: 10.1109/7.220926 [22] DAEIPOUR E,BAR-SHALOM Y. IMM tracking of maneuvering targets in the presence of glint[J]. IEEE Transactions on Aerospace and Electronic Systems,1998,34(3): 996-1003. doi: 10.1109/7.705913 -