WEI Wen-jun1， GAO Xue-ze1， GE Li-min2， GAO Zhong-jun3
（1. School of Automaton and Electrical Engineering, Lanzhou Jiaotong University, Lanzhou 730070, China；2. School of Mechatronics Engineering, Lanzhou Jiaotong University, Lanzhou 730070, China；3. China Academy of Railway Sciences, Beijing 100081, China）
Abstract： To solve low precision and poor stability of the extended Kalman filter (EKF) in the vehicle integrated positioning system owing to acceleration, deceleration and turning (hereinafter referred to as maneuvering), the paper presents an adaptive filter algorithm that combines interacting multiple model (IMM) and non-linear Kalman filter. The algorithm describes the motion mode of vehicle by using three state space models. At first, the parallel filter of each model is realized by using multiple nonlinear filters. Then the weight integration of filtering result is carried out by using the model matching likelihood function so as to get the system positioning information. The method has advantages of nonlinear system filter and overcomes disadvantages of single model of filtering algorithm that has poor effects on positioning the maneuvering target. At last, the paper uses IMM and EKF methods to simulate the global positioning system (GPS)/inertial navigation system (INS)/dead reckoning (DR) integrated positioning system, respectively. The results indicate that the IMM algorithm is obviously superior to EKF filter used in the integrated positioning system at present. Moreover, it can greatly enhance the stability and positioning precision of integrated positioning system.
Key words： vehicle; integrated positioning system; information fusion algorithm; extended Kalman filter (KEF); interacting multiple model (IMM)
CLD number： TP873Document code： A
Article ID： 1674-8042（2018）03-0279-07doi： 10.3969/j.issn.1674-8042.2018.03.010
References
［1］Zhang L S, Li B, Gong J H, et al. Maximum likelihood identification of nonlinear model for high-speed train. Acta Automatica Sinica, 2014, 40(12): 2950-2959.
［2］Lu H H, Zhang S, Wang H, et al. WiFi integrated positioning system using a federated Kalman filter. Journal of Tsinghua University(Science and Technology), 2011, 51(3): 420-423.
［3］Lai L, Qu S R. Moving vehicle location method based on traffic wireless sensor network. Journal of Traffic and Transportation Engineering, 2013, 13(1): 114-116.
［4］Zhai D L, Lei H M, Li j, et al. Trajectory prediction of hypersonic vehicle based on adaptive IMM. Acta Aeronautica et Astronautica Sinica, 2016, 37(11): 3466-3468.
［5］Jwo D J, Tseng C H. Interacting multiple model nonlinear filter design for ultra-tightly coupled integrated navigation. In: Proceedings of the 8th Asian Control Conference, Taiwan, China, 2011.
［6］Zhang Y Y. Research on the application of GPS/INS integrated navigation system based on IMM_AF in mobile robot. Haerbin: Harbin Institute of Technology, 2016: 5-50.
［7］Malleswaran M, Vaidehi V, Irwin S. IMM-UKF-TFS model-based approach for intelligent navigation. Journal of Navigation, 2013, 66(6): 859-877.
［8］Zhang H F, Zhang L W, Wang X L, et al. Optimization design and error analysis of strapdown inertial navigation system mechanization in polar region. Jouranl of Chinese Inertial Technology, 2015, 23(6): 701-706.
［9］Liu T, Meng F L, Liu J X. New combined location algorithm of SINS/DR for land vehicle. Modern Defense Technology, 2015, 43(4): 80-85.
［10］Zang R C, Cui P Y. Research on adaptive Markov parameter IFIMM algorithm. Acta Electronica Sinica, 2006, 34(3): 521-524.
［11］Huang X P, Wang P. Kalman filtering principle and application. Beijing: Publishing House of Electronics Industry, 2015: 85-121.
［12］Yvo B, Hans D. A multiple model hypothesis filter for Markovian switching system. Automatica, 2004, 11(18): 709-715.
［13］Li X R, Bar-Shalom Y. Design of an interacting multiple model algorithm for air traffic control tracking. IEEE Transactions on Control Systems Technology, 1993, I(3): 186-194.
［14］Li X R, Zhao Z L, Zhao Z L. Model-set design for multiple-model method II example. In: Proceedings of the 5th International Conference on Information Fusion, Washington DC, USA, 2002: 1347-1354.
［15］Dai D S, Yao M L. Improved adaptive Markov IMM algorithm. Acta Electronica Sinica, 2015, 45(5): 1198-1203.
交互多模型在车辆组合定位系统中的应用
魏文军1， 高学泽1， 葛立民2， 高忠军3
(1. 兰州交通大学 自动化与电气工程学院， 甘肃 兰州 730070；2. 兰州交通大学 机电工程学院， 甘肃 兰州 730070；
3. 中国铁道科学研究院， 北京 100081)
摘要:为解决扩展卡尔曼滤波器（extended Kalman filter, EKF）在车辆组合定位系统中因车辆加减速、 转弯（以下简称机动）而存在的精度低、 稳定性差等问题， 设计了一种将交互多模型（interacting multiple model, IMM）算法与非线性卡尔曼滤波器相融合的自适应滤波算法。 该算法使用三种状态空间模型来描述车辆的运动模式， 采用多个非线性滤波器对每个模型并行滤波， 通过模型匹配似然函数对滤波结果进行加权融合， 最终得到系统的定位信息。 该方法具备非线性系统滤波器优点， 克服了单一模型滤波算法对机动目标定位效果差的缺点。 利用该方法和EKF算法分别对GPS/INS/DR车辆组合定位系统中进行了仿真实验， 结果表明， 该算法的滤波定位精度明显优于目前组合定位系统中所用的EKF滤波器， 大幅提高了组合定位系统的稳定性和定位精度。
关键词:车辆； 组合定位系统； 信息融合算法； 扩展卡尔曼滤波器； 交互多模型
引用格式：WEI Wen-jun， GAO Xue-ze， GE Li-min， et al. Application of interacting multiple model in integrated positioning system of vehicle. Journal of Measurement Science and Instrumentation, 2018, 9(3)： 279-285. ［doi：10.3969／j.issn.1674-8042.2018.03.010］286Journal of Measurement Science and InstrumentationVol.9 No.3, Sept. 2018FAN Ze-yuan, et al. ／ Information fusion of train speed and distance measurements based on fuzzy adaptive ...Journal of Measurement Science and InstrumentationVol.9 No.3, Sept. 2018
[full text view]
