基于EKF的初始状态自适应室内融合定位算法

Initial state adaptive indoor fusion location algorithm based on EKF

针对扩展卡尔曼滤波(EKF)系统滤波初始值偏差会导致滤波结果发生偏离的问题,对融合定位系统的初始状态偏差进行了定量分析,并提出了一种初始状态自适应的EKF融合定位算法,通过对Wi-Fi定位初始多点协同卡尔曼滤波(KF)从而获得精准的初始位置与初始航向角,该方法能够良好适应不同的初始定位状态。同时提出了一种改进的EKF算法,以邻近状态RSSI欧氏距离作为度量动态调整EKF系统参数,以降低Wi-Fi数据波动对EKF系统所造成的影响。实验表明:系统可以获得精确的初始状态,能够在4步内收敛到状态的真实值,该方法能很好的降低Wi-Fi定位的波动与PDR的累积误差,从而提高定位精度,在相同条件下,该融合定位算法的平均误差为0.97 m,相较于最新的EKF算法,能够提高17.2%的定位精度。

The deviation of initial value of extended Kalman filtering(EKF)system may result in distinct difference of the state filtering.Therefore, the fusion of positioning system has carried on the quantitative analysis of the initial state problem, and an EKF fusion algorithm with adaptive initial state is proposed.Through the initial Wi-Fi location points with Kalman filter to obtain the accurate initial position and heading angle, the method can well adapt to different initial state.Finally, an improved EKF algorithm based on dynamic system parameters measured by the received signal strength indication(RSSI)Euclidean distance of adjacent states is designed to reduce the influence of Wi-Fi fluctuation on the EKF system.Experimental verification shows that the system can obtain accurate initial state and quickly converge to the true value in four steps.The result can well reduce the fluctuation of Wi-Fi positioning and the accumulated error of PDR,and improve positioning accuracy.Under the same conditions, the average error of the fusion positioning algorithm is 0.97 m, which can improve the positioning precision by 17.2 % compared with the latest EKF algorithm.