基于多新息理论的EKF算法研究

Research of Extended Kalman Filter Based on Multi-innovation Theory

扩展卡尔曼滤波算法(EKF)是将卡尔曼滤波理论(KF)进一步应用到非线性系统中.然而当系统为强非线性时,EKF就会违背局部线性假设,引起误差增大,从而使得其精度降低,最终导致滤波发散.针对上述问题,提出结合多新息(multi-innovation)理论的改进EKF算法,即多新息扩展卡尔曼滤波(M I-EKF),使系统在原先只利用单个新息的情况下,扩展为能够利用之前多个时刻的新息,从而大大提高了滤波的精度.另外本文同时也从理论上证明了改进的多新息扩展卡尔曼滤波算法的收敛性.最后仿真结果表明,改进的多新息扩展卡尔曼滤波较标准扩展卡尔曼滤波算法更有效.

The Extended Kalman filter（ EKF） is based on Kalman filter（ KF） add extended it to nonlinear system. However,when the system is strongly nonlinear,the estimation accuracy of EKF algorithm is very low. Based on this problem,combining multi-innovation theory,we propose an improved extend Kalman filter algorithm（ MI-EKF）,making the system not just iterative one time of the state and take full advantage use of prior information to greatly improve the accuracy of filtering. Further,the convergence properties of the proposed Multi-innovation Extends Kalman Filter has been proved in theory. Finally,simulation results showthat the improved algorithm Multi-innovation Extends Kalman Filter is superior to the traditional Extends Kalman Filter