Improvement of extended Kalman filter based on multi-innovation theory

Because of the low estimation accuracy of normal extended Kalman filter( EKF) in strong nonlinear system,this paper developed an improved extended Kalman filter( MI-EKF) to solve the problem,and it improved the filtering accuracy greatly. It proposed MI-EKF by combining multi-innovation theory and the standard EKF. MI-EKF had better precision and stability,because MI-EKF considered not only the current measured value,but also gave full consideration to the time before state of motion. Finally,it discussed the impact of algorithm precision which included different numbers of innovations. Simulation results show that the improved algorithm MI-EKF included two innovations is optimal.