Dual Estimation and the Unscented Transformation

摘要: Dual estimation refers to the problem of simultaneously estimating state a dynamic system and model which gives rise dynamics. Algorithms include expectation-maximization (EM), dual Kalman filtering, joint methods. These methods have recently been explored in context nonlinear modeling, where neural network is used as functional form unknown model. Typically, an extended filter (EKF) or smoother for part algorithm that estimates clean given current estimated An EKF may also be estimate weights network. This paper points out flaws using EKF, proposes improvement based on new approach called unscented transformation (UT) [3]. A substantial performance gain achieved with same order computational complexity standard EKF. The illustrated several