Abstract:

The analytical value of the posterior density function cannot be obtained in the nonlinear nonGaussian, and needs to approximate by the exact importance density function. The traditional particle filter (PF) directly employs the state transition prior distribution function which does not include the latest measuring information as an importance density function to approximate the posterior density function. For the lack of measuring information of PF, a re-sampling Cubature particle filter (CPF) algorithm based on Cubature Kalman filter (CKF) is proposed. The new algorithm that incorporates the latest observations into a prior updating phase develops the importance density function by CKF that is more close to the posterior density. Simulation results show that the accuracy of CPF is higher than PF and extended particle filter (EPF). Compared with the unscented particle filter (UPF), the precision is similar, but the running time of CPF reduces by about 20%.