Particle Filter for Nonlinear Systems with Multi-Step Randomly-Delayed and Missing Measurements
【摘要】：For nonlinear dynamic systems, this paper develops a new particle filter to deal with the case that multi-step random measurement delay and missing exist simultaneously and are induced by the same reason. A measurement model with multi-step random delay and missing is formulated by introducing a sequence of independent and identically distributed Bernoulli variables with a certain latency probability. Based on this model, a new particle filter is proposed by exploiting a new weighting scheme for particles. When the latency probability is unknown, an identification method is designed by the maximum likelihood criterion.The superiority of the proposed particle filter and the effectiveness of the proposed identification method are illustrated in the univariate non-stationary growth model.