As auxiliary particle filtering techniques needs a large computation, and takes long time when applied to parameter identification, a novel parameter identification approach is proposed. The proposed approach is based on a novel optimal filtering which can be applied to system with nonlinearity. It tries to reach the real probabilistic distribution by estimating the map between priori distribution and posterior distribution and reduce the distance between estimated probabilistic distribution and true probabilistic distribution. Here augmented states together with filtering method and auxiliary variable are used to handle parameter identification problem; it is proved by theory that the precision of the used filtering method has an inverse relationship with double the number of supporting points in use. The proposed algorithm has the advantages of lower complexity of computation, less time consumption and similar precision compared to auxiliary particle filtering; it is verified by simulation that the proposed algorithm is effective, and outperforms auxiliary particle filtering.