Improved Maximum Likelihood Method for Ship Parameter Identification
Hongli Chen, Qiang Li, Ziyuan Wang · 2018
An improved maximum likelihood method is proposed to estimate the state estimation problem of the unknown hydrodynamic parameters in the nonlinear state space equation of a ship. On this basis, a new nonlinear filtering algorithm, Belief Condensation Filtering, is studied. A new likelihood probability formula is derived. Further, a new calculation strategy of expectation and maximization of two stages is proposed by using Monte Carlo method and two parameter updating methods. And the theory proves that the two strategies are convergent with the updating of the parameters. The proposed algorithm is an iterative process of two stages: using the improved maximum likelihood method to identify the current time parameters, and using nonlinear filtering and identification parameters to estimate the state of the current time. The algorithm does not use all observation data to do maximum likelihood iterative computation. It only uses the observed value of current time to calculate. It is simple in calculation and can be applied to real-time online parameter identification. In view of a 3 degree of freedom nonlinear state space model for Underwater Unmanned Aerial Vehicle (AUV), the realization and results of specific identification of ship parameters are given. The simulation results show that the proposed method is effective and can be used to identify the nonlinear parameters of the ship.