Markov Chain Monte Carlo (MCMC) Method for Parameter Estimation of Nonlinear Dynamical Systems

摘要

This manuscript is concerned with parameter estimation of nonlinear dynamical system. Bayesian framework is very useful for parameter estimation, Metropolis-Hastings (MH) algorithm is proposed for constructing the posterior density, which is main working procedure of Bayesian analysis. Extended Kalman Filter (EKF) gives better results in non-linear environment at each time step in which Taylor series approximation for nonlinear system is used. A performance comparison of EKF in linear and non-linear environment is proposed. This study will give us the solution for nonlinear systems, numerical integration of complex integrals and parameter estimation of stochastic differential equations (SDE).