Estimating Parameter And Application Of SMC-GWO Algorithm Based On Nonlinear Dynamical System

Markov chain Monte Carlo(MCMC)algorithm,sequential Monte Carlo sampling(SMC)algorithm and swarm intelligence algorithm(SI)can be used for uncertainty estimation and parameter estimation of complex system.Though methods for estimating parameter are countless,it is important to choose efficient algorithm to estimate unknown parameter of nonlinear dynamical system.As for parameter estimation of nonlinear dynamic system,this paper is divided into two situations: one is the nonlinear dynamic system with fixed parameters estimation;the other is the discrete nonlinear state-space stochastic model with constantly adjusted parameters according to the new observation data.This paper selects Lorenz chaotic model to study while discussing algorithms which is suitable for estimation for nonlinear dynamic system with fixed parameter.This paper presents:(1)an mixing algorithm which combines SMC,grey wolves optimization(GWO)algorithm with M-H rule(in short as SMC-GWO);(2)comparison of practicability and validation among adaptive Metropolis(AM)algorithm,differential evolution Markov chain(DE-MC)algorithm,SMC,GWO,SMC-GWO,SMC-RWM and SMC-DEMC mixing algorithm for parameters estimation of Lorenz model.This paper selects the propagation models about Ebola virus : discrete SIR model and Richards model while discussing methods which is suitable for estimation for discrete nonlinear state space model.This paper presents comparison of simulated results among SMC-RWM,SMC-DEMC,SMC-GWO algorithms for parameters estimation in two models,and illustrates that SMC-GWO algorithm is the most suitable algorithm for discrete nonlinear models by estimating basic reproductive number of the two models.