Improved particle filtering for state and parameter estimation- CSTR model

This paper addresses the problem of states and parameters estimation for a continuously stirred tank reactor using Bayesian methods. The performances of various conventional and state-of-the-art state estimation techniques are compared when they are utilized to achieve this objective. These techniques include the Particle Filter (PF), and the developed improved particle filter (IPF). Unlike the PF which depends on the choice of sampling distribution used to estimate the posterior distribution, the IPF yields an optimum choice of the sampling distribution, which also accounts for the observed data. The proposal sampling distribution is obtained by minimizing the Kullback-Leibler divergence (KLD) distance. The simulation results show that the new improved particle filter superiors to the standard particle filter. In addition, IPF can still provide both convergence as well as accuracy related advantages over other estimation methods.