Sequential Bayesian Inference for Uncertain Nonlinear Dynamic Systems: A Tutorial.

In this article, an overview of Bayesian methods for sequential simulation from posterior distributions nonlinear and non-Gaussian dynamic systems is presented. The focus mainly laid on Monte Carlo methods, which are based particle representations probability densities can be seamlessly generalized to any state-space representation. Within context, a unified framework the various Particle Filter (PF) alternatives presented solution state, state-parameter input-state-parameter estimation problems basis sparse measurements. algorithmic steps each filter thoroughly simple illustrative example utilized inference i) unobserved states, ii) unknown system parameters iii) unmeasured driving inputs.