A particle approach to the analysis and estimation of nonlinear dynamical systems

The PhD thesis deals with the general model based estimation problem, which is solved here using particle filters as basic tool. Particle filters are sequential Monte Carlo methods able to solve the nonlinear filtering problem with the appealing feature of dropping otherwise mandatory assumptions of linear models and/or Gaussian distributions. The main contribution of this PhD thesis is to develop new algorithms where particle filters are used within a set-membership framework. Thanks to the combined approach, it is possible to estimate the probability distribution of the unknown real state and at the same time keeping a bound on the maximum error that can be committed.