Set-membership identification: Bayesian approach vs subpavings approach

This paper deals with the problem of nonlinear set-membership identification. To solve this problem, a Bayesian approach is introduced and compared with the subpavings approach. The paper illustrates how the Bayesian approach can be used to determine the feasible parameter set and to check the consistency between measurement data and model. In particular, it is shown that the Bayesian approach, assuming uniform distributed estimation error and flat model prior probability distributions, leads to the same feasible parameter set than the subpavings technique. Main issues and performance of both approaches are compared and discussed by means of an application example.