Fault detection with non-linear nuisance parameters and safe train navigation
Abstract
Fault Detection and Identification (FDI) problems often have to take into account the interference of nuisance parameters in the elaboration of decision processes. There are many works addressing cases in which nuisance parameters interfere in a linear and additive way, most of them in a deterministic framework. The main contribution of the paper is to propose a fully statistical methodology for dealing with non-linear nuisance parameters. The results obtained allow an analysis of the risks (in terms of non-detection and false alarm probabilities) attached to a statistical test designed for such non-linear models. The developed method is applied to the integrity monitoring of GNSS train navigation and some simulations demonstrate the worthiness of this approach.