Abstract
Fault identification in rotor systems has been studied by many authors, but the considered malfunction is one single fault only, generally an unbalance. Real machines can be affected by several different types of faults; moreover sometimes also two different faults may develop simultaneously. A model based method for identifying multiple faults acting simultaneously on a rotor system in the frequency domain is briefly described and its robustness with regards to measuring and modelling errors is evaluated, by means of numerical simulations performed on the models of two typical power plant machines: a steam turbogenerator and a gas turbogenerator.