Servo systems play an important role in many automated processes. In order to fulfil the hard demands on reliability and fast and precise operation, intelligent concepts for the control, supervision and (re)configuration are necessary. In this paper, an approach is presented which integrates different levels of signal processing in an electromechanical servo system. The digital controller and the model-based fault detection scheme are designed taking into account model-uncertainty and the time variant process behaviour, which is caused by temperature influences, wear, aging, etc. After a brief description of the theoretical basis an experimental application shows results for an automobile servo system which is driven by a d.c. motor.