The trend to increase systems complexity while requiring higher reliability increases the scope to use model based fault diagnosis methods. As a sufficiently good modeling of complex plants can be very demanding or even impossible, data based methods are being widely used, but the computational time and the quality can still be too poor for reliable fault detection. In this paper, we propose an iterative multilayer approach, which is characterized by two elements: fault detection and fault isolation consist of sequentially triggered cascaded processes, whose speed and quality rely on dynamical modeling abilities. The presented methods have been developed in the framework of an industrial project in the field of engine test benches, from which examples are shown.