The goal of this project is to design and implement an automatic system for on-line fault detection in multidimensional (nD) processes based on nD neural networks (NN), with particular interest in aerospace, civil and mechanical structures, but of general application. As a consequence of environmental and structural conditions civil infrastructures are often subject to damage. The identification of cracks in their early growth stage allows repairing the structure without high costs. In this way the development of automatic identification techniques for online fault detection based on measured vibration data is of crucial importance. The project proposal builds on the combination of the Principal Investigator experience in nD systems with the team experience in identification, NN, structural modeling and custom high-performance embedded systems development.