Sparse paradigm for change detection applications

Fault detection and diagnosis is a research area that has a significant impact due to commercial and industrial implications. The study presented in this paper is based on the idea that mathematical models can be assigned to any processes so, in this context, we have started from the assumption that the fault monitoring and detection in industrial processes could be made using frequency model of the signals. This hypothesis has some limitations, especially due to signal resolution, therefore an original approach based on sparse model of a signal is proposed. The results obtained for the bearing faults detection from sparse point of view are presented.