Classification of Fault Signals by Chaotic Oscillators Synchronization

Among ways of technical systems monitoring, methods using sensors of mechanical vibrations are widely used. Such sensors make it possible to inspect vibrations in electrical machines, gearboxes and other mechanisms and to identify signs of malfunctions. Simple and efficient algorithms for finding signs of faults, suitable for use in low-cost hardware, are in demand. One of the newest techniques in this field is the diagnostics based on chaotic systems. We propose a new method of this family based on the observing the synchronization error between two chaotic oscillators. For numerical study, recordings of vibrations of a motor with faulty bearing were used. We show that the proposed method reliably identifies the signs of a defect in the presence of background noise.