| Popis: |
Gears are main components in the industrial machines and considered among the most critical elements. Therefore their diagnosis is vital to avoid failure and prevent catastrophic failure of the machinery. The vibration analysis is the most used approach for the gears diagnosis. However the vibration signal captured from a gear mesh with faulty gears is in general non-stationary and noisy in nature. In this frame, this work addresses the gears diagnosis by the combination of the classical indicators with a denoising method. For this a gear dynamic model, including the evolution of a localized tooth defect, is developed. The model consists of a spur gear pair, two inertias which incorporate the effects of time-varying mesh stiffness and damping, excitation due to the gear errors. The Newmark integration scheme is used to calculate a dynamic model. Then, the obtained simulation signals were drawn in a random noise which simulates the vibration background noise. For the simulation signals, the temporal and the frequency indicators were proposed for the early detection of the crack evolution. By using the wavelet technique these indicators were improved and the comparison of its performance was made. Several types of wavelet functions were used for this purpose and it was seen that the most appropriate is the adaptive wavelet function. The simulation results are validated with experimental signals which consist in a test rig with gear transmissions. |
