Morphological covering based generalized dimension for gear fault diagnosis

This investigation presents a new generalized dimension estimation method based on morphological covering (MC) technique for characterizing the nonlinearity and complexity of vibration signals measured from gearbox. A synthetic fractal signal is employed to evaluate and compare the proposed MC technique with the traditional box-counting (BC) method and a similar approach developed in literature. Results revealed that the presented MC method is the one providing the most reliable generalized dimension estimation results. Furthermore, we applied this scheme to analyze the vibration signals from a gearbox with three operation states. The estimated general dimensions are used as the input feature vector for classifiers to the gear working states. Experimental results showed that our presented scheme achieves the best performance on discriminating the gear conditions. We also explore the calculational efficiency of the MC method. Results demonstrated that the MC method requires much less computational cost than BC method and seems to be more suitable for on-line condition monitoring of gearboxes.