Low-variance version of the RCC index and form factor index for machine condition monitoring

The distinctive symptom of the machine fault signal is cyclic transients. The ratio of cyclic content(RCC) and the form factor indexes have been demonstrated as two successful condition monitoring indicators dedicated to characterizing the non-stationarity and impulsiveness of the machine fault signal, respectively. However, the original version of RCC and form factor indexes rely on estimating the fourth-order and second-order moments, respectively, whose estimation variance properties are poorer than that of the first-order moment. The current work proposed the low-variance version of the RCC and form factor indexes to remedy this gap. As just estimating the first-order moment, the proposed indicators fluctuate less than their original version at the machine's normal stage. This work also illustrates an intrinsic connection between the low-variance version of the RCC index and the wavelet scattering convolutional network, providing a novel perspective to explain the black-box CNN model prevalent in the intelligent fault diagnosis community. Finally, the proposed indicators are applied to detecting the incipient bearing defect and the escalator roller defect. The experimental results verified the proposed indicators' effectiveness and low-variance advantages.