Gear fault diagnosis using spectral Gini index and segmented energy spectrum

Abstract Fault diagnosis of gears is crucial for maintaining the stable operation of a gearbox within a mechanical system. Traditional envelope demodulation methods depend on the distribution of sidebands around a central frequency. However, due to various interferences such as amplitude modulation, frequency modulation and assembly errors, the sidebands do not always distribute regularly. To circumvent dependence on sidebands distribution, a novel method, based on spectral Gini index (SGI) and segmented energy spectrum, is proposed to extract fault features from the perspective of energy variation in a specific frequency band to achieve fault diagnosis. Considering the operational characteristics of gears, the vibration signal is segmented into a series of short-time vectors according to the meshing frequency, to calculate the frequency response during each gear engagement. The SGI is employed as a new method to determine the optimal frequency band. An energy sequence is obtained by calculating the energy values of the segmented vectors within the optimal frequency band. Subsequently, the spectrum of the energy sequence is computed to identify the fault characteristic frequency. For comparison, methods based on band-pass filtering and envelope demodulation are also conducted and discussed. The effectiveness of the proposed method is validated through numerical and experimental studies.