LAUSR

Fault diagnosis is essential for the safe operation and subsequent maintenance of mechanical equipment. However, multifault mutual interference brings great challenges to fault diagnosis when multicomponent faults occur simultaneously. Furthermore, since multiple components operate under different varying-speed conditions, it is difficult to match fault information from different components to their respective speeds, making fault diagnosis even more difficult. A dual-guided adaptive decomposition method of fault information and fault sensitivity (FIFS-ADM) is proposed to address the above issues. First, a fault information-driven empirical wavelet transform whose segmentation boundaries are optimized by fault richness index is presented to separate multicomponent faults, thereby eliminating multifault interference. Then, multiorder tracking that refers to using multiple speed signals to resample the decomposed subsignals, respectively, is performed to convert these speed-varying subsignals into angular-domain stationary signals. Finally, a health recognition matrix composed of multicomponent fault sensitivity indicators is introduced to achieve not only the matching of fault component and speed but also fault location and health assessment of multiple components. Furthermore, simulated signals and actual signals are built to validate the proposed method, which shows the effectiveness and superiority of FIFS-ADM.