Fast Eserogram: A novel adaptive spectrum segmentation method for rolling bearing fault diagnosis

Early fault diagnosis of rolling bearings remains a critical challenge in industrial applications due to weak fault features and strong background noise interference. While the Fast Kurtogram (FK), as a classic fault diagnosis method, has been widely employed for fault detection in rotating machinery, its fixed frequency-band segmentation strategy may discard fault-related components, and conventional statistical indices often lack robustness under strong noise interferences. To overcome these challenges, a novel Fast Eserogram method for early fault detection is proposed in this paper. First, the Fourier model-fitting method is employed to extract spectral trends. Second, the local minima points of the spectral trend are integrated with a 1/3-binary tree filter bank to construct an adaptive filter group, maximizing the preservation of fault-related frequency bands. Subsequently, a robust statistical index—the Envelope Spectrum Energy Ratio (ESER) is established to evaluate the fault information richness in narrowband signals, enabling the selection of the optimal demodulation frequency band (ODFB). Finally, fault diagnosis is achieved through envelope spectrum analysis of the selected ODFB. The Fast Eserogram method exhibits superior performance to FK, Autogram, Accurgram, RCCgram and Log-cycligram in detecting early rolling bearing faults, as evidenced by experimental results.