Aiming at the complex fault signal components and difficulty in identifying the fault features of a hoist spindle device, this study proposes a method based on a filtering algorithm, Hilbert-Huang transform (HHT), energy entropy, and support vector machines (SVM). The filtering method is applied to the vibration signal under different fault conditions. Then, the Hilbert-Huang transform is applied to the noise-reduced signal. The empirical mode decomposition (EMD) method decomposes the noise-reduced vibration signal into a set of intrinsic mode functions (IMF). Then, the Hilbert transform (HT) calculates the envelope spectrum of the first few IMFs. Afterward, it evaluates and extracts the fault characteristic frequencies. Finally, the identification of different fault defect types is determined by combining the intrinsic modal energy entropy and SVM. The experimental results show that the method can accurately identify the faults in the rotor bearing system and is an effective fault signal processing method.