Superharmonic resonance study and fault diagnosis induced by slant crack in rotor-bearing system

Crack faults in rotor-bearing systems can cause major safety hazards, which makes it necessary to monitor and diagnose them promptly. This paper investigates the crack fault mechanism and further proposes a diagnosis between crack faults and others. A cracked stiffness model was developed based on fracture mechanics, and the cracked rotor was analyzed in detail for stresses. The multi-failure rotor test bench was established and related experiments were carried out to verify the model's accuracy. The results show that the critical rotational speed decreases obviously as the crack angle increases for crack depth ratios from 0.8 to 1. For the slant cracked rotor-bearing system with crack depth coefficient of a/R (Ratio of crack depth to shaft radius) = 1 and angle of 45^∘, superharmonic resonance and horizontal resonance were observed around 1/3ω_n (critical speed), and the resonance peaks were hysteretic in horizontal direction. Based on the harmonic resonance characteristics, three neural networks, Back Propagation (BP), Kernel Extreme Learning Machine (KELM) and Random Forest (RF), are used to classify different faults including cracks, and they are optimized by Sparrow Search Algorithm (SSA). The results show that all three models have high classification accuracy, while the optimized KELM model is the most efficient.