Temperature Rise-Aware Compensation Control for Active Magnetic Suspension Bearing System Based on Neural Network Identification

Significant temperature rise can lead to parameter variations in the rotor of magnetic levitation bearings and cause thermal drift in eddy current sensors, thereby affecting the stability of active magnetic suspension systems. To enhance the stability of magnetic suspension systems under high-temperature conditions, a temperature compensation control method based on neural network identification is proposed. Firstly, a linearized magnetic force model of active magnetic bearings and a single-degree-of-freedom axial magnetic levitation rotor system model are established to analyze the influence of temperature on the stiffness parameters of magnetic suspension rotors, power amplifiers, and displacement sensor amplification gains. Secondly, an offline identification method using neural networks is adopted to identify the resistance values of the stator coil and the probe coil of the eddy current sensor under different temperature conditions. Then, based on the identified resistance values, a temperature rise compensation control scheme for the active magnetic suspension system is designed using a combination of series and parallel compensation. Finally, through Simulink simulationand experimentation, it is demonstrated that the temperature compensation control method based on neural network identification can improve the stability of the system.