Decoupling Control of Six-Pole Radial Active Magnetic Bearings Based on BP Neural Network Inverse Optimized by Improved Mind Evolutionary Algorithm

To solve the coupling issues of radial displacements in the two degrees of freedom of a six-pole radial active magnetic bearing (AMB) rotor, a decoupling control strategy based on back propagation neural network (BPNN) inverse optimized by an improved mind evolutionary algorithm (MEA) is proposed. Firstly, the structure and working principle of the six-pole radial AMB are introduced, and its mathematical model is derived. Secondly, an inverse model is constructed using a BPNN. To solve the sensitivity to initial weights and thresholds effectively, the improved MEA is designed with an adaptive subpopulation size strategy and an adaptive variance adjustment strategy. Finally, simulation studies of the six-pole radial AMB system are performed, and experiments on rotor floating, anti-disturbance, and decoupling performance are complemented. The results verify the effectiveness and reliability of the proposed control strategy.