Unbalance vibration compensation control using deep network for rotor system with active magnetic bearings

Unbalance vibration directly affects the operational precision, stability and life of rotary machinery. Profiting from the active control speciality of active magnetic bearing (AMB), unbalance vibration of rotor system with AMBs can be compensated and controlled automatically. This paper considers unbalance vibration minimum for rotor system with AMBs. Deep learning theory is utilized to design a compensation controller, which is added to the PID feedback control. The structure of the compensation controller is established by a deep neural network with 2 hidden layers, and its operation algorithms are designed. Model of a 4-DOF rigid rotor with AMBs is established for controller parameter setting and simulation. The unbalance vibration control of different controllers at fixed rotational speed is simulated, and the control effects of the proposed controller are demonstrated via unbalance vibration analysis and control current analysis. This research provides a new adaptive control approach for AMB control of unbalance minimum compensation, and it can also be applied in other multi-dimension vibration control.