Accurate compensation of moving-gimbal effects based on FXLMS algorithm in magnetically suspended control moment gyroscope

In magnetically suspended control moment gyroscope (MSCMG), disturbance to rotor supporting system resulted from gimbal revolve can be depressed by means of gimbal angular-rate feedforward, but compensation precision of feedforward with invariant characteristic is reduced by system model errors significantly. The purpose of this study is to introduce adaptive feedforward into MIMO magnetic bearings system based on FXLMS algorithm to eliminate moving-gimbal disturbance. The algorithm of feedforward weight iteration was deduced via gradient estimation and steepest descent strategy, and convergence factor and initial feedforward weight were designed according to convergence property and convergence velocity. Simulation results showed that feedforward with FXLMS adaptation converged swiftly and rotor run-out decreased to 20% than that without adaptation corresponding to the same current stiffness errors. It is concluded that FXLMS feedforward could raise compensation precision remarkably so as to improve precision of output moment of MSCMG.