Neural net-based robust controller design for brushless DC motordrives

摘要

A nonlinear neuro-controller is developed for controlling thenspeed of brushless dc motors operating in a high performance drivesnenvironment. The control inputs and the identification parameters of thensystem are adjusted simultaneously in real time using a system composednof three hidden-layer dynamic neural networks while the system is innoperation. The control architecture adapts and generalizes its learningnto a wide range of operating conditions and provides the necessarynabstraction when measurements are contaminated with noise. The problemnof persistently spanning excitation faced with the use of an onlinenneuro-controller is addressed. In particular, the ability of thenneuro-controller to “remember” previously trained referencentracks when confronted with an input excitation that is markedlyndifferent from what it was trained with is investigated. The intent isnto capture the nonlinear dynamics of a brushless dc motor over anynarbitrary time interval in its range of operation. The sensitivity ofnreal time neuro-controllers to random changes in the load torque also isninvestigated and very promising results are observed