H∞ control of active magnetic bearings using artificial neural network identification of uncertainty

A novel approach is presented that couples the "hard computing" features of robust H_∞ control with the "soft computing" characteristics of intelligent system identification, and realizes the benefits of both. Radial basis function networks (RBFNs) are used to experimentally identify linear parameter varying (LPV) dynamic models and uncertainty models associated with an active magnetic bearing (AMB) test rig. These models are used to synthesize robust H_∞ controllers that are gain-scheduled with rotor setpoint. Experimental results confirm the stability robustness and exceptional tracking performance of these "intelligent robust " controllers.