State variable constraints on the performance of optimal Maglev suspension controllers

The influence of the choices of state variables when designing linear optimal controllers for low speed magnetically suspended (Maglev) vehicles is examined. The design process takes account of both random and deterministic track inputs, and shows that the controllers which are produced by a LQG approach do not in general give a satisfactory response for deterministic inputs. The performance for a number of state space formulations is evaluated, and compared in the context of the suspension design requirements. Two of the formulations include the effect of integral control to reduce the low frequency errors and improve the response to deterministic inputs.<>