Due to the accelerated growth in the field of power electronics and controller design techniques, the usage of the active magnetic bearing has picked up in industries. Active magnetic bearing helps the rotor to rotate freely without any physical contact. In brief, this paper develops a model of an active magnetic bearing using the finite element method, and its associated reduced order model, followed by the development of a robust control strategy. COMSOL software is used to perform three-dimensional simulation of an active magnetic bearing system. The state space system matrices are extracted from the finite element method, and a linear time-invariant state-space system is generated in MATLAB. Since the original system is large, the reduced order model is constructed. Then, based upon the reduced order model, a sliding mode control is designed to improve the regulation performance of an active magnetic bearing under unmodeled uncertainties. The stability analysis of closed-loop reduced order model with unmodeled uncertainties guarantees the finite time convergence of system states using Lyapunov theory. Further, it is proved that the same control law will also provide satisfactory performance for the original model using the reduced order model as an observer. The numerical simulation is carried out to illustrate the effective performance of the proposed controller for the reduced model as well as the original model with multiple initial conditions. The proposed work offers an alternative approach of using the reduced order model instead of the original model for the controller design of an active magnetic bearing. INDEX TERMS Active magnetic bearing, finite element method, full order model, 3D simulation, model order reduction, reduced order model, sliding mode control, finite time convergence.