© 2017 This paper presents a improved Bernstein global optimization algorithm based model predictive control (MPC) scheme for the nonlinear systems. A new improvement in the Bernstein algorithm is the introduction of a box pruning operator, which during a branch-and-bound search, discard portions of the solution search space that do not contain global solution, thereby speeding up the algorithm. The applicability of this MPC scheme is demonstrated with a simulation studies on a nonlinear single machine infinite bus power system over a wide range of operating conditions. The simulation results show improvement in the system damping and settling time compared with the classical power system stabilizer and partial feedback linearization control schemes.