Acceleration of MPC using graphic processing unit

Model Predictive control (MPC) algorithm owns good performance but causes heavy computational burden. This disadvantage makes it only applied for small systems or slow processes. Due to great calculation capability of Graphic Processing Unit (GPU), this paper proposes a parallel calculation method for the acceleration of MPC based on CUDA (Compute United Device Architecture): using CPU to organize and control the procedure and using GPU to calculate matrixes and to solve variables. Simulations give calculation time and speedup of the proposed method as well as the serial. It indicates that this method can greatly deduce the calculation time and broadens the application range of MPC.