Reformulation of the long-horizon direct model predictive control problem to reduce the computational effort

For direct model predictive control schemes with current reference tracking, the underlying integer least-squares (ILS) problem is reformulated to reduce the computational complexity of the solution stage. This is achieved by exploiting the geometry of the ILS problem and by reducing the computations needed for its formulation and solution. A lattice reduction and a sphere decoding algorithm are implemented. A variable speed drive system with a three-level voltage source inverter serves as an illustrative example to demonstrate the effectiveness of the proposed algorithm.