Skip to Main Content
Model predictive control (MPC) for a two-motor drive, supplied from a five-leg inverter, is presented in this paper. As an alternative to existing methods, use of MPC in multimachine drives has the advantages of independent fast current control of the machines, elimination of the closed-loop system's cascaded structure, and a reduced number of microcontrollers. A vector control algorithm is required, necessitating state-space modeling, with each machine's direct- and quadrature-axis currents chosen as state variables. Prediction of future states is via a discrete-time model of the five-leg inverter and a piecewise-affine model of two permanent-magnet synchronous motors (PMSMs). A method which eliminates unfeasible switching states inherent in reduced-switch-count inverters while reducing computation and sampling times is proposed. The algorithm is implemented in a TMS320F28335 DSP microcontroller, which controls the five-leg inverter and the two PMSMs. Simulation and experimental results validate the presented control concept.