FPGA implementation of a real-time model predictive controller for hybrid power systems

Abstract

Hybrid power systems can find wide applications in hybrid electric vehicles, more-electric aircraft, electric ships, and microgrids, where power converters are important in controlling the instantaneous power flow. Model predictive control (MPC) provides promising benefits over traditional controllers such as PIDs to dramatically improve dynamic control performances of those nonlinear multi-input, multi-output systems (MIMO). The MPC approach is to proactively adjust control actions based on internal model prediction with the advantages of simultaneously optimizing the operation for multiple objectives and generally guaranteeing an optimal path towards these control objectives. This paper presents the recent work on FPGA implementation of a real-time nonlinear model predictive controller for the power management of hybrid power systems. The MPC formulation, FPGA architecture of a quadratic programming solver, and real-time implementation issues will be discussed in detail. The FPGA based MPC implementation will be tested and compared with PID controllers, showing promising results. Finally, based on testing results, the effects of load prediction and different control parameters on system performances will be elaborated.