Improved accuracy of model predictive control of Induction motor drive using FPGA

Finite control set model predictive control (FCS-MPC) is one of successful model predictive control approaches in electric drives which offers effective solution to multi variable multi criteria problems. The optimal control is found by “brute force” search over the limited set of possible control actions. Due to a discrete nature of power converters FCS-MPC is particularly well suited for use in electric drives. The performance of the control is closely related to accuracy of the model of controlled system. Conventional way of modeling electric drives is to include only simple model of the converter with ideal components with no voltage drops or effect of dead times. This simple mathematical converter description is computationally cheap enough to be implemented by conventional control hardware. On the other hand, the accuracy of the prediction is limited which may negatively impact the performance of the control. In this paper, we propose to design detailed mathematical model of the drive including the mathematical description of the inverter which allows us to address the problems associated with dead times and semiconductor voltage drops. Modeling those inverter non-linear effects can enhance the control accuracy especially in non-nominal drive conditions (e.g. low speeds). On the other hand the computational requirements increases. We propose to use FPGA to implement the control algorithm using fixed-point arithmetics with high level of pipelining resulting in very fast execution times while keeping FPGA resources low. The performance of proposed solution is verified by simulations and experiments on the laboratory prototype of induction motor drive.