Investigation of Disturbance Observers for Model Predictive Current Control in Electric Drives

Abstract

Model predictive control (MPC) of power electronic converters has obtained much attention in many applications and especially in electric motor control. As the control loop is closed by predicting the future plant behavior by means of a mathematical model, disturbances and uncertainties are important aspects when using any MPC strategy. The plant model may be inaccurate due to plenty of reasons, such as parameter mismatches or the inverter nonlinearity. If these disturbances are not properly addressed during the MPC design process, the control performance is deteriorated. Hence, a suitable disturbance observer (DOB) is required to compensate for model inaccuracies. This contribution is comparing different lumped-DOB designs in the context of a continuous-control set MPC for induction motor current control. As a baseline for comparison, a field-oriented proportional-integral (PI)-type regulator is utilized which does not require a DOB due to its integral feedback.Comprehensive experimental results prove the necessity of a proper DOB; however, it is also shown that the overall transient and steady-state control improvement due to MPC has to be bought at a high price since the computational burden is at least doubled compared to the PI-baseline.