Thermal Stress Based Model Predictive Control of Power Electronic Converters in Electric Drives Applications
Power electronics is used increasingly in a wide range of application fields such as variable speed drives, electric vehicles and renewable energy systems. It has become a crucial component for the further development of emerging application fields such as lighting, more-electric aircrafts and medical systems. The reliable operation over the designed lifetime is essential for any power electronic system, particularly because the reliability of power electronics is becoming a prerequisite for the system safety in several key areas like energy, medicine and transportation. The thermal stress of power electronic components is one of the most important causes of their failure. Proper thermal management plays an important role for more reliable and cost effective energy conversion. As one of the most vulnerable and expensive components, power semiconductors, are the focus of this thesis. Active thermal control is a possibility to control the junction temperatures of power semiconductors in order to reduce the thermal stress. For this purpose the finite control-set model predictive control (FCS-MPC) is chosen. In FCS-MPC the switching vector is selected using a multi-parameter optimization that can include non-linear electric and thermal stress related models. This switching vector is directly applied to the physical system. This allows the direct control of the switching-state and the current through each semiconductor at each time instant. For cost-effective control of the thermal stress a measure for the degradation of the semiconductor's lifetime is necessary. Existing lifetime models in literature are based on the thermal cycling amplitudes and maximum values of recorded junction temperature profiles. For online estimation of the degradation, a method to detect the junction temperatures of the semiconductors during operation is designed and validated. An existing and proven lifetime model is adapted for online estimation of the thermal stress. An algorithm for the FCS-MPC is written that utilizes this model to drive the inverter with reduced stress and equalize the degradation of the semiconductors in a power module. The algorithm is demonstrated in simulation and validated in experiment. A technique to find the optimal trade-off between reduction of the thermal stress and allowing additional losses in the system is given. The effect of rotor flux variation of the machine on the junction temperatures of the driving inverter is investigated. It can be used as another parameter to control the junction temperature. This allows increasing the maximal thermal cycling amplitude that can be compensated by an active thermal controller. A suitable controller is proposed and validated in experiment. The integration of this technique into the FCS-MPC is presented.