چكيده لاتين
Conventional AC-DC-AC converters, commonly used in electric drive applications, suffer from several limitations including bulky passive components, reduced efficiency due to multiple power conversion stages, limited bidirectional power flow capability, and higher total harmonic distortion (THD). Additionally, they often generate significant reactive power and require large electrolytic capacitors, which reduce system lifespan and reliability. In contrast, matrix converters offer a compact, efficient, and fully bidirectional AC-AC conversion solution without the need for intermediate energy storage, making them an attractive alternative.
This thesis addresses the limitations of conventional AC-DC-AC converters in electric drive applications by utilizing a matrix converter. Two model predictive control (MPC) strategies are designed and evaluated: Model Predictive Current Control (MPCC) and Model Predictive Torque Control (MPTC). These controllers rely on an accurate model of the system and solve an optimization problem at each time step to simultaneously achieve multiple objectives, including accurate tracking of reference current or torque, reduction of reactive power on the grid side, and minimization of switching frequency. A cost function is formulated that incorporates tracking error, reactive power, and the number of switching transitions. Simulation results show that using MPCC, the matrix converter can reduce switching frequency by up to 40% compared to a conventional AC-DC-AC converter, without compromising current quality. Additionally, under equal switching frequencies, the stator current THD is reduced by up to 58% with the matrix converter. The MPTC method also achieves precise torque control and successfully maintains accurate tracking of torque and flux under load variations and direction reversals. Moreover, the reactive power on the grid side is nearly eliminated in the matrix converter, a feature not achievable with traditional converters. It is observed that increasing the weight of reactive power in the cost function further reduces it, though this may slightly compromise current tracking accuracy—highlighting a trade-off between competing objectives.
In conclusion, the application of model predictive control to matrix converters improves dynamic response, reduces switching losses, and lowers current distortion, thereby enhancing overall system efficiency and extending the lifespan of power equipment. These advancements lay the groundwork for future experimental validations, multi-drive system integration, renewable energy applications, and even hybridization with artificial intelligence techniques.
