Optimal modulation for AC-DC matrix converters

This thesis investigates modulation methods and switching pattern selection for three-phase AC–DC matrix converters, with the goal of minimizing output current ripple while considering switching losses. Existing research either focuses on a limited set of predefined patterns or optimizes individual patterns, but does not provide a comprehensive evaluation of all feasible switching patterns. One challenge addressed in this work is the extremely large number of possible patterns satisfying operational constraints, particularly for patterns with four or more switching states per half cycle, where manual optimization becomes impractical. An automated procedure is proposed for systematic pattern generation and optimization, evaluating all feasible patterns to identify those that achieve the optimal trade-off between output current ripple and switching losses. Analytical optimization for the selected patterns enables real-time implementation, and experimental validation confirms close agreement with theoretical predictions. Further ripple reduction is achieved by selecting an optimal set of one, two, or three patterns to be applied within a single input voltage period, with maps developed to determine the appropriate pattern based on input voltage angle and modulation index. The resulting optimized modulation strategies are compared to standard methods, demonstrating superior performance across different power factors. Finally, a modulation approach based on discontinuous conduction mode (DCM) is introduced to improve efficiency and reduce input current distortion at low output currents. A current-control-based modulator and an output-inductance estimation method are developed, providing robust and experimentally validated steady-state and transient performance.