Lumped-parameter network-based design and optimization of synchronous reluctance machines

The thesis develops a methodology for the design and optimization of synchronous reluctance machines, using lumped-parameter network models to describe their electromagnetic behavior and a PSO (Particle Swarm Optimization) algorithm to optimize the design. Firstly, the work focuses on rotor design, particularly on defining the geometry of air barriers and carriers to optimize the magnetic flux path. Subsequently, three lumped-parameter network models were developed to analyze the machine under different conditions, investigating magnetic saturation and cross-coupling effects. The PSO algorithm was employed to optimize the machine's geometry and electrical parameters by iteratively generating new configurations and evaluating their performance to maximize a specific objective. The codes, implemented in MATLAB, were validated through FEM simulations performed using the FEMM 4.2 software.