Model Order Reduction for Simulation and Control of Synchronous Machines

Model order reduction approaches aim at reducing the computational complexity of numericalmodels. The reduction is achieved by lowering the dimension of the state space or the numberof degrees of freedom in the original high-order model, which in return generates a reducedorder model. This reduced order model is an appropriate substitution for the original model inthe applications with restricted computational resources or a demanding large number of simulations. This thesis proposes a novel method, named the orthogonal interpolation method, to reducethe computational burden of numerical models, which makes the models suitable for real-timeexecution. This method is applied to a 2-D finite element model of a 2.2 kW interior permanentmagnet synchronous motor. According to the simulation results, the resulting reducedmodel accurately imitates the behaviour of the finite element model of the machine, and successfully lowers the computational time and memory requirements. Furthermore, the proposedmethod grants a more significant reduction in the computational complexity comparedto other model order reduction techniques, such as proper orthogonal decomposition coupledwith a discrete empirical interpolation method. As an application, the proposed reduced model is employed in a control system for real-timecontrol of the motor. The high computational efficiency of the reduced model allows direct implementation of the resulting control system in the embedded processor of the drive. Moreover,the simulation and the experimental results show the capability of the developed controlsystem in considering the magnetic cross-coupling and saturation phenomena of the motor,and therefore produces higher torque and output power.