A principal study using Matlab/Simulink and FEM calculations

Today the interest for Hybrid electric vehicles (HEV) is high due to their ability to reduce the fuel consumption, which means a reduction in the operation cost as well as in the environmental impact. Weight and volume is important and the machine type with smallest size for a given performance is usually the permanently magnetized synchronous machine (PMSM). To be able to control the PMSM with good performance the position of the rotor has to be known. In terms of space, reliability and cost it would be favorable to replace the mechanical position sensor used today with a so called sensorless control system. The thesis evaluates the possibilities to use sensorless control for a Permanent Magnet Synchronous Machine (PMSM) used in a hybrid powertrain for heavy vehicles. Different machine configurations are investigated to see if changes can be made facilitating the position detection. This has been done with simulations in Matlab, Simulink and FEM software. A sensorless method has been implemented in simulations and evaluated on a realistic model of the PMSM in Simulink attained by FEM calculations. As the problem with sensorless control primarily is related to the signal injection method used at low and zero speed, the focus of the thesis has been at rotation speeds up to 5 electrical Hz. Problems associated with the signal injection method is explained using illustrations describing the machines electromagnetical behavior during loading conditions. Signal injection methods use the fact that the directions where the permanent magnets are placed in the rotor have the lowest reluctance. It is found that cross saturation effects will change this direction causing an angle error depending on the applied current. This disturbance can however be compensated for to a large extent. The sensorless control system used in the thesis introduced a limitation in the response time caused by the non ideal filters used to filter the high frequency current response. Optimizations could be done to make the system faster, however it has not been established whether it is possible to make the system equally fast as when using mechanical position sensors. Evaluation of different machine configurations showed that alterations of the rotor can provide better sensorless performance without deteriorating the machines torque production capabilities