Beitrag zur Berechnung und Übertragung von Unsicherheiten am Beispiel der Magnetisierung von permanentmagneterregten Synchronmaschinen

This thesis deals with methods for the investigation of production tolerances and uncertainties in the calculation of electrical machines. In this context, production tolerances describe the differences between the ideally designed machine and the actual occurring geometries, excitations and material properties in the production. The described deviations are stochastic - they differ from machine to machine and often are referred to as uncertainties. If these production uncertainties become too large, the spatial-temporal periodicity of the electrical machine's air gap field can be destroyed. In some cases this leads to parasitic force density waves that eventually result in undesired vibrations or noises. To solve the presented problem, this work presents and applies a polynomial-chaos meta-modelling tool chain for the calculation of electrical machines. It enables the easier development of robust machine designs as well as a straightforward and easy-to-use calculation of sensitivity indices. The examination focuses on magnet deviations of permanent magnets in electrical machines, since their uncertainties have a significant influence on the machine's performance. In order to determine the occurring magnetization deviations a test bench is built. It allows to measure the air-field of all magnets. On this basis, the inverse problem that translates air-field measurements into the magnet's excitation sizes is solved and used in the polynomial-chaos meta-modelling tool chain. Finally, three rotors with entirely measured magnets are produced and their cogging torque is measured in another test bench to evaluate the simulation method's quality and accuracy.