A Gradient Descent Based Method For Maximum Efficiency Calculation

The power losses and the efficiency of electrical machines are becoming increasingly important. Furthermore the possibilities of automated (machine design) optimization processes are growing. For combining both, a calculation of the electrical machine's maximum efficiency is necessary. The efficiency calculation should be done without human interaction, like a visual evaluation of efficiency maps. Because of this a gradient descent based optimization was chosen for the calculation of the maximum efficiency of an electrical machine. The method is applied to a permanent magnet synchronous machine, nevertheless it could be applied to other types of machines. It can be shown that the maximum efficiency can be found. The studies show that the area can be found independently of the starting values. Further the results as well as conspicuous features are discussed. Especially the efficiency dependencies and the resulting optimization path are highlighted. The gained results can be used for the subsequent machine design optimization process.