A Method for Optimization of Five-Phase Induction Machines Based on Genetic Algorithms

Abstract

The optimization of five-phase induction machines is addressed using a new procedure based on genetic algorithms. A constrained optimization model is introduced which considers the main machine dimensions as free variables. Number of stator and rotor slots, winding pitch, and rotor bar inclination angle are among the free design variables. In addition, the relationship between fundamental and third harmonic component of the airgap induction is also considered as a free variable. This relationship is used to shape the airgap induction making it near to a trapezoid, thus potentially increasing the output torque. The underlying machine model used in the optimization process is detailed in previous works and includes the effect of losses and saturation on the steady state performance. Thus, a mixed-integer optimization problem is defined, in which the continuous variables are codified as integer variables making the optimization problem easier to solve. Three objective functions are defined and tested: efficiency, cost of conductor material, and a weighted combination of efficiency and material costs; other objective functions can be defined, too. The proposed method was applied to the optimization of a 5.5-kW prototype machine, and the results are presented and discussed.