Structural Optimisation of an Induction Motor using a Genetic Algorithm and aFinite Element Method
, Acta Polytechnica Scandinavica, Electrical Engineering Series, No. 84,Helsinki, 1996, 99 p. ISBN 951-666-490-3. ISSN 0001-6845. UDC 621.313.33:519.62/.64:519.853.Keywords: numerical simulation, finite element method, non-linear optimisation, genetic algorithm,structural optimisation, induction motor, slot shape, torque, electromagnetic losses
Several dozen variables affect the characteristics of an electric motor. The magnetic circuit of anelectric motor is highly non-linear and analytically it is not possible to calculate the torque or lossesin motors with sufficient accuracy for optimisation of the near air gap region. Only with the finiteelement method (FEM) is it possible to obtain sufficient accuracy. To be able to accurately evaluatethe losses caused by higher harmonics the time-stepping method is needed to simulate the rotationof the rotor. The purpose of this work is to design and to test a method for structural optimisationand to use this method for the design of a new slot shape for induction motors, especially in theoptimisation of the near air gap region. This method enables the design of more efficient andsmaller motors, or vice versa, design of motors with a higher shaft power from the same amount of materials. This optimisation method is based on a genetic algorithm, and it is applied to theoptimisation of the slot dimensions and the whole slot geometry with different voltage sources andoptimisation constraints. In the genetic algorithm, optimisation is based on a population. Thealgorithm changes an entire population of designs instead of one single design in optimisation. TheFEM is not accurate, i.e. all the changes in the mesh do not necessarily correspond realimprovements in the characteristics of a motor. To improve the reliability of the optimisationresults with FEM, the average design of the population is studied. The results obtained clearlyindicate the usefulness and the effectiveness of both the optimisation method selected and the FEMin a design for induction motors.