Design Improvement of Synchronous Reluctance Motor Geometry, Using Neural-Network, Genetic Algorithm and Finite Element Method

appropriate approach to reach high efficiency in Synchronous Reluctance (SynRel) machines is to enhance these machines’ magnetic saliency. This is usually done by changing the geometry of machine and especially by changing the number and shape of rotor flux barriers. In this paper an intelligent- method have been used to optimizing the design of SynRel motors based on magnetic saliency ratio. To achieve this aim, all of the motor parameters including stator geometry, axial length of machine, winding type, and number of flux barriers in rotor are assumed constant and just position of the rotor flux barriers are optimized. These positions have been defined by six parameters. Changing these parameters, the magnetic saliency of machine is calculated by finite element analysis (FEA). Using these values to train a neural network (NN), a modeling function is obtained for magnetic saliency of SynRel machine. Considering this NN as the target function in genetic algorithm (GA), the parameters of SynRel machine have been optimized and the best rotor structure with highest magnetic saliency has been obtained. Finally the abilities of NN in correct estimation of magnetic saliency and motor synchronization were approved by FEA and dynamic simulation.