Skip to Main Content
This paper presents a novel strategy of switched reluctance motor optimal design. The strategy is based on the so called flux linkage characteristic. The flux linkage characteristic contains most of the information of the machine and thus could be regarded as the “footprint” of the machine performance. In this work, first the desired flux linkage characteristic is identified, and then the optimal design parameters are sought after starting form this “idealized characteristic”. This could be regarded as an inverse problem. In this paper, neural networks are proposed to identify the mapping between the design variables and the flux linkage curve of the machine, and thus overcoming the nonlinearities that are inherent to this type of problems. Finite elements analysis is used to validate this approach.