This paper presents switched reluctance motor features improvement by presenting new geometries dimension optimization. The design goals are ripple of torque and weight reduction and average torque increase. To achieve these goals not only design of previous research of a normal SRM is improved, but also new geometries with optimised dimension are suggested. The performance analysis of SRM is carried out by a finite element method software. A multi objective genetic algorithm coded by MATLAB, has been employed to perform the numerical analysis to obtain solutions. COMSOL 4.3 has been selected for FEM analysis in order to establish a better online link to MATLAB. To reach a reliable solution, the being optimized parameters of each suggested geometry, has been widely changeable by MOGA as long as there is no conflicts between others. The method of optimization and the new suggested geometries has been applied to a 6/4 SRM and simulation results have been declared and discussed.