Transverse flux permanent magnet machines (TFPMs) are categorized as synchronous machines that benefit from having high value of torque density and capablity of accommodating high pole numbers. These characteristics make TFPMs suitable candidates for low-speed applications where high torque density value is requred such as direct drive wind turbine application. Despite the aforementioned advantages, TFPMs suffer from intrinsically high cogging torque value which is an important concern for wind turbine application. This paper focuses on axial PM segmentation technique to minimized cogging torque of TFPM topologies. Concept of the proposed method is discussed using analytical equations and optimum segmentaion angle is formulized. Non-linear magnetic equivalent circuit (MEC) is adopted where the PM segmentation, armature reaction, rotor transition and iron saturation effect are carefully modeled. The results of the MEC simulation are compared with the finite element method (FEM) results in terms of accuracy and computational time. The results from the analysis indicate that the proposed MEC method is almost ten times faster than FEM with reasonable level of precision. Taguchi method is adopted as a fast-response optimization method to improve the generator torque characteristics. The results show that the cogging torque has reduced by 97% with respect to the initial design while the average torque has only dropped by 8% which is an acceptable side effect due to the significant improvement in machine cogging torque.